Aperture Vertical Array Research Articles

Very-low-frequency under-ice reflectivity

This paper describes a direct method to model under-ice reflection loss from analysis of ice draft data taken from a region of the Arctic near the FRAM IV experiment site. The water–ice boundary is modeled as a random distribution of infinite elliptical half-cylinders fixed to a free surface. Burke and Twersky’s theory [J. Acoust. Soc. Am. 40, 883–895 (1966)] of scattering from a single cylindrical protuberance is used to calculate reflectivity from a distribution of scatterers. Individual ice ridge keels are identified from ice draft data resulting in a ridge keel depth distribution function spatially coincident with FRAM IV acoustic propagation paths. The scattering amplitude is weighted by the keel depth distribution function providing an effective ridge keel depth which is used to calculate the under-ice reflectivity. The predicted reflection loss is in good agreement with those inferred from normal-mode methods applied to FRAM IV acoustic field data received on a larger aperture vertical array.

Sound-speed profile inversion using a large aperture vertical line array

In this paper, the single vertical slice ocean acoustic tomography problem is investigated with a large aperture vertical line array and a narrow-band low-frequency acoustic source deployed at a known location. Within the framework of matched-field processing and adiabatic normal mode theory, the self-cohering technique of Bucker is shown to provide a smooth measure of sound-speed mismatch. Simple simulation results lead to the development of a minimization procedure to invert for the sound-speed profile using a good starting solution (e.g., based on historical data). The technique then is demonstrated with additional simulations for sound-speed structures characteristic of the NE Pacific.

Spatial smoothing and minimum variance beamforming on data from large aperture vertical line arrays

Various approaches to the beamforming of data from large aperture vertical line arrays are investigated. Attention is focused on the conventional beamforming problem where the angular power spectrum is estimated, in this case by the adaptive minimum variance processor. The data to be processed are 200 Hz CW transmissions collected at sea by a 900 m vertical line array with 120 equally spaced sensors. Correlated multipath arrivals result in signal cancellation for the adaptive processor, and spatial smoothing techniques must be used prior to beamforming. The processing of subapertures is proposed. Full aperture and subaperture processing techniques are used on the 200 Hz data. Multipath arrivals are found to illuminate only parts of the array, thus indicating that the wavefield can be highly inhomogeneous with depth.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Depth measurement of remote sources using multipath propagation

A submerged acoustic source radiates narrowband Gaussian noise. Its signal propagates to a remote, large aperture vertical array over a multipath channel whose characteristics may or may not be fully known. The primary concern of this study is the accuracy of source depth estimates obtainable from the array output. Cramer-Rao bounds for the depth estimate are calculated. When the velocity profile is known exactly, the value of the bound is quite insensitive to the precise form of the velocity profile. A bound calculated from a constant velocity profile yields an excellent approximation for many situations likely to be encountered in practice. Introduction of an unknown parameter into the velocity profile has little effect on the Cramer-Rao bound for depth. However, a maximum likelihood estimator of depth working with an inaccurate value of the unknown parameter performs poorly. To obtain satisfactory performance, one must estimate the unknown parameters along with the source depth. Simulations demonstrate the success of this approach.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Estimation of acoustic source signatures in an ocean waveguide

Some experimental results are presented on source signature extraction using vertical array data obtained from the acoustic field emitted by a point source in a deep ocean waveguide. Passively sensed field data obtained with a large aperture vertical array is processed by wideband, coherent, and incoherent matched-field methods, and a multichannel deconvolution approach in order to localize the source and reconstruct its temporal signature [P. Mignerey and S. Finette, J. Acoust. Soc. Am. 92, 351–364 (1992)]. A wideband normal modes code was implemented on a fine-grained parallel computer and used to calculate both the replica field vectors and the deconvolved estimates of the source signature. Signals in the band 25–110 Hz are processed for sources located at either the first or second convergence zones, approximately 48 and 97 km from the array. Normalized cross correlations between the measured signatures and their deconvolved estimates indicates that faithful reconstructions of the source signature can be obtained by processing a subaperture of the array data. The method depends on reliable estimates of the ocean Green’s functions linking the source and receivers, which in turn depend on good environmental knowledge concerning the ocean waveguide properties. [Work supported by ONR and NRL.]

