On the limits of distinguishing seabed types via ambient acoustic sound.

This article presents a theoretical analysis of optimally distinguishing among environmental parameters from ocean ambient sound. Recent approaches to this problem either focus on parameter estimation or attempt to classify the environment into one of many known types through machine learning. This classification problem is framed as one of hypothesis testing on the received ambient sound snapshots. The resulting test depends on the Kullback-Leibler divergence (KLD) between the distributions corresponding to different environments or sediment types. Analysis of the KLD shows the dependence on the signal-to-noise ratio, the underlying signal subspace, and the distribution of eigenvalues of the respective covariance matrices. This analysis provides insights into both when and why successful hypothesis testing is possible. Experiments demonstrate that our analysis provides insight as to why certain environmental parameters are more difficult to distinguish than others. Experiments on sediment types from the Naval Oceanographic Office Bottom Sediment type database show that certain types are indistinguishable for a given array configuration. Further, the KLD can be used to provide a quantitative alternative to examining bottom loss curves to predict array processing performance.

Optimal environmental estimation with ocean ambient noise.

This article presents an asymptotically optimal technique for estimating environmental parameters from ocean ambient noise. Noise from wind and breaking waves propagates through the water column and reflects off the bottom over a wide range of angles and frequencies and, in doing so, imparts information about the environment to the noise covariance matrix for a receiver array. Most environmental estimation techniques focus on spatial filtering methods aimed at recovering the vertical noise directionality. However, an often overlooked fact is that the noise covariance matrix fully characterizes the probability density function of each snapshot, which forms the basis for an information-theoretic approach. In this light, it is possible to obtain the theoretical bounds on optimal estimator performance while also providing a basis for assessing the utility of different parameterization schemes. Most importantly, it provides a natural definition for a maximum likelihood estimator that meets the optimal bounds in an asymptotic sense. This technique outperforms beamforming-based methods by a significant margin. It also remains unbiased in the presence of strong white noise, is tolerant to array tilt, can operate beyond the array design frequency, but does suffer greater sensitivity to model mismatch. These trade-offs are explored with simulations and analyses of experimental data.

Environmental information content of ocean ambient noise.

In recent years, methods have been developed to estimate a variety of environmental parameters based on measurements of the ocean ambient noise. For example, noise has been used to estimate water depth using the passive fathometer technique and bottom loss estimated and used to invert for seabed parameters. There is also information in the noise about the water column sound speed, volume attenuation, and the sea-state. The Fisher information can be used to quantify the basic information available in the noise measurements and its inverse, the Cramér-Rao lower bound (CRLB), provides the lower limit on the variance of an unbiased estimator of a particular parameter. The CRLB can be used to study the feasibility of various measurement configurations and parameter sensitivities. In this paper, the CRLB is developed for ocean ambient noise and the environmental information contained in the measurements is determined. The CRLBs provide an estimate of the underlying information in the data, however, it is independent of the estimation methodology. This is useful to determine if a given estimation method is reaching the lower bound. Results illustrating the bounds as well as sensitivities and performance of estimators are demonstrated using both simulations and data.

An analysis of beamforming algorithms for passive bottom reflection-loss estimation.

This study provides an argument cautioning against the use of adaptive-beamforming (ABF) techniques in conjunction with a known method for estimating the bottom reflection loss from natural marine ambient noise. This application of ABF has been investigated in the past with rather inconsistent results. Furthermore, no formal proof that ABF algorithms do indeed provide an estimate of the bottom reflection loss is available. This study moves from a recent derivation of the relationship between the bottom reflection coefficient and the Fourier transform of the marine-noise spatial coherence function. The circumstances under which the beamforming operation approximates a discrete Fourier transform (DFT) of the spatial coherence function estimated from array data are analyzed. It is shown that, under certain conditions, conventional beamforming is equivalent to directly computing the DFT of the coherence function, as long as some subtle details are properly taken into account. Furthermore, it is shown that ABF cannot be guaranteed, in general, to perform this operation, and therefore provide an estimate of the bottom reflection coefficient. The conclusions are demonstrated on simulated and measured data.

