Ocean Tomography Research Articles

Experimental studies of the characteristic features of vector receivers in application to ocean tomography

Experimental data on the angular structure of the sound field on the shelf of the Sea of Japan are discussed. The data are obtained by using the vector-phase processing of complex signals in the form of M-sequences with a central frequency of 2500 Hz. With the parameters of the water bulk being monitored, an unambiguous relation is established between the variability of the temperature regime on a fixed acoustic path and the arrival times and angles. It is shown that the use of vector receivers in ocean tomography provides an additional independent parameter of the pulse response of the underwater waveguide and leads to an increase in the efficiency of solving the problem of reconstructing the hydrophysical fields from the data of acoustic sensing.

Ray and finite frequency kernel stability in the presence of realistic eddies

Many recent studies have focused on the instability of ray paths (and therefore ray sampling kernels) in an ocean sound-speed field perturbed by internal waves. In many parts of the world’s oceans there is synoptic scale eddy and planetary wave activity that produces sound-speed anomalies of larger scale than the internal waves, and of equal or larger magnitude. These are the features to be imaged by ocean tomography experiments, but the linearized forward problem that relates the travel time anomaly of the observed rays to the sound-speed anomalies may be invalid for realistic levels of eddy variability. This problem has been studied in the past for ray kernels by a number of authors, some of whom have concluded that nonlinearity requires iteration of the inversion process. We will compare and contrast the linearity of ray travel time kernel, finite frequency travel time kernels, and finite frequency kernels for other observables, such as pulse shape, in the presence of realistic variability for the N. Philippine sea, as estimated by ocean models and historical data. The travel time signals and nonlinearity due to the synoptic variability will be compared to that for internal waves. [Work supported by ONR.]

The coupled oceanographic-tomographic analysis and prediction system: A practical means for utilizing high-frequency tomography in the world’s oceans

Over the last several years, technology has been developed to transition ocean tomography out of the laboratory and into practical use. This technology, the coupled oceanographic-tomographic analysis and prediction system (COTAPS), couples an ocean circulation model in a region with travel time information from a network of high-frequency acoustic nodes in the same region. A data assimilation technique is used to impose corrections to the ocean model temperatures and salinities based on observed arrival time anomalies and a Kalman gain matrix. Details of the features of COTAPS are presented and discussed. This includes the technique and associated software for assessing ray path information in the area in which the network of nodes is deployed, an automated method for calculating the Kalman gain, and a suggested design of the acoustic nodes. We also discuss a few issues that remain to be researched. These include significant variability of observed arrival times over short time intervals (less than 5 min) and how to mitigate the impact of such variability, the use of observed arrival times to account for errors in the locations of the nodes in a network, and the effect of assimilating arrival time information from multiple source-receiver pairs.

Model-oriented ocean tomography using higher frequency, bottom-mounted hydrophones

A tomographic scheme is presented that ingests ocean acoustic measurements into an ocean model using data from bottom-mounted hydrophones. The short distances between source-receiver pairs (1-10 km) means arrival times at frequencies of 8-11 kHz are readily detectable and often distinguishable. The influence of ocean surface motion causes considerable variability in acoustic travel times. Techniques are presented for measuring travel times and removing the variability due to surface waves. An assimilation technique is investigated that uses differences in measured and modeled acoustic travel times to impose corrections on the oceanographic model. Equations relating travel time differences to oceanographic variables are derived, and techniques are presented for estimating the acoustic and ocean model error covariance matrices. One test case using a single source-receiver pair shows that the tomographic information can have an impact on constraining the solution of the ocean circulation model but can also introduce biases in the predictions. A second test case utilizes knowledge of a bias in a model-predicted variable to limit grid cells that are impacted by the tomographic data. In this case, using the tomographic data results in significant improvements in the model predictions without introducing any biases.

Ocean tomography, inverse methods, and broadband ocean acoustics

Ocean acoustic tomography, as proposed by Munk and Wunsch in 1979, and as implemented by the Ocean Tomography Group, uses ray travel times to estimate ocean sound speed and currents. Earlier work by Medwin (1970) and Hamilton (1977) used pulse travel times as measures of integrated sound speed along paths at short and long range, respectively. Munk and Wunsch (1979) recognized that broadband transmissions between many instruments could be used with inverse methods [Backus and Gilbert (1967); Liebelt (1967)] to reconstruct 3D ocean sound speed fields from the travel times along multiple paths. Inverse methods are widely used in ocean acoustics and in physical oceanography, and the modern challenge is to incorporate time dependence into the inverse methods to take advantage of the improving ocean dynamical models. In addition, the understanding of broadband acoustic propagation has improved to the point of refining the sensitivity kernel for travel time measurements beyond the simple geometrical optics of ray paths. This paper will review the evolving use of forward and inverse methods in acoustical oceanography, primarily with application to acoustic tomography.

