Knowledge of the geoacoustic properties of the ocean bottom is essential for accurate modeling of acoustic propagation in shallow-water environments. Estimates of these properties can be obtained through geoacoustic inversion. Among the various inversion methods, the ones based on matched-field processing (MFP) have been increasingly used due to their relatively easy implementation and their good performance. In matched-field inversion (MFI), the objective is to maximize the match between the measured acoustic pressure field and the modeled field calculated for trial sets of geoacoustic parameters characterizing the environment. This thesis investigates the technique of matched-field tomographic inversion, a recent application of MFI that takes advantage of a multiple array-multiple source configuration to estimate range-dependent geoacoustic parameters. A two-stage inversion method based on the ray approach adopted to calculate the modeled pressure fields is developed to increase the efficiency of the estimation. The first stage consists of matching measured and modeled amplitudes of waterborne rays propagating between each source-array pair to estimate the parameters at the seafloor. The second stage consists of matching measured and replica pressure fields corresponding to rays that penetrate the sediment to estimate deeper parameters. In the first stage, the match is quantified using a least-squares function whereas in the second stage the robust pairwise processor is used. Both stages use a simplex genetic algorithm to guide the search over the parameter space. The inversion method is first applied to the two-dimensional (2-D) problem of vertical-slice tomography where four sets (2 sources x 2 vertical line arrays) of multi-tone pressure fields are used to estimate the depth and range variations of geoacoustic parameters. The method is validated via simulation studies that show its good performance in the ideal case where every model parameter except the ones to be estimated are exactly known, and quantify its limitations in non-ideal cases where noise in the data or errors in the array positions are present. The inversion results show that the parameters to which the pressure field is the most sensitive are well estimated for signal-to-noise ratios greater than or equal to 5 dB or for array position uncertainties less than two wavelengths of the source wavelet. The inversion method is then applied to a 3-D environment problem. From the different array configurations studied, it is found that the accuracy of the parameter estimates increases with decreasing propagation range. Finally, the method is applied to experimental data for a vertical-slice configuration. The relatively poor match obtained between the replica and measured data is attributed to the large uncertainty in the array position and the simplistic parameterization of the environment. / Graduate
| Identifer | oai:union.ndltd.org:uvic.ca/oai:dspace.library.uvic.ca:1828/9920 |
| Date | 16 August 2018 |
| Creators | CorreĢ, Vanessa |
| Contributors | Chapman, N. Ross |
| Source Sets | University of Victoria |
| Language | English, English |
| Detected Language | English |
| Type | Thesis |
| Format | application/pdf |
| Rights | Available to the World Wide Web |
Page generated in 0.0017 seconds