Head wave correlations in ambient noise.

Ambient ocean noise is processed with a vertical line array to reveal coherent time-separated arrivals suggesting the presence of head wave multipath propagation. Head waves, which are critically propagating water waves created by seabed waves traveling parallel to the water-sediment interface, can propagate faster than water-only waves. Such eigenrays are much weaker than water-only eigenrays, and are often completely overshadowed by them. Surface-generated noise is different whereby it amplifies the coherence between head waves and critically propagating water-only waves, which is measured by cross-correlating critically steered beams. This phenomenon is demonstrated both experimentally and with a full wave simulation.

A two-hydrophone range and bearing localization algorithm with performance analysis.

An automated, passive algorithm for detecting and localizing small boats using two hydrophones mounted on the seabed is outlined. This extends previous work by Gebbie et al. [(2013). J. Acoust. Soc. Am. 134, EL77 - EL83] in which a similar two-hydrophone approach is used to produce an ambiguity surface of likely target locations leveraging multipath analysis and knowledge of the local bathymetry. The work presented here improves upon the prior approach using particle filtering to automate detection and localization processing. A detailed analysis has also been conducted to determine the conditions and limits under which the improved approach can be expected to yield accurate range and unambiguous bearing information. Experimental results in 12 m of water allow for a comparison of different separation distances between hydrophones, and the Bayesian Cramér-Rao lower bound is used to extrapolate the performance expected in 120 m water. This work demonstrates the conditions under which a low cost, passive, sparse array of hydrophones can provide a meaningful small boat detection and localization capability.

Passive localization of noise-producing targets using a compact volumetric array.

A technique is presented for passively localizing multiple noise-producing targets by cross-correlating the elevation beams of a compact volumetric array on separate bearings. A target's multipath structure inherently contains information about its range; however, unknown, random noise waveforms make time separation of individual arrivals difficult. Ocean ambient noise has previously been used to measure multipath delays to the seabed by cross-correlating the beams of a vertical line array [Siderius, Song, Gerstoft, Hodgkiss, Hursky, and Harrison, J. Acoust. Soc. Am. 127, 2193-2200 (2010)], but this methodology has not been applied to distant noise sources having non-vertical arrivals. The technique presented in this paper uses a compact volumetric array mounted to an autonomous underwater vehicle to measure the three-dimensional directionality and time delays of multipath arrivals, while adaptively rejecting clutter and multi-target interference. This is validated with experimental results in a shallow ocean environment in which a small workboat maneuvered in the vicinity. Short ranges could be estimated reliably using straight ray paths, but longer ranges required accounting for ray refraction.

Localization of a noisy broadband surface target using time differences of multipath arrivals.

Previous studies [Tiemann et al., J. Acoust. Soc. Am. 120, 2355-2365 (2006)] have reported the localization of marine mammals in 3-D from their clicks using multipath arrivals. Bathymetric variations were advantageously used to predict multipath arrival times with a raytracer. These arrivals are directly discernible from the time series for impulsive sources, such as whale clicks, but extension of the method to continuous broadband sources presents additional complications. By pulse compressing noise emitted from a small boat using two hydrophones, the hyperbolic direct-arrival ambiguity can be refined in both range and bearing. Acoustic-derived results are validated with target GPS measurements.

Aspect-dependent radiated noise analysis of an underway autonomous underwater vehicle.

This paper presents an analysis of the acoustic emissions emitted by an underway REMUS-100 autonomous underwater vehicle (AUV) that were obtained near Honolulu Harbor, HI using a fixed, bottom-mounted horizontal line array (HLA). Spectral analysis, beamforming, and cross-correlation facilitate identification of independent sources of noise originating from the AUV. Fusion of navigational records from the AUV with acoustic data from the HLA allows for an aspect-dependent presentation of calculated source levels of the strongest propulsion tone.