Seabed Characterization Experiment Research Articles

Estimation of bottom parameters in the New England Mud Patch using geophone and hydrophone measurements

Measurements of acoustic pressure and particle velocity were made during the Seabed Characterization Experiment (SCEx) in the New England Mud Patch south of Cape Cod in about 70 m of water. The University of Rhode Island and Woods Hole Oceanographic Institution deployed the “geosled” with a four-element geophone array, a tetrahedral array of four hydrophones, and Several Hydrophone Receive Units (SHRUs) as data acquisition packages. In addition, a recently developed low frequency source, the interface Wave Sediment Profiler (iWaSP), was deployed to excite interface waves (Scholte waves) on the seabed. The geophone array was localized using the known locations of the small explosive sources and noise from the research vessel. Modal arrivals from broadband sources on geophones and hydrophones were used to invert for compressional and shear wave speeds and attenuation in the mud layer and the underlying sand layer. The Scholte wave arrivals from the iWaSP source measured by the geophone array also provided estimates for these parameters. The inversion results are compared to modeling using the viscous grain shearing (VGS) theory. [Work supported by the Office of Naval Research, Code 322 OA.]

Vector acoustic observations of underwater ship noise from the 2017 seabed characterization experiment

The Intensity Vector Autonomous Recorder (IVAR) measures acoustic particle velocity and pressure simultaneously. IVAR was deployed on the seabed at depth of 75 m during the 2017 Seabed Characterization Experiment (SBCEX) with objective to study sound propagation within underwater waveguides for which the seabed consists of fine-grained, muddy sediments. In this presentation, vector acoustic observations of ship generated noise recorded by IVAR from a cargo ship traversing the central region of the SBCEX are discussed. Pressure and particle velocity data as a function of frequency are converted into a bounded, non-dimensional form known as circularity, a quantity that is independent of the ship noise source spectrum and which can be interpreted as the normalized curl of active intensity. The frequency dependent pattern in circularity is sensitive to the sediment layering structure, and bathymetry. This pattern, as the vessel closes and opens in range, is inverted for sediment properties using a Bayesian framework, with inversion model space consisting of a low-compressional speed elastic layer, above an elastic half-space. Change in water depth between the location of the ship source and IVAR over the 20-min observation period is addressed using adiabatic mode theory. [Study supported by Office of Naval Research.]

Mud as a porous medium

Since marine mud supports shear waves, it must have an elastic, though fragile, frame. It is also saturated with seawater, therefore it could be modeled as a porous medium. It is an unusual porous medium for a number of reasons. The boundary between the skeletal frame and the pore fluid is difficult to define. The effective porosity of the skeletal frame is not the same as the bulk fluid fraction. The chemistry of mud suggests that a fraction of the water is an integral part of the skeletal frame. Similarly, a fraction of the solid particles may be suspended in the pore fluid. The structure of the skeletal frame is a function of several factors, including salinity of the water and mineralogy of the clay particles. Porous media have two main loss mechanisms: losses in the skeletal frame and viscous losses due to relative motion between the frame and the pore fluid. In the case of marine mud, the frame losses may be modeled as a relaxation process. The measurements from the Seabed Characterization Experiment show it has a spectrum of relaxation frequencies, consistent with a stationary creep process. [Work supported by ONR, Ocean Acoustics Program.]

Semi-supervised deep learning of source location and seabed class from unlabeled merchant ships in the shallow ocean

Understanding sound propagation in a shallow ocean environment is complicated but important for naval technologies. Some researchers are currently applying a variety of supervised deep learning methods, such as convolution neural networks, for either source localization or seabed classification. One of the costs of supervised learning is the requirement for large amounts of labeled data. In the present study, unlabeled data are used to train a Transformer model. Transformer-based models have demonstrated ability for predicting sequential data; examples are Google’s BERT and OpenAI’s GPT-3. In our technique, we train a transformer-based model in a twopart process. First, self-supervised learning is implemented using synthetic ship spectrograms for various shallow ocean environments. The model is trained as an encoder/decoder to perform sequence-to-sequence prediction. Second, the transformer model is fine-tuned to predict the source location and seabed class from a small set of labeled synthetic samples. Data samples measured during the Seabed Characterization Experiment 2017 are utilized as a testing dataset. The advantages of this approach are the ability to train a model with a larger variety of data, including the use of unlabeled data and data of variable input length.

