Traveltime tomography has been extensively applied in both global tomography and seismic exploration. The traveltime Fréchet derivative, derived from the Born approximation, is only valid for weak perturbations and small phase-shifts. Although the small phase-shift restriction can be handled with the first-order Rytov approximation, the weak velocity perturbation assumption is still a major limitation. To break the weak-perturbation limitation, we combine the recently developed generalized Rytov approximation and the finite frequency theory, and propose a new finite-frequency traveltime sensitivity kernel, which is valid for arbitrary velocity perturbation as long as the scatter-angles are small. Then, we solve the associated finite-frequency tomography problem with the Gauss-Newton algorithm. To obtain the Gauss–Newton descent direction, we present a matrix-free Hessian-vector product method, which can avoid the explicit calculation of the Hessian matrix or the sensitivity kernels. Numerical examples demonstrate that the traveltime-shifts predicted by the proposed sensitivity kernel are more accurate than the results obtained by the first-order Rytov approximation, which can further improve the inversion accuracy.
Presentation Date: Wednesday, September 18, 2019
Session Start Time: 1:50 PM
Presentation Time: 4:45 PM
Presentation Type: Oral