Santa Barbara Channel Experiment Research Articles

Focusing on motion and averaging the diagonal: Lisa Zurk’s many contributions to passive sonar array processing

Lisa Zurk’s numerous contributions to passive sonar signal processing include motion compensation and regularizing sample covariance matrices (SCM) in snapshot deficient environments. While working at MIT Lincoln Labs, Lisa and colleagues compared the performance of several motion compensation methods including mode-based rank reduction and target motion compensation, or focusing, on the Santa Barbara Channel Experiment dataset [Zurk, Lee & Ward, JASA, 2003]. Later at Portland State, Jorge Quijano and Lisa developed methods for regularizing the SCM through Toeplitz averaging, followed by maximum entropy extrapolation of the SCM to additional lags, [Quijano & Zurk, JASA, 2017]. Toeplitz regularization found significant application in developing “augmented covariance matrices” for DOA estimators on sparse arrays. This talk reviews some of Lisa’s contributions to adaptive beamforming and highlights how some of her contributions inspired research in the UMass Dartmouth Signal Processing Group. [Work supported by ONR Code 321US.]

Moving ship tomography in the Santa Barbara Channel Experiment 2016

Ocean acoustic travel-time-tomography attempts to map heterogeneities such as ocean temperature with mesoscale resolution. The Santa Barbara Channel Experiment in 2016 (SBCEx16) contained both controlled and uncontrolled (i.e., ships of opportunity) sources, in addition to environmental data collected by a variety of sensors. The path to developing a methodology for utilizing noise from ships of opportunity for oceanographic imaging involves benchmarking our processing with a controlled source-tow. This latter moving source tomography, itself, is a procedure that still remains to be quantitatively validated and documented. We present results from the source-tow tomographic component of SBCEx16 with the added objective of understanding the requirements for minimizing the uncertainties associated with uncontrolled, random sources of opportunity.

Coherent processing of ocean ambient noise for sediment profiling and head-wave detection using a vertical line array in the Santa Barbara Channel

In this study, sediment profiling using the passive fathometer and the head-wave phenomenon are investigated. Head waves are created by waves propagating in sediments parallel to the water-sediment interface (possibly at a speed higher than the water sound speed), and propagating back in the water at critical angles. Ocean ambient noise recorded by a vertical line array is processed using a generalization of the known “passive fathometer.” The technique can detect coherent time-separated arrivals at specific angles that are consistent with the propagation of head waves. The arrivals structure detected at endfire provides information on the sediment layering, while the head wave can be used to estimate the sediment sound speed, and may provide information about the attenuation. The technique is demonstrated using simulation and data from the Santa Barbara Channel Experiment of 2016, using both natural noise generated at the surface, and—for the first time—the noise generated by a ship. [Work supported by the Office of Naval Research, Ocean Acoustics Program.]

Ship localization in Santa Barbara Channel using machine learning classifiers.

Machine learning classifiers are shown to outperform conventional matched field processing for a deep water (600 m depth) ocean acoustic-based ship range estimation problem in the Santa Barbara Channel Experiment when limited environmental information is known. Recordings of three different ships of opportunity on a vertical array were used as training and test data for the feed-forward neural network and support vector machine classifiers, demonstrating the feasibility of machine learning methods to locate unseen sources. The classifiers perform well up to 10 km range whereas the conventional matched field processing fails at about 4 km range without accurate environmental information.

Active vs passive moving source tomography: Comparing results from the Santa Barbara Channel Experiment (SBCEx16) on sources of opportunity

An experiment was performed in the Santa Barbara Channel (SBCEx16) using four vertical line arrays (VLAs) of hydrophones placed midway between the in- and out-going shipping lanes of a maritime highway. The goal of this experiment was to study whether these sources of opportunity can be used for passive tomographic purposes. In addition to the continuous passive observations, active acoustic source-tows were conducted. The environment was monitored continuously throughout the course of the experiment using thermistor strings, and ship tracks were obtained from the Automatic Identification System (AIS). The results from both types of tomography are presented and compared.

