Selected Abstracts

A Parallel Hypothesis Method of Autonomous Underwater Vehicle Navigation
Cara LaPointe, Ph.D., 2009
Dana Yoerger, Advisor

This thesis presents a parallel hypothesis method for autonomous underwater vehicle navigation that expands the operating envelope of existing long baseline acoustic navigation systems by incorporating information that is not normally used. The proof of concept was done using real-world data obtained by the Autonomous Benthic Explorer (ABE) and Sentry vehicles during operations on the Juan de Fuca Ridge. This algorithm uses a nested architecture that breaks the navigation solution down into basic building blocks for different types of available external information. A proof of concept was conducted using acoustic time-of-flight measurements in a hypothesis generation process and a priori low-resolution bathymetric data in a grid arbitration process.
The major contributions of this research include in-situ identification of acoustic multipath time-of-flight measurements, the multiscale utilization of a priori low-resolution bathymetric data in a high-resolution navigation algorithm, and the design of a navigation algorithm with a flexible architecture. This flexible architecture allows the incorporation of multimodal beliefs without requiring a complex mechanism for real-time hypothesis generation and culling, and it allows the real-time incorporation of multiple types of external information as they become available in situ into the overall navigation solution.

Forward Sound Propagation around Seamounts: Application of Acoustic Models to the Kermit-Roosevelt and Elvis Seamounts
Hyun Joe Kim, Ph.D., 2009
Arthur Baggeroer, Advisor

The Basin Acoustic Seamount Scattering Experiment (BASSEX) of 2004 was conducted to measure forward sound propagation around the Kermit-Roosevelt and the Elvis seamounts in the Northeast Pacific, focusing on bathymetric effects on sound propagation. In this thesis, the reconciliation between the measured broadband pulses and the simulated pulses using various two-dimensional (2D) and three-dimensional (3D) theoretical models is carried out to investigate the physical characteristics of the sound propagation around seamounts. First, the shadow and convergence zones behind the seamounts are explored using the 2D and 3D sound propagation models. Second, the broadband pulses using the 2D parabolic equation (PE) for the SPICEX source are reconciled with the measurement in the BASSEX experiment, which shows good correlation within the convergence zone. Those pulses are identified with the help of the 2D ray tracing method. Third, the 3D spectral coupled-mode model (W. Luo, PhD Thesis, MIT, 2007) is applied to sound propagation around the seamount showing wider shadow zone than the Nx2D computation. The increase of computational efficiency of the 3D spectral coupled-mode model is achieved using parallel computing. Through these analyses, we examine the characteristics of the sound propagation around the seamount as well as the limit of the application of the acoustic models.

Cooperative Navigation for Autonomous Underwater Vehicles
Alexander Bahr, Ph.D., 2009
John Leonard, Advisor

Self-localization of an underwater vehicle is particularly challenging due to the absence of GPS reception or features at known positions that could be used for position computation. This typically requires the pre-deployment of a set of beacons. This thesis examines the scenario in which the members of a group of AUVs exchange navigation information with one another so as to improve their individual position estimates.
We describe how the underwater environment poses unique challenges to vehicle navigation and how cooperation can improve the performance of self-localization. We also address the constraints of the communication channel and the effect that these constraints have on the design of cooperation strategies. The classical approaches to underwater self-localization of a single vehicle are presented. We then examine how methods used for cooperating land-vehicles can be transferred to the underwater domain. An algorithm for distributed self-localization is proposed. We also address how correlated position estimates of cooperating vehicles can lead to overconfidence in individual position estimates. Finally, key to any successful cooperative navigation strategy is the incorporation of the relative positioning between vehicles. A distributed algorithm for the dynamic positioning of vehicles is proposed.

