Ambient Noise Differential Adjoint Tomography Applied to a 2D Dense Array

Abstract:

The time derivative of the seismic noise interferometry correlation function has been widely used as the approximate Green’s function between the virtual source and receiver, but it suffers from bias due to a tradeoff between bias from noise sources and structural properties. In short, the unknown distribution and temporal variation of noise sources compound the difficulties in deriving unbiased velocity structure from noise-based Green’s functions. In this study, we extend a new technique (Liu, 2020) using differential time kernels based on one virtual source and two neighboring receivers to a dense nodal array in Albania. We compute sensitivity kernels for differential travel time and amplitude measurements. These differential sensitivity kernels cancel the overlapping part of the original source and structure kernels for pairs of stations in interferometry, thus significantly reducing the effect of non-isotropic intensity illumination and non-stationary noise sources. We use the differential sensitivity kernel based on multiple virtual sources in an iterative inversion for surface wave phase velocity and from that derive the shear velocity. The differential adjoint tomography results are robust with respect to noise source distributions and provide faster convergence than for adjoint tomography based on station-pair travel time kernels.