Channel impulse response arrival uncertainty using source of opportunity for tomography

Passive acoustic tomography exploits the acoustic energy generated by sources with unknown spectral content such as sources of opportunity (e.g., cargo ships) to study the ocean. The recording at each sensor within a vertical line array (VLA) is the channel impulse response (CIR) convolved with the noise generated by the moving random radiator. Using an incremental approach, we estimate the source signal locally at three VLAs by beamforming on the direct ray-path to deconvolve each CIR. CIR arrival uncertainty is inversely proportional to the bandwidth of the source and SNR, the latter estimated from the deconvolved time domain waveform. Over the 10 min source track, we present the time evolution of CIR arrival uncertainty computed at three VLAs horizontally separated by 1.5 km and discuss constraints on integration time. Data are presented using the Noise Correlation 2009 Experiment and application to the Santa Barbara Channel Experiment 2016 is discussed.

Toward underwater channel impulse response estimation using sources of opportunity

Passive acoustic tomography exploits the acoustic energy generated by sources with unknown spectral content such as sources of opportunity (e.g., cargo ships) to study the ocean. However, separating the channel impulse response (CIR) from a possibly moving, random radiator is non-trivial for coherent processing and the amount of unknown variables may leave the problem intractable. Using an incremental approach, we study the sparse CIR estimation problem for a single receiver. Second, we add multiple receivers, relax source spectrum knowledge, and study processor performance in resolving possibly coherent multipath arrivals across the vertical line array. Third, we discuss the extension to include multiple VLAs and sequential processing for a source with known location. Application of the Noise Correlation 2009 Experiment and the Santa Barbara Channel Experiment 2016 are discussed.

Source motion mitigation for adaptive matched field processing.

Application of adaptive matched field processing to the problem of detecting quiet targets in shallow water is complicated by source motion, both the motion of the target and the motion of discrete interferers. Target motion causes spreading of the target peak, thereby reducing output signal power. Interferer motion increases the dimensionality of the interference subspace, reducing adaptive interference suppression. This paper presents three techniques that mitigate source motion problems in adaptive matched field processing. The first involves rank reduction, which enables adaptive weight computation over short observation intervals where motion effects are less pronounced. The other two techniques specifically compensate for source motion. Explicit target motion compensation reduces target motion mismatch by focusing snapshots according to a target velocity hypothesis. And time-varying interference filtering places time-varying nulls on moving interferers not otherwise suppressed by adaptive weights. The three techniques are applied to volumetric array data from the Santa Barbara Channel Experiment and are shown to improve output signal-to-background-plus-noise ratio by more than 3 dB over the standard minimum-variance, distortionless response adaptive beam-former. Application of the techniques in some cases proves to be the difference between detecting and not detecting the target.

Analytic prediction of sidelobe statistics for matched-field processing

Underwater source localization using matched-field processing (MFP) is complicated by the relatively high sidelobe levels characteristic of MFP ambiguity surfaces. An understanding of sidelobe statistics is expected to aid in designing robust detection and localization algorithms. MFP sidelobe levels are influenced by the underwater channel, array design, and mismatch between assumed and actual environmental parameters. In earlier work [J. Acoust. Soc. Am. 108, 2645 (2000)], a statistical approach was used to derive analytic expressions for the probability distribution function of the Bartlett ambiguity surface. The distribution was shown to depend on the orthogonality of the mode shapes as sampled by the array. Extensions to a wider class of array geometries and to broadband processing will be shown. Numerical results demonstrating the accuracy of the analytic results and exploring their range of validity will be presented. Finally, analytic predictions will be compared to data from the Santa Barbara Channel experiment. [Work sponsored by DARPA under Air Force Contract F19628-00-C0002. Opinions, interpretations, conclusions, and recommendations are those of the authors and are not necessarily endorsed by the Department of Defense.]