Geoacoustic inversion of modal dispersion on the New England Mud Patch. II: Hybrid parameterization for gradient estimation

This paper applies Bayesian geoacoustic inversion with a hybrid seabed model parameterization to modal-dispersion data from the New England Mud Patch to attempt to estimate gradient structure in the upper mud layer. The hybrid seabed parameterization is based on an upper layer with a general representation of gradients based on Bernstein-polynomial basis functions, above an unknown number of discrete (uniform) layers formulated trans-dimensionally. The dataset, collected during the 2017 Seabed Characterization Experiment (SBCEX17), involves high-order modes (up to mode 21) extracted via warping time-frequency analysis from recordings of a combustive sound source on a vertical hydrophone array. The inversion results are compared to those from a trans-dimensional inversion of the same data with no gradient layer, and to results from other acoustic data sets collected in the region as well as to nearby core measurements.

An overview of seabed characterization experiment

An overview of the Spring 2021 component of the Seabed Characterization Experiment will be presented. The project is focused on the acoustics of fine-grained ocean bottom sediments. The study site is the New England Shelf, south of Martha’s Vineyard, MA, USA. The goals of the project are (1) understand the physical mechanisms that control acoustic propagation in fine-grained sediments, (2) quantify uncertainties in the estimation of seabed parameters, (3) correlate the observed horizontal variations in the acoustical field with the measured horizontal variability of the seabed and (4) assess the performance of the resulting geoacoustic models, and inversion and statistical inference methods. These goals have been pursued by obtaining direct measurements and/or inferring the values of the following parameters over a wide frequency band: compressional wave speed and attenuation, shear wave speed and attenuation, and seabed layering and gradients. Oceanographic data were collected to support both forward and inverse modeling efforts. A major part of this effort was completed in 2017 and a follow-on effort is planned for 2021 and 2022. This presentation provides an overview of the project, a summary of past results, and a discussion of the most recent work. [Work supported by ONR.]

Geoacoustic inversion of modal dispersion on the New England Mud Patch. I: Experimental quantification of information content in higher-order modes

This paper presents the results from a geoacoustic inversion study performed using a combustive sound source signal recorded on a vertical line array during the 2017 Seabed Characterization Experiment (SBCEX17). A single receiver modal estimation method (source deconvolution and warping) is recursively applied along the array channels to estimate modal time-frequency dispersion, resulting in modal dispersion data for 18 modes between modes 1 and 21. These data are then used as an input for trans-dimensional Bayesian geoacoustic inversion. The paper compares inversion results obtained with subsets of modes (1 to 7 and 1 to 15) to those obtained with the whole set (modes 1 to 21) to explore the data information content associated with high-order modes. The study shows that high-order modes enable the resolution of fine details in the seabed sound-speed profile, such as a small sound speed increase over the first 8 m of the seabed (i.e., within the upper portion of the so-called “mud layer,” an important feature of the experiment area). [Work supported by the Office of Naval Research.]

Correlation of deep seabed structure with waveguide invariant β in ±∞ regime

An examination of the received spectrogram levels of merchant ship recordings has identified an important acoustic feature at the very-low frequencies (VLF). The data were recorded on vertical line arrays deployed in the New England Mudpatch during the Seabed Characterization Experiment 2017 in about 75 m of water. The VLF feature occurs when the group velocities of modes 1 and 2 are equal at a frequency f = F. The average value of F from about twenty ship signatures observed in the Mudpatch is about 24.5 Hz. Generally, the observation of this VLF feature is a result of βmn( F) = infinity where βmn is the waveguide invariant for modes n and m. The sign of the waveguide invariant is determined by the sign of the difference between the phase speeds at F. Also, we predict what F would be in two areas south of the Mudpatch, including a second smaller Mudpatch in about 125 m of water and an area near the slope in about 250 m of water where the sediment is reported to be sand over multiple layers of softer sediments. Finally, F is predicted on the Atlantis II seamount with a limestone seabed in about 1650 m of water.

Information content of reflection-coefficient data versus angle and frequency for geoacoustic inversion

This paper considers the information content of seabed reflection-coefficientdata (as a function of angle and frequency) to resolve fine seabed structure through geoacoustic inversion. During the 2017 Seabed Characterization Experiment conducted on the New England Mud Patch, two types of measurement systems were employed to collect reflection data. One is a bistatic system with bottom-moored hydrophones and an omni-directional towed source, and the other is a towed system that uses the same source and a 15.4 m acoustic array on the same cable (source separation to the closest sensor is ∼ 27 m). Limited by the source–receiver geometry and data quality, the angular coverages of the reflection data for the bistatic and towed systems are ∼25–58° and ∼43–55°, respectively. This study aims to assess the capabilities of these systems to infer sediment structure using simulations over a frequency band of 1–6 kHz. Inversions show that the data simulated for the bistatic system better resolve fine structure of the seabed even with fewer frequency components, while the data for the towed system must include reflections at higher frequencies to get acceptable results. [Work supported by the Office of Naval Research.]