Acoustic tomography via matched field processing

This paper suggests a new approach based on narrow-band, low-frequency data using air-deployed shots recorded on widely distributed large aperture vertical arrays. This approach uses fast, cheap, and high S/N data. Results to date with a simulated three-dimensional (3-D) eddy environment show that efficient characterization of the environment plus careful selection of the source/array geometry can lead to highly accurate estimates of the 3-D sound-speed profiles, e.g., maximum errors less than 0.2 m/s.

Modal shading coefficients for high-resolution source depth localization

In this article the minimum variance principle is applied to the modal shading coefficients to improve source depth localization. It is shown that for a long, well-populated vertical array, the source depth localization function based on the minimum variance principle is approximately the same as the depth function based on normal mode theory. For a limited aperture vertical array, the modes are spatially undersampled. In this case, the modal shading coefficients based on the minimum variance principle are shown to significantly improve the depth resolution assuming sufficient signal-to-noise ratio. But, as the signal-to-noise ratio decreases, the depth resolution for the minimum variance case approaches the (conventional) modal beamformer result.

Acoustic navigation for a large aperture array

The acoustic navigation system for a large aperture (900 m) vertical array and the resulting data, describing the motion of the array and research platform FLIP during a 20‐day deployment in the NE Pacific, are presented. Acoustic navigation data are acquired by detecting the high‐frequency signals emitted by near bottom transponders. Recorded travel times are converted to xyz positions by a nonlinear least‐squares approach that adjusts the transponder, FLIP, and the array positions, minimizing the rms error. Simulation results highlight error sensitivities. The movement of FLIP, although constrained by a three‐point moor, exhibits a wind‐driven component and a clockwise semidiurnal tidal component with horizontal displacements in excess of 300 m. Suspended vertically from FLIP under tension, the array has a damped response to tidal and internal wave forces. [Work supported by ONT.]

Vertical directionality of low‐frequency wind‐induced noise

The fine‐scale structure of the directional ambient noise field is revealed using a large aperture (900 m) vertical array. Frequency and spatial spectral estimates are calculated during the passage of a local storm, providing a detailed study of ambient noise levels at low frequencies (15–130 Hz) as wind speed increases from 2 to 12 m/s over a 21‐h period. Spectral levels of beams directed horizontally show the time and spatial variability of distant sources. Spectral levels of beams directed toward the surface reflect local noise sources and display a threshold type behavior with level changes up to 5 dB, suggesting the abrupt onset of a source mechanism such as breaking waves. Subsequent thresholds may indicate a change in source mechanism such as the conversion from spilling breakers to plunging breakers. [Work supported by ONT.]

Modal matched‐field processing with small aperture vertical arrays

In the past years, much attention has been given to mode space matched‐field processing in shallow‐water, low‐frequency environments. In this type of situation, mode space matched‐field processing has several advantages over conventional matched‐field processing such as: lower processing dimension, better sidelobe rejection, and the ability to discriminate signal against modal noise. However, extraction of modal amplitudes from data by the method pioneered by Shang [J. Acoust. Soc. Am. 77, 1413–1418 (1985)] is not very reliable for small aperture arrays. Attempts have been made to overcome this problem in order to make mode space matched‐field techniques more applicable to actual physical environments [T. C. Yang, J. Acoust. Soc. Am. 82, 1736–1745 (1987)]. A fresh approach is brought to this problem here using computer simulations of matched‐field processing in a Pekeris waveguide to demonstrate a new method for extracting modal amplitudes from data. These simulations demonstrate that this new method of modal amplitude extraction allows successful mode space matched‐field detection and localization at array apertures significantly smaller than those allowed by previously used algorithms. Finally, the algorithm will be discussed in the general context of dimensional reduction techniques.

Source localization: Matched field versus matched mode Synthetic and real data performance analysis

The present study compares the matched field and the matched mode techniques for passively localizing a narrow‐band point source in a shallow‐water, range‐independent environment. The matched mode technique is fully characterized in terms of sidelobe ambiguity performance and robustness against both system and environment parameter variation and mismatch. Comparative results are also shown for real data detection of a cw source immerged at different depths in a 120‐m depth channel using a 62‐m aperture vertical array. The results of this study indicate that the matched mode method is much less sensitive to the environmental conditions than the matched field method, and in particular, the result is less degraded by the effects of partial water column sampling (short array). Results obtained on real data showed good agreement with the corresponding tests from simulated data. However, a large sidelobe coverage was found for some situations leading to detection losses. Major causes of performance degradation are the uncertainty on the array sensor position due to array motion and correlated noise due, mainly, to surface‐generated noise.