Source motion compensation using waveguide invariant theory

Passive detection and localization of quiet, submerged sources moving in shallow water in the presence of shipping noise is a critical problem for the ASW community. Past work has demonstrated that adaptive processing can be applied to detect and localize stationary sources while mitigating loud interferers. However, source motion can cause significant signal gain degradation, loss of localization accuracy, and incomplete interference rejection. The severity of the motion loss is channel dependent, and a function of the amount of motion during the observation interval. Motion compensation techniques have been demonstrated in past work with data from the Santa Barbara Channel Experiment. These algorithms utilized a normal mode propagation model and a velocity hypothesis to successfully compensate for source motion. However, the technique requires good environmental knowledge and can be computationally expensive if a large hypothesis space must be searched. In this paper, the waveguide invariance principle is used to combine the time-frequency snapshots acquired from a broadband source to compensate for motion. The technique does not require knowledge of the underwater environment for the compensation and the search across the velocity space can be structured so it is computationally efficient. Results from this technique will be presented and discussed.

A comparison of modal, beam, and PE modeled replicas for matched-field processing in the Santa Barbara Channel

In order to evaluate the use of different propagation models in matched-field processing (MFP) applications, source localization was attempted using data from the Santa Barbara Channel Experiment and replicas generated by the adiabatic normal mode model KRAKEN, the beam model BELLHOP, and the parabolic equation model RAM. A conventional matched-field processor using all these models successfully located the source position in range and depth at both short (1 km) and intermediate (7 km) range, with marginal results at the longest 15 km range. Because signal-gain degradation varies with the model used, simulations were performed to quantify the degradation and relate that amount to a model mismatch, i.e., an error in array element localization.

Matched field inversion with a moving source in a shallow-water environment

Geoacoustic inversion based on matched field processing (MFP) is examined for shallow-water, low-frequency environments. Specific interest lies with resolving geoacoustic parameters from cw tones projected from a moving source. Data obtained using vertical line arrays (VLA) are available from both the 1996 Haro Strait PRIMER Experiment (HSX) and the Santa Barbara Channel Experiment (SBCX) of 1998. The environmental complexity associated with these experiments, namely, strong range-dependent bathymetry and current flow, is typical to littoral environments in general and must be appropriately addressed. Parabolic equation modeling is used as it provides range-dependent results. The VLA receiver configuration is described as a catenary which is typical of bottom-anchored arrays drifting in uniform current flow. Both SBCX and HSX are useful for benchmarking geoacoustic inversion techniques since results from other techniques are available. Estimation of bottom properties is discussed as a function of propagation range, ship track with respect to receiver position, and general bathymetric features. Results for both tangential and radial tracks are presented. [Work supported by ONR.]

Geoacoustic parameters by inversion of bottom reflectivity data from the Santa Barbara channel experiment

A method is described to estimate the range dependence of geoacoustic parameters of the seafloor from bottom reflectivity data. The data were obtained using light bulbs as sound sources in the Santa Barbara channel experiment (SBCX). Light bulbs were deployed around the SBCX FFP array that consisted of five 30-element vertical line arrays. Signals from the direct path, the surface reflected path, and the four components of the first bottom reflection were clearly resolved at each array. The inversion integrates the results from several source–array combinations to generate estimates of the seafloor geoacoustic parameters in the area around the FFP array. The reflection coefficients were determined versus angle from the direct path and bottom-reflected path signals, and inverted using a simple fluid model to estimate the seafloor sound speed and density. The estimates are consistent with the known ground truth information for the SBCX site. [Work supported by ONR.]

Array design and motion effects for matched field processing

While matched field processing (MFP) algorithms have most commonly been applied to data collected on vertical line arrays, there is increasing interest in applying MFP to more general array geometries. One challenge for novel array design is to determine the fundamental performance of volumetric arrays in a shallow water channel. To address this issue, recent work by Tantum and Nolte [J. Acoust. Soc. Am. 107, 2101–2111 (2000)] has been extended to allow more general array geometries, including horizontal arrays steered away from endfire. The correlation among acoustic modes across the array can be related to MFP resolution and sidelobe levels. Another trend in underwater array design is the consideration of larger arrays which provide additional array gain. However, large arrays are susceptible to the effects of motion [Zurk et al., Oceans 99]. A new time series simulation that rigorously treats source and receiver motion is used to quantify the effects on adaptive MFP. Simulation results are compared to data from the Santa Barbara Channel experiment (SBCX). [This work was sponsored by DARPA under Air Force Contract F19628-95-C0002. Opinions, interpretations, conclusions, and recommendations are those of the authors and are not necessarily endorsed by the United States Air Force.]