Seabed type and source parameters predictions using ship spectrograms in convolutional neural networks.

Broadband spectrograms from surface ships are employed in convolutional neural networks (CNNs) to predict the seabed type, ship speed, and closest point of approach (CPA) range. Three CNN architectures of differing size and depth are trained on different representations of the spectrograms. Multitask learning is employed; the seabed type prediction comes from classification, and the ship speed and CPA range are estimated via regression. Due to the lack of labeled field data, the CNNs are trained on synthetic data generated using measured sound speed profiles, four seabed types, and a random distribution of source parameters. Additional synthetic datasets are used to evaluate the ability of the trained CNNs to interpolate and extrapolate source parameters. The trained models are then applied to a measured data sample from the 2017 Seabed Characterization Experiment (SBCEX 2017). While the largest network provides slightly more accurate predictions on tests with synthetic data, the smallest network generalized better to the measured data sample. With regard to the input data type, complex pressure spectral values gave the most accurate and consistent results for the ship speed and CPA predictions with the smallest network, whereas using absolute values of the pressure provided more accurate results compared to the expected seabed types.

A porous medium model for mud.

The extended Biot model for sands and silts is repurposed to include mud, but modifications are needed. The boundary between pore water and skeletal frame needs to be redefined because a significant fraction of the pore fluid is adsorbed onto the solid frame by electrostatic forces, and a proportion of the solid particles may be suspended in the pore fluid. Revil's relationships are used to simplify the input parameters. The frame elasticity equations are corrected to accommodate the sparse skeletal frame, which is supported by electrostatic forces, and behaves differently to a mechanical packing of grains. The corrected, Revil, extended Biot model has just four fitting parameters and is compared with published measurements of wave speeds and attenuations in the literature of clay, silty clay, and clayey silt sediments including recently published measurements from the Seabed Characterization Experiment. The results indicate that the skeletal frame in clay has a high water-content and the pore water contains suspended particles. To fit all the currently available data, it was necessary to modify corrected, Revil, extended Biot by flattening the creep related relaxation loss spectrum. There is a similarity with the Viscous Grain Shearing models in the use of a fractional exponent.

Learning location and seabed type from a moving mid-frequency source.

While source localization and seabed classification are often approached separately, the convolutional neural networks (CNNs) in this paper simultaneously predict seabed type, source depth and speed, and the closest point of approach. Different CNN architectures are applied to mid-frequency tonal levels from a moving source recorded on a 16-channel vertical line array (VLA). After training each CNN on synthetic data, a statistical representation of predictions on test cases is presented. The performance of a single regression-based CNN is compared to a multitask CNN in which regression is used for the source parameters and classification for the seabed type. The impact of water sound speed profile and seabed variations on the predictions is evaluated using simulated test cases. Environmental mismatch between the training and testing data has a negative impact on source depth estimates, while the remaining labels are estimated tolerably well but with a bias towards shorter ranges. Similar results are found for data measured on two VLAs during Seabed Characterization Experiment 2017. This work shows the superiority of multitask learning and the potential for using a CNN to localize an acoustic source and detect the surficial seabed properties from mid-frequency sounds.

Differences in regression, classification, and multi-task deep learning on pressure time series for range and seabed type

Estimating source-receiver range and predicting seabed type are important problems in underwater acoustics. Deep learning solutions for simultaneous predictions have been previously shown some success for these problems due to their ability to learn patterns from large datasets. However, one important question to consider in deep learning is how predictions should be made: A network can be configured to classify or regress predictions. In this study, that question is explored by comparing predictions from networks trained on simulated SUS charge pressure time-series configured to either regress or classify range and seabed predictions from measured data taken with the IVAR system during the 2017 Seabed Characterization Experiment. To further this inquiry, networks configured to regress range and classify seabed type (via “multi-task” learning) are also explored. Separating the two predictions proves to be useful, as the networks using multi-task learning perform better at predicting range and seabed class simultaneously than those configured only for classification or regression. The results of this experiment illustrate the need to use the proper type of network outputs depending on the desired predictions. [Work supported by the Office of Naval Research.]

