Inverse problem of Ocean Acoustic Tomography (OAT) - A numerical experiment

Abstract

Acoustic model simulation experiments related to the forward and inverse aspects of ocean tomography have been taken up with a view to estimate the vertical sound speed field by inverting the travel time data. Two methods of inversion have been attempted. These are: (1) deterministic inverse and (2) stochastic inverse. To solve the former, singular value decomposition (SVD) is used to arrive at the model parameter(s). The data kernel consisting of acoustic path lengths in a pre-set number of ocean layers, or grids developEd. by solving the forward problem of the acoustic model enable build the generalized inverse operator (GIO) that operates on the travel time perturbation data. Resolution matrices obtained through SVD helped to examine the correctness of the solution. In the stochastic inverse case, for random model parameter and data, the vertical structure of the ocean is modeled through the quasi-geostrophic theory while the horizontal structure is assumed to have Gaussian covariance, with length scales of 100 km. In both cases, a good agreement has been observed between the assumed and the reconstructed sound speed perturbation field