Model-based and data-based processing of moving sources in a shallow water channel

Adaptive MFP algorithms are applied to passively detect and localize a moving source while mitigating interference from surface shipping. A model-based motion compensation algorithm is applied to prevent smearing loss from target motion and extend the observation interval for adaptation. Moving interference is removed by applying a time-varying spatial filter at each data snapshot. The time-variation of the filter decreases the size of the instantaneous interference subspace and allows for a larger signal subspace. The interference subspace is formed by one of three methods. In the first method, knowledge of the ship position is combined with a propagation model to predict the time-varying acoustic interference. A sector null is formed to protect against mismatch degradation. In the second method, a data decomposition is used to determine the strong interferers. The third method combines model-based and data-based approaches. For each method, interference rejection is quantified by presenting SINR values for the candidate algorithm and comparing to conventional approaches. The results use data obtained from the Santa Barbara Channel Experiment. [This work was sponsored by DARPA under Air Force Contract F19628-95-C00022E. Opinions, interpretations, conclusions, and recommendations are those of the author and are not necessarily endorsed by the United States Air Force.]

Broadband adaptive matched field processing with modal filtering

This paper will demonstrate 3D localization of a moving, broadband source in a shallow water channel using adaptive matched field processing (AMFP) with modal filtering. The adaptation is desirable to mitigate the high sidelobes of the conventional processor, but it also introduces the requirement for long observation times, which can be difficult to satisfy for moving sources. Rank reduction can be achieved through modal preprocessing by exploiting the sparsity of propagating modes. In addition, the modal decomposition provides a mechanism for filtering the signatures of surface interferers from those of a submerged source. Results will be presented that demonstrate rejection of surface interference as well as quantifying the relationship between localization accuracy and the size of the modal subspace. The acoustic data was obtained during the Santa Barbara Channel Experiment (SBCX) which employed a 150-hydrophone volumetric array consisting of five vertical line arrays (VLAs) instrumented for element localization. The VLAs fully spanned the 209-m water column and allowed accurate modal decomposition. The MFP results will be presented for exercises involving a research vessel, the Acoustic Explorer (AX), towing an acoustic source. The results of broadband AMFP show accurate localization of both the towed source and towship.

Source localization in a stochastic ocean using data from the Santa Barbara Channel experiment

Source localization performance using matched field processing is degraded when: (1) incorrect environmental information is used in replica generation; and (2) the propagation model fails to correctly predict energy transfer from one spatially derived eigenvalue to another. Krolik [J. Acoust. Soc. Am. 92] addressed (1) by developing a robust adaptive matched field processor assuming a stochastic propagation medium and a perturbed sound speed profile. Czenszak [J. Acoust. Soc. Am. 101] extended Krolik’s work to include multiple narrow-band frequencies. We approach both (1) and (2) by direct decomposition of the simulated pressure field based on environmental perturbations. This yields information on energy distribution from a single point source on spatially derived eigenvalues in a simulated environment. We will explore the application of both processors and extensions of them to range-dependent shallow water acoustic data taken from the 1998 Santa Barbara Channel experiment. Source localization performance of the stochastic ocean beamformer is contrasted with both Bartlett and White Noise Constrained processors at ranges to 15 km, with emphasis on performance metrics in the face of uncertain environmental knowledge.

Tracking results from the three-dimensional full field processing (FFP) array and the Santa Barbara Channel experiment

In March of 1998, SAIC under DARPA sponsorship deployed five 170-m, 30-element, vertical line arrays (VLAs) in 200 m of water in the Santa Barbara Channel. The VLAs were arranged in a pentagon with a 200-m horizontal aperture. Each hydrophone package contained tilt sensors for array localization as well as a thermistor for water temperature measurement. Each VLA also contained four compasses that were used in conjunction with the tilt sensors for passive localization of the array elements. In this presentation the acoustic experiments will be described and a number of results including a comparison of acoustic and nonacoustic localization methods, an environmental characterization, and the results of acoustic measurements using a towed source and surface ships of opportunity will be presented. Specific acoustic results to be presented will include measurements of array signal and noise gain, and the results of tracking the towed source and surface ships using adaptive matched field methods.