Seabed and range estimation of impulsive time series using a convolutional neural network.

In ocean acoustics, many types of optimizations have been employed to locate acoustic sources and estimate the properties of the seabed. How these tasks can take advantage of recent advances in deep learning remains as open questions, especially due to the lack of labeled field data. In this work, a Convolutional Neural Network (CNN) is used to find seabed type and source range simultaneously from 1 s pressure time series from impulsive sounds. Simulated data are used to train the CNN before application to signals from a single hydrophone signal during the 2017 Seabed Characterization Experiment. The training data includes four seabeds representing deep mud, mud over sand, sandy silt, and sand, and a wide range of source parameters. When applied to measured data, the trained CNN predicts expected seabed types and obtains ranges within 0.5 km when the source-receiver range is greater than 5 km, showing the potential for such algorithms to address these problems.

Estimation of seabed properties and range from vector acoustic observations of underwater ship noise.

The Intensity Vector Autonomous Recorder (IVAR) simultaneously measures acoustic particle velocity and pressure. IVAR was deployed during the 2017 Seabed Characterization Experiment (SBCEX) with the primary objective to study sound propagation in fine-grained, muddy sediments. In this study a Bayesian inversion framework is applied to ship underwater noise recorded by IVAR. The data are relative phase of pressure and vertical particle velocity, a quantity that is independent of the ship noise source spectrum. Inversion estimates for the sediment layer and underlying basement properties are in agreement with other reports from SBCEX.

Guest Editorial An Overview of the Seabed Characterization Experiment

A broad overview of the field work conducted during the SBCEX has been provided. SBCEX included a comprehensive subbottom chirp profile of the experimental area, about 200 sediment cores, and a number of direct in situ measurements of the acoustic properties of the seabed. Numerous acoustic propagation measurements throughout the experimental site provided data for inversion processes and statistical inference techniques to remotely sense the acoustic properties of the bottom. The scientific plan of these measurements and the supporting oceanographic and environmental measurements allowed the various principal investigators to begin to achieve the goals outlined in the introduction of this editorial, and thereby increase the community's understanding of the acoustic properties of the fi ne-grained sediment at the site.

Measurement of Sound Speed in Fine-Grained Sediments During the Seabed Characterization Experiment

The Seabed Characterization Experiment was carried out from March 5 to April 10, 2017 (SBCEX17) on the New England Mud Patch, approximately 90 km south of Martha's Vineyard. The SBCEX17 experimental site covers an area of 11 km × 30 km with water depth in the range of 75–80 m. The Sediment Acoustic-speed Measurement System (SAMS) is designed to measure sediment sound speed and attenuation simultaneously over the surficial 3 m of sediments. During SBCEX17, SAMS was successfully deployed at 18 sites, which were chosen to coincide with coring locations, with the goal of developing a geoacoustic model for the study area. In this article, a summary of SAMS operation during SBCEX17 is presented, as well as preliminary results for sediment sound speed and its spatial variation in the frequency band of 2–10 kHz. It is found that in mud, the sound-speed ratio is in the range of 0.98–1. Little dispersion was observed in this frequency band. Using the preliminary SAMS sound-speed results measured at different depths, the sound-speed gradient in mud within the surficial 3 m favors an exponential rather than a linear dependence at SBCEX17 site. Large gradients are observed for depth shallower than 1.5 m. For the sandy basement beneath the mud layer, the sound-speed ratio is as high as 1.105.

On the sound speed dispersion and the frequency dependence of sound attenuation in a fine-grained sediment in the New England Mud Patch