Fourier component fluctuations of wideband ocean tomography signals: Empirical temporal statistics from AST96

Current theories for wave propagation in random media can predict fluctuations for narrow-band (single-frequency) signals, but do not explicitly address wideband signals. Recent ocean acoustic tomography experiments, however, have employed wideband ‘‘m-sequence’’ signals. Processing here assumes a periodic signal, and hence, involves a unique set of Fourier series components. A technique is presented for decomposing such wideband signals into their Fourier components and then analyzing the space–time fluctuations of these ‘‘narrow-band’’ components. The technique is applied to AST96 data transmitted over 150-km and 1-Mm paths for single-point and two-point (temporal separation) statistics. [Work support by ONR.]

A cross-relation matched field inversion for geoacoustic parameter estimation in shallow water

In this paper the application of a matched field (MF) processor for geoacoustic parameter estimation from a signal of opportunity such as the broadband random noise radiated by passing ships is considered. A novel cross-relation (CR) based matched field processor is introduced for the purpose of ocean tomography based on broadband random source. This class of MF processors is based on the cross-relation property of sensor outputs at an array and their corresponding transfer functions from the true source location to the array. The processors are developed for nonstationary (NS), and wide sense stationary (WSS) random signals. For each formulation, two processors are proposed, a self-CR and a cross-CR. The performance of the proposed MF processors for environmental parameter estimation is demonstrated for real ocean environments using data collected by a 16-element vertical line array during one of the experimental tracks from the Pacific Shelf experiment that was carried out in shallow water off the west coast of Vancouver Island in the Northeast Pacific Ocean. The high-resolution property makes the cross-CR processor an excellent candidate for inversion for model parameters of the ocean waveguide. The replica or modeled fields are calculated using the normal mode model, ORCA.

Sound scattering by moving inhomogeneity

This report considers the propagation of sound in the vicinity of a rather general class of localized flows induced by the motion of spherically symmetric structures of the type of solid spheres and droplets or vortices in the liquid, say, ocean environment. Acoustic scattering from such inhomogeneities, situated inside various flows, is analyzed with the aid of the M. J. Lighthill equation boundary value problem solution—both in the presence of vorticity and without it. The role of a moving body generating inhomogeneity as well as the role of the surface at which the tangential component of flow velocity could suffer a discontinuity are demonstrated in a most general form. Comparison of body and flow contribution to the observed scattered field is presented. The physical condition of flow presumably yields in a forward scattering direction—a situation of the type encountered in a conventional ocean tomography system, comprising a source–detector line, is proposed. The angular structure of the scattering amplitude and the frequency dependence of the scattered sound are evaluated.

Vertical synthetic aperture sonar for ocean acoustic tomography

This paper presents the result of a first attempt to achieve a vertical synthetic aperture in the ocean for SOFAR multipath identification. The experiment was conducted during the deployment of a tomographic array in the Mediterranean Sea. Drifting the hydrophone up or down from a ship while listening to the transmitted signal created a powerful synthetic aperture at 400 Hz. Wide-band phase-coded signals, typically used in ocean tomography, were found suitable for this application. The displacement length was 100 m and the hydrophone velocity 1 m/s. The obtained resolution of 1/spl deg/ enabled all the rays in the tested middle range configuration to be resolved and identified. Most of them could not have been resolved with a static hydrophone. Several Doppler processing methods are presented. The narrowband approximation leading to fast algorithms is discussed. Phase time series of individual paths obtained with an array-like wave separation method show that the phase coherence of the different multipaths is nearly perfect. An angle/velocity calibration method and a first rough inversion are finally presented.

Tomographic imaging of the ocean volume

Ocean acoustic tomography has become a highly developed tool for observing the ocean’s interior. It has been used in the Atlantic, Pacific, and Indian Oceans, the Mediterranean, Greenland, and Norwegian Seas, and in a number of smaller bodies of water including Puget Sound. There are active ocean tomography research programs in the U.S., France, Germany, Japan, and India. In the same sense that x-ray, or ultrasonic acoustic tomography is used to image parts of the human body, ocean tomography provides images of the ocean’s interior. This presentation will describe the methodology and present examples of tomographically derived ocean images.