Sound‐speed determination using matched field processing techniques

Matched field processing is shown to be effective for estimating sound speed in a deep, range‐independent ocean environment. The amplitude and phase of signals from a distant source measured on a large aperture vertical array are sensitive to changes in sound speed. This sensitivity is exploited to infer the environmental sound‐speed profile by matching predicted and measured amplitude and phase. A sound‐speed profile model is developed based on a modified version of Munk's canonical sound field equation. This model is used to determine sound‐speed profile using a 15‐Hz signal from a 240‐m explosive source detected on a 675‐m vertical array at a range of 50 km in a deep‐water Pacific environment, characterized by a classical range‐independent sound channel at 700 m. The search for the best estimated profile is conducted by varying the sound channel axis strength and depth in the modified Munk equation while maintaining a constant sound‐speed profile at great depths. Differences between the estimated and measured profiles are less than ± 2 m/s; the sound channel axis depth is determined within 20 m of the measured axis depth. Extension of this approach to include mesoscale features and greater range is discussed.

A comparison of matched‐mode and matched‐field processor performance against sound‐speed and sensor position mismatch

Recent results due to T. C. Yang [J. Acoust. Soc. Am. 82, 1736–1745 (1987)] have shown that matched‐mode processing can provide range‐depth source localization with good sidelobe and array gain performance. His simulations and success with limited data suggest that the matched‐mode processor is relatively robust with respect to environmental and hydrophone position mismatch. The objective of this study is to compare the performance of matched‐mode versus matched‐field processing under identical array geometry and environmental conditions. Assuming a range‐independent deep ocean environment, sound‐speed mismatch and sensor location mismatch were simulated at second convergence zone ranges. Eigenvalue truncated (reduced rank inverse) matched‐mode processor performance was compared to conventional matched‐field processor performance at 30 Hz on a large aperture vertical array. In the absence of mismatch, the two methods yield near identical results on an array spanning half the water column. The results in the presence of mismatch will be discussed and quantified in terms of signal gain degradation, sidelobe level, and speckle.

Laboratory scale investigation of the amplitude and phase of signals reflected from sea ice ridges

Acoustic reflectivity from plastic models of sea ice ridges on a thin plate were investigated for 0.67 &amp;lt; ka &amp;lt; 6 (where k is the wavenumber and a is the mean ridge depth). Models were constructed of lucite, which has acoustic velocities consistent with analytically estimated velocities of sea ice ridges, and of “fast cast,” a material with much lower velocities. For ka &amp;gt; 1 measurements from randomly oriented and parallel lucite and fast cast ridges on a thin plate and on an infinite half-space were nearly identical, and in good agreement with Twersky theory prediction. At the lowest frequency, ka = 0.67, and at grazing angles less than 20°, measured reflection losses were significantly greater than predictions: twice as large for randomly oriented lucite ridges on a thin plate. The phase change upon reflection was strongly dependent on the material properties of the ridges and their orientation. Results for randomly oriented lucite ridges were found to be reasonably consistent with previously reported inferences of reflection loss and phase based on matched field processing of data recorded on a large aperture vertical array in the Arctic [Livingston and Diachok, J. Acoust. Soc. Am. Suppl. 1 81, S85 (1987)].

Measurements of the vertical fields of acoustic signals and normal mode propagation in the Arctic Ocean

The vertical sound fields of cw signals and shots propagating in the ice‐covered Arctic Ocean were measured during the FRAM IV experiments in April 1982. The measurements were made with a vertical array consisting of 28 hydrophones extending to a depth of 960 m. This experiment marked the first successful deployment of a large aperture vertical array in the Arctic environment. This paper will present the results and interpretations of the vertical array data. The lowest order normal mode amplitudes and phases were measured from the shot signals received simultaneously on the 28 hydrophones of the vertical array. The measured mode amplitudes arc compared with the computed normal mode and used as the basis for spatial filtering of normal modes from the cw signals. Intensity distributions of the cw signals versus depth are presented for several frequencies below 100 Hz. Spatial coherence of the cw signals are studied and compared with that of the ambient noise. The arrival angles of the lowest order normal modes are measured from the cw signals using the mode filtering technique. The angular arrival pattern of the cw signals are plotted using conventional beamforming. Temporal behavior of the cw signals will be shown; phase tracking beamforming is used to enhance the vertical array gain.