Range-dependent geoacoustic inversion using the parabolic equation and genetic algorithms - application to GAIT test cases

Abstract

Rapid in-situ assessment of sonar conditions in shallow water is an area of interest for military applications. The goal is through-the-sensor techniques that collect acoustic data from sonar, infer the relevant ocean environment parameters, and provide environment model data for subsequent use in sonar performance predictions or in environmentally adapted signal processing. A successful inversion method requires good solutions to several problems: how to efficiently parameterize the ocean environment, how to identify the most relevant parameters for the given mode of sonar operation, and how to most efficiently obtain estimates of geoacoustic parameters from given sonar data. In this report, an inverse method is applied to synthetic low-frequency data for two shallow-water range-dependent environments, consisting of a range-varying bathymetry over a laterally invariant seabed with depth-dependent sound speed, density and attenuation profiles. (The first two GAIT workshop 2001 test cases.) The SAGA inversion package from NURC is used with the RAM forward model from NRL. Low-frequency data within 50 Hz to 400 Hz is used for matched-field inversion on two model linear arrays (a vertical array and a horizontal array). It is found that the sound speed profiles in seabed can be estimated from acoustic data, whereas density and attenuation profiles are not well estimated. The long horizontal array provides results comparable to those of a VLA spanning the water column. A limiting factor to further applicability of range-dependent inversion methods is the execution time induced by the forward model.