Ocean acoustic tomography based on peak arrivals

The recently introduced notion of peak arrivals [Athanassoulis and Skarsoulis, J. Acoust. Soc. Am. 97, 3575–3588 (1995)], defined as the significant local maxima of the arrival pattern, is studied here as a modeling basis for performing ocean tomography. Peak arrivals constitute direct theoretical counterparts of experimentally observed peaks, and offer a complete modeling of experimental observables, even in cases where ray or modal arrivals cannot be resolved. The coefficients of the resulting peak-inversion system, relating travel-time with sound-speed perturbations, are explicitly calculated in the case of range-independent environments using normal-mode theory. To apply the peak-inversion scheme to tomography the peak identification and tracking problem is examined from a statistical viewpoint; maximum-likelihood and least-square solutions are derived and discussed. The particular approach adopted treats the identification and tracking problem in close relation to the inversion procedure; all possibilities of associating observed peaks with background arrivals are examined via trial inversions, and the best peak identification is selected with respect to a least-square criterion. The feasibility of peak tomography is subsequently demonstrated using first synthetic data and then measured data from the THETIS-I experiment. In the synthetic case the performance of the overall scheme is found to be satisfactory both with noise-free and noisy data. Furthermore, the identification, tracking, and inversion results using experimental acoustic data from THETIS-I are in good agreement with independent field observations.

Robust ray path identification for ocean acoustic tomography.

Acoustic tomography in a ray environment is a two‐step process. First, each measured ray arrival time in the received multipath structure is identified with a particular predicted ray path. Second, the differences between the measured arrival times and the predicted arrival times for the represented ray paths are used in a linear inversion to calculate corrections to parameter values describing the sound‐speed structure. If measured arrival times are identified with the wrong ray paths, errors will result in the linear inversion. In deep ocean tomography, the time spacing between ray arrivals is typically large compared to the parameter induced changes in arrival times, so ray path identification is not difficult. In shallow water, however, ray path identification can be more challenging. An algorithm is presented that combines the ray path identification step with the linear inversion step to allow tomography using rays where the parameter induced arrival time shifts may be larger than the time spacing between arrivals. The performance of the algorithm is simulated for a typical shallow‐water environment. [Work supported by ONR.]

Interferential algorithm for ocean bottom diagnostics

The efficiency of determining sound velocity, sound density, and acoustic attenuation factor in the bottom is evaluated by a numerical experiment, using the method of interferential ocean bottom tomography proposed by the authors. Simulation is performed in the presence of random noise when the source power is unknown.

Ocean tomography via matched‐field processing

The tomographic determination of 4‐D ocean sound speeds depends upon the nature of the system design, the analysis technique, and upon the resources used for generating and measuring the acoustic fields sampling the ocean region of interest. Matched‐field tomography has recently been proposed as a high‐resolution method potentially capable of synoptically imaging ocean regions as large as 1000 by 1000 km with rms accuracies less than 1 m/s [A. Tolstoy, J. Comput. Acoust. 2(1), 1–10 (1994)]. However, all results to date for the determination of deep ocean sound‐speed profiles have been simulated. A major drawback to the actual implementation of the technique has been the requirement for several long (1000 m) vertical arrays, and these arrays have been neither affordable nor readily available. New technology is on the verge of producing relatively inexpensive, air‐deployable, lightweight versions of these arrays suggesting that the MFP method may soon be testable. [Work supported by ONR.]

Oceanic tomography using model‐based matched‐filter processing of wideband acoustic signals

As originally conceived, ocean acoustic tomography relies on determining the travel times of pulses that propagate along identifiable multipaths between pairs of transducers. In principle, other properties of acoustic propagation such as amplitude and phase over a broad frequency band can be used to infer environmental parameters. These parameters include range‐dependent sound‐speed profiles and geoacoustic parameters of the sea bottom. Recent experimental work has demonstrated that the distortion of time‐dispersed wideband signals can be compensated effectively by using a model‐based matched filter (MBMF). A multi‐channel MBMF receiver that incorporates the modeled Green’s function of the medium, determined the correct range and depth of the source and receiver. In this paper, the possibility of applying model‐based matched‐filter processing to the tomographic inversion problem was investigated. The acoustic data are low‐frequency, large time‐bandwidth product, linear‐frequency‐modulated signals transmitted through a range‐independent waveguide at a deep water site west of Sardinia. Compared to time‐of‐flight tomography, the amplitude and phase distortion undergone by the propagated waveforms were fully exploited to reconstruct, in part, the range‐averaged sound‐speed profile between source and receiver. This was achieved by searching for the set of sound‐speed profile parameters that maximized the gain at the output of a multi‐channel MBMF receiver.

