Extraction And Application Of Green’s Function In Random Fluctuating Ocean

Ocean ambient noise and acoustic scattering is one important reason of random fluctuation of sound field. Although it is considered that the background field is a nuisance traditionally, there are more and more studies which extract ocean environment information from background field in recent years. The main research of this thesis is to extract and apply Green’s function from random fluctuant ocean environment.Above all, ambient noise interferometry or Green’s function retrieval is considered. This theory indicates that the time domain Green’s function of two points can be recovered from cross-correlations of diffuse noise recorded at these two points. Given the non-diffusivity of ocean ambient noise, a new data preprocessing method called diffuse noise filtering (DNF) is proposed on the basis of KL transform and random matrix theory (RMT). Experimental results demonstrates that the DNF can reconstruct diffuse noise component from unideal ocean ambient noise, enhance Green’s function retrieval and passive fathometer processing.Then, passive scatter imaging is studied by numerical simulation and experimental data. On account of time reverse D.O.R.T and MUSIC algorithm cease disabled when Green’s function retrieval is unsatisfactory, the conventional Kirchhoff emigration imaging method is improved which bases on compressive sensing theory. And it produces good passive scatter imaging when it is applied to experimental data.Finally, bottom backscattered echoes of 41 sites located in the north Yellow Sea and Bohai Sea were acquired by a vertically oriented echosounder working at 20 kHz. Based on a time domain incoherent intensity model of seafloor high frequency backscattering, the sediment classification with inversion of seafloor sediment geoacoustic parameters is investigated. By maximum likelihood matching between the averaged echo envelope and the model output with a global simulated annealing- downhill simplex optimization (SIMPSA), the sediment mean grain size is estimated. It is found that the fluctuation of the echo and the variance of the estimated parameters are effectively reduced once the echo envelope is normalized by the total echo energy before calculating the averaged envelope. And the estimated parameters are consistent with the ground truth. Simultaneously, the percentage of totally correct classification is 75.6%.