This paper examines array gain and detection performance of single vector sensors and vector sensor line arrays, with focus on the impact of nonacoustic self noise and finite spatial coherence of the noise between the vector sensor components. Analytical results based on maximizing the directivity index show that the particle motion channels should always be included in the processing for optimal detection, regardless of self noise level, as long as the self noise levels are taken into account. The vector properties of acoustic intensity can be used to estimate the levels of nonacoustic noise in ocean measurements. Application of conventional, minimum variance distortionless response, and white-noise-constrained adaptive beamforming methods with ocean acoustic data collected by a single vector sensor illustrate an increase in spatial resolution but a corresponding decrease in beamformer output with increasing beamformer adaptivity. Expressions for the spatial coherence of all pairs of vector sensor components in homogeneous, isotropic noise show that significant coherence exists at half-wavelength spacing between particle motion components. For angular intervals about broadside, an equal spacing of about one wavelength for all components provides maximum directivity index, whereas each of the component spacings should be different to optimize the directivity index for angular intervals about endfire.

No abstract

In this article, a method of passively localizing a narrow-band source in range and depth in a waveguide is presented based on "matching" predicted normal mode amplitudes to measured mode amplitudes. The modes are measured by using a vertical array of hydrophones and performing mode filtering. Previous studies of mode filtering have considered only the overdetermined case, i.e., where there are more hydrophones than discrete modes present in • the waveguide. In this study, mode filtering is considered for the underdetermined case, i.e., where there are fewer hydrophones than the total number of discrete modes in the waveguide, but only a subset of the total number of modes is to be estimated. Previous studies of matched field localization have been based on matching the entire pressure field. In this study, the pressure field is expressed in terms of normal modes, and only a subset of the total number of modes is used for localization. Using a subset of modes allows trade-offs to be made between localization accuracy, computational complexity, and sensitivity to environmental mismatch. In this article, the matched mode localization method is presented, and the dependence of its localization accuracy on the number of modes used and environmental conditions is demonstrated. The effects of array length and hydrophone spacing on mode estimation error, and hence on localization accuracy, are also demonstrated for the particular method of mode estimation used here. Other methods of mode estimation may produce different results. Finally, the effects of mismatch between the assumed and actual environment due to water depth variation are explored. It is shown that localization accuracy in range is proportional to the mode interference distance between the lowest and highest modes used to localize, and that as few as six modes can be used for ranging. It is also shown that the array length need not be any longer than the depth extent of the highest mode to be estimated, and that the hydrophone spacing must be no greater than half the vertical wavelength of the highest mode that contributes significantly to the sound field (not just the highest mode to be estimated). Localization is most sensitive to environmental mismatch effects that contribute to changes in the phase of the horizontal component of the mode amplitudes. Because a subset of modes is used for localization instead of the entire pressure field, this method of localization can be fairly insensitive to certain kinds of environmental mismatch.

No abstract

Bispectral analysis is a statistical tool for detecting and identifying a nonlinear stochastic signal-generating mechanism from data containing its output. Bispectral analysis can also be employed to investigate whether the observed data record is consistent with the hypothesis that the underlying stochastic process has Gaussian distribution. From estimates of bispectra of several records of ambient acoustic ocean noise, a newly developed statistical method for testing whether the noise has a Gaussian distribution, and whether it contains evidence of nonlinearity in the underlying mechanisms generating the observed noise is applied. Seven acoustic records from three environments are examined: the Atlantic south of Bermuda, the northeast Pacific, and the Indian Ocean. The collection of time series represents both ambient acoustic noise (no local shipping) and noise dominated by local shipping. The three ambient records appeared to be both linear and Gaussian processes when examined over a period on the order of a minute, but were found to be nonlinear and non-Gaussian when examined over shorter time periods on the order of a second. In each case the time series dominated by local shipping noise tested to be nonlinear and non-Gaussian over both short and long time periods.