Scattering from frontal structures, internal tides, and internal waves during the 1992 Barents Sea polar front experiment

In August 1992, a coastal ocean tomography experiment was conducted in the Barents Sea over the steep northwestern slope of the Bear Island Trough, about 100 km east of Bear Island. The oceanographic objective of the experiment was to study the dynamics of the polar front and its vicinity using both acoustic tomography and traditional hydrographic techniques. The acoustic objective was to study the effects of the front and other coastal oceanography on the acoustic propagation. These objectives are strongly coupled, in that to effectively perform an inverse for the ocean features, one must first obtain a good understanding of the ‘‘forward problem.’’ In this paper, the effect of the strong frontal interface is examined, including its corrugations (‘‘interleaving’’ features), internal tides, and internal waves on the observed acoustic propagation. Effects on both the modal and ray propagation pictures will be examined, using both theoretical predictions and experimental data.

Tomographic measurements of frontal variability in the Barents Sea

In August 1992 a coastal ocean tomography experiment was conducted in the Barents Sea over the steep northwestern slope of the Bear Island Trough, about 100 km east of Bear Island. The objective of the experiment was to map and study the oscillations of the Barents Sea Polar Front using acoustic tomography coupled with traditional hydrographic techniques. Because mesoscale ocean variability has shorter spatial and temporal scales in a coastal environment, a vertical receiving array and frequently transmitting tomography sound sources were used to achieve an enhanced system resolution appropriate for coastal monitoring. The vertical array data were processed using plane‐wave beamforming to separate ray arrivals in both time and angle. In addition, modal arrivals were separated using broadband modal beamforming techniques. The processed travel time data were then ‘‘inverted’’ using a hybrid ray‐mode inverse technique to produce a time series of maps of the polar front. In this presentation, the hybrid ray‐mode inverse method and the frontal variability as imaged by this shallow‐water tomography system are discussed. [Work supported by ONR 1125AR.]

Analysis of finite-duration wide-band frequency sweep signals for ocean tomography

A group of amplitude and frequency modulated signals which generate narrow synthesized pulses are described. The pulse-compression properties of these signals should approach those of maximal (M) sequence phase-modulated signals now commonly used in ocean experiments. These amplitude-tapered linear frequency-sweep (chirp) type signals should be accurately reproducible with most acoustic sources since they have controllable limited-bandwidth frequency content and differentiable phase. The Doppler response of the signals is calculated using a wideband approach, where the frequency shift from relative motion is not constant throughout the waveform. The resultant Doppler effect on the matched-filter output is a function of the signal duration. The signals are suitable for use with tunable resonant transducers, and have adequate Doppler response for use with Lagrangian ocean drifters.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Impedance analysis of acoustic vibration element using giant magnetorestrictive material

Giant magnetoresistive material offers great magnetostriction over 1000 p.p.m., making it well-suited to acoustic transducers for acoustic ocean tomography systems. The acoustic vibration element using giant magnetostrictive material (AVEG) is a transducer is a drive unit developed for this purpose. The authors discuss the following three points through measurement of AVEG impedance: measurement of mechanical resonance frequency based on electrical impedance characteristics; calculation of equivalent circuit constraints based on analysis of motional impedance characteristics; and comparison between calculated and measured values for the displacement-current sensitivity of the vibrating terminal. Around the resonance frequency, the displacement-current sensitivity values measured with a laser Doppler vibrometer almost exactly matched the values calculated by using the equivalent circuit constants. This indicates that mechanical vibration can be estimated from the impedance characteristics.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Travel time corrections for ice scattering in the 1988–1989 Greenland Sea tomography experiment.

Previous work by Guoliang and Wadhams [Prog. Oceanog. 22, 249–275 (1989)] has shown that reflections of acoustic rays by a sea ice cover can significantly affect acoustic travel times in an ocean tomography experiment. This can lead to warm or cold biases in the ocean inversions, if unaccounted for. This paper presents a discussion of how one deals with correcting the travel times of acoustic arrivals seen in the 1988–1989 Greenland Sea tomography experiment for ice effects. The discussions will include: generalization of the beam displacement formalism of Guoliang and Wadhams, determination of the reflection coefficient versus incidence angle of the rays using scattering amplitude data, estimation of beam displacements and time delays, and effects of the corrections on inversions for the ocean temperature field. Directions for future work will also be discussed. [Work supported by ONR and NSF.]