The study on the sound speed dispersion and the frequency dependence of sound attenuation in marine sediments is of great importance in understanding the physics of sound propagation in the sea bottom. Modal dispersion curves (MDC) have only been utilized to estimate the low-frequency sound speed. At higher frequencies (i.e., above a few hundred hertz), arrival time difference for different modes generally become smaller and MDC is difficult to extract because of the cross-mode interference. This paper applies time-frequency representations (TFR) with high time-frequency resolution (e.g., optimal-kernel and center-affine-filtered TFR) to broadband acoustic signals (e.g., 31g explosive and combustive source signals) detonated along the 15-km Northwest-Southeast tracks during the Seabed Characterization Experiment in the New England Mud Patch, which has a surface fine-grained sediment layer overlying a sandy bottom. The resulting high-resolution broadband (150–1500 Hz) MDC, in conjunction with TL data, are fed to a previously developed two-step dimension-reduced geoacoustic inversion algorithm including modal dispersion-based inversion for sound speed and energy-based inversion for attenuation [IEEE J. Ocean. Eng., 44, in print, (2019)]. The estimated sound speed and attenuation are compared with the theoretical results from seabed geoacoustic models for fine-grained sediments. [Work supported by ONR Ocean Acoustics (322OA).]The study on the sound speed dispersion and the frequency dependence of sound attenuation in marine sediments is of great importance in understanding the physics of sound propagation in the sea bottom. Modal dispersion curves (MDC) have only been utilized to estimate the low-frequency sound speed. At higher frequencies (i.e., above a few hundred hertz), arrival time difference for different modes generally become smaller and MDC is difficult to extract because of the cross-mode interference. This paper applies time-frequency representations (TFR) with high time-frequency resolution (e.g., optimal-kernel and center-affine-filtered TFR) to broadband acoustic signals (e.g., 31g explosive and combustive source signals) detonated along the 15-km Northwest-Southeast tracks during the Seabed Characterization Experiment in the New England Mud Patch, which has a surface fine-grained sediment layer overlying a sandy bottom. The resulting high-resolution broadband (150–1500 Hz) MDC, in conjunction with TL data, are ...

In situ measurements of compressional and shear wave properties during gravity coring operations

Sediment cores provide valuable insight about the physical properties of the seabed, and laboratory measurements of sediment wave speed from cores are often considered “ground truth.” However, sound-speed estimates obtained from cores can be inaccurate due to changes in pressure, temperature, and mechanical properties of the sediment caused by the removal of the core from the seabed and its subsequent transport to the laboratory. To address these issues, the Acoustic Coring System (ACS) was developed. The system uses sets of transducers mounted below the penetrating tip of a sediment corer to make in situ measurements of geoacoustic properties as the corer penetrates the seabed. Compressional wave measurements are obtained with rod-mounted piezoelectric cylinders, and shear wave measurements are obtained with bender elements mounted in flat blades. The ACS was deployed during the environmental survey for ONR Seabed Characterization Experiment (SBCEX) in 2016 and in support of the ONR Canada Basin Acoustic Propagation Experiment (CANAPE) in 2017. In both cases, the system provided measurements of the sediment sound-speed profile as a function of depth within the seabed. On-going work focused on the measurement of shear wave speed profiles and compressional and shear wave attenuation will also be presented. [Work supported by ONR.]Sediment cores provide valuable insight about the physical properties of the seabed, and laboratory measurements of sediment wave speed from cores are often considered “ground truth.” However, sound-speed estimates obtained from cores can be inaccurate due to changes in pressure, temperature, and mechanical properties of the sediment caused by the removal of the core from the seabed and its subsequent transport to the laboratory. To address these issues, the Acoustic Coring System (ACS) was developed. The system uses sets of transducers mounted below the penetrating tip of a sediment corer to make in situ measurements of geoacoustic properties as the corer penetrates the seabed. Compressional wave measurements are obtained with rod-mounted piezoelectric cylinders, and shear wave measurements are obtained with bender elements mounted in flat blades. The ACS was deployed during the environmental survey for ONR Seabed Characterization Experiment (SBCEX) in 2016 and in support of the ONR Canada Basin Acoustic...

Source level predictions of surface ships using seabed characterization in the New England Mudpatch from viscous grain shearing model

Acoustic inferences from remote sensing of surface ship radiated noise in bottom-limited ocean environments are difficult due to a generally unknown geoacoustic structure of the seabed. The idea here is to first calibrate an acoustic sensing system by estimating geoacoustic parameter values by utilizing controlled sources with known levels. Then, the acoustic field that results from a surface ship is utilized to estimate the ship source levels where the transmission loss is constructed based on the previous estimate of the seabed parameterization and measurements of the sound speed profile at the time that the ship noise measurements were made. This concept was tested with acoustic data collected on two vertical line arrays (VLAs) during the Seabed Characterization Experiment in the New England Mudpatch during April 2017, in 75 m of water. The acoustic field was produced with a controlled source radiating in the 1.5–4 kHz band. A maximum entropy method provided estimates of viscous grain shearing (VGS) model parameter values. Then, using this VGS parameterization, ship source levels as a function of aspect and frequency out to about 3 kHz were extracted for the RV Endeavor and several merchant ships passing near the VLAs. [Work supported by ONR N00014-16-C-3065.]