Inversion of Acoustic Field Data for Bathymetry and Bottom Sound Speed via Simulated Annealing

Abstract

Geoacoustic models of the ocean bottom involving parameters such as bathymetry and bottom sound speed are required for reliable acoustic propagation modelling and matched field processing. Bottom parameters can be difficult to determine in range-dependent environments, particularly under Arctic ice cover. We are investigating methods of inverting acoustic field data to obtain estimates of these parameters. The data typically consist of measurements of the total acoustic field due to sources at known ranges, as measured in a standard propagation experiment. The inversion is formulated as an optimization problem by assuming a discrete search space for the geoacoustic model, and seeking the set of parameter values which minimizes the mismatch with the measured fields. The optimization is solved using the method of simulated annealing. This is accomplished in an efficient manner by computing the matching fields with the adiabatic normal mode model and making use of a pre-computed ‘look-up’ table of mode values. Several examples of inversion in realistic Arctic environments are presented.