Numerical solution of the inverse kinematic problem in a linearized formulation for inhomogeneous media using refractive and scattered waves
Abstract
The inverse kinematic problem for vertically inhomogeneous media, in which wave propagation velocity depends only on depth, has been investigated using refracted and scattered wave signals. Unlike classical formulations of inverse kinematic problems, which primarily focus on direct or reflected waves, this study centers on scattered waves – those whose trajectories have been altered due to interaction with local medium inhomogeneities. This approach extends the applicability of the method to media exhibiting waveguide effects, where waves are “trapped” within a layer and propagate over long distances with minimal losses. A computational algorithm based on the linearization of the Fermat functional has been developed, enabling determination of the velocity distribution V?(z) in such media. For the direct problem, both analytical solutions (in the linear case V?(z) = 1 + Bz) and numerical methods (for nonlinear cases), including the 4th-order Runge-Kutta method, have been applied. The implemented solution demonstrated the feasibility of reconstructing the medium velocity distribution from measured wave arrival times, while also revealing errors associated with the inaccuracy of linearization. The analysis showed that linearization error increases with depth, which, in turn, may be compensated by introducing additional regularization procedures – for example, noise smoothing using splines. The obtained results may be of interest for solving problems in seismology, acoustic sounding, and non-destructive testing. Prospects for further research are primarily associated with transitioning to more complex medium models, applying more advanced regularization techniques, and hybridizing this approach with artificial intelligence models – all of which open prospects for expanding the method’s applicability and resolution capability.