CFD Study of Hydrodynamic Signal Perception by Fish Using the Lateral Line System
Mark Rapo, Ph.D., 2009
Houshou Jiang, Advisor

The lateral line system on fish has been found to aid in schooling behavior, courtship communication, active and passive hydrodynamic imaging, and prey detection. The most widely used artificial prey stimulus is the vibrating sphere, which some fish are able to detect even when the signal induced velocities to its lateral line are orders of magnitude smaller than background current velocities. This thesis uses a series of computational fluid dynamic (CFD) simulations, matched with recent experiments, to quantify the effects of 3D fish body parts on the received dipole signals, and to determine signal detection abilities of the lateral line system in background flow conditions. Analytical models were developed for the Mottled Sculpin canal and superficial neuromast motions, in response to hydrodynamic signals. When the background flow was laminar, the neuromast motions induced by the stimulus signal at threshold had a spectral peak larger than spectral peaks resulting from the background flow induced motions. When the turbulence level increased, the resulting induced neuromast motions had dominant low frequency oscillations. For fish using the signal encoding mechanisms of phase-locking or spike rate increasing, signal masking should occur.

Sound Propagation around Underwater Seamounts
Joseph J. Sikora III, Ph.D., 2009
Arthur Baggeroer, Advisor

Seamounts are ubiquitous throughout the world’s oceans and can absorb and scatter acoustic energy, offering many interesting acoustic modeling challenges. The goal of the research performed in support of this thesis is to measure the acoustic scattered field of a large, conical seamount at long-range, and reconcile observations with 2-D range-dependent acoustic models, for the purpose of understanding the effects of highly range-dependent bathymetry.
The Basin Acoustic Seamount Scattering Experiment (BASSEX) was conducted to measure the scattered fields of the two seamounts which form the Kermit-Roosevelt Seamount Complex in the Northeast Pacific Ocean during September and October of 2004. BASSEX is the first experiment to measure acoustic arrival patterns in the scattered field of a seamount at many locations at sound path ranges of order 500 km, utilizing a rich bathymetry and sound velocity database.
Convergence zones in the forward-scattered field of seamounts at long-range are observed, created by higher order mode coupling and blockage. Acoustic ray arrival angles, travel times, and amplitudes show good agreement with parabolic equation (PE) acoustic modeling results inside the forward-scattered fields; in particular, simulated results are fairly accurate for weak surface-reflected-bottom-reflected acoustic rays.

Acoustic Scattering from Sand Dollars(Dendraster excentricus): Modeling as High Aspect Ratio Oblate Objects and Comparison to Experiment
Gregory Dietzen, S.M., 2008
Timothy Stanton, Andone Lavery, Advisors

Benthic shells can contribute greatly to the scattering variability of the ocean bottom, particularly at low grazing angles. Among the effects of shell aggregates are increased scattering strength and potential subcritical angle penetration of the seafloor. Sand dollars (Dendraster excentricus) occur commonly in the ocean and have been shown to be significant scatters of sound. In order to understand more fully the scattering mechanisms of these organisms, the scattering from individual sand dollars was studied using several methods.
Using an approximation to the Helmholtz-Kirchhoff integral, the Kirchhoff method gives an analytic integral expression to the backscattering from an object. This integral was first solved analytically for a disk and a spherical cap, two high aspect ratio oblate shapes which simplify the shape of an individual sand dollar. A method for solving the Kirchhoff integral numerically was then developed. An exact three dimensional model of a sand dollar test was created from computed tomography scans. The Kirchhoff integral was then solved numerically for this model of the sand dollar.
The finite element method, a numerical technique for approximating the solutions to partial differential equations and integral equations, was used to model the scattering from an individual sand dollar as well. COMSOL Multiphysics was used for the implementation of the finite element method.
Modeling results were compared with published laboratory experimental data from the free field scattering of both an aluminum disk and a sand dollar. Insight on the scattering mechanisms of individual sand dollar, including elastic behavior and diffraction effects, was gained from these comparisons.

Lossy Compression and Real-Time Geovisualization for Ultra-Low Bandwith Telemetry from Untethered Underwater Vehicles
Christopher Murphy, S.M., 2008
Haunmant Singh, Advisor

Oceanographic applications of robotics are as varied as the undersea environment itself. As underwater robotics moves toward the study of dynamic processes with multiple vehicles, there is an increasing need to distill large volumes of data from underwater vehicles and deliver it quickly to human operators. While tethered robots are able to communicate data to surface observers instantly, communicating discoveries is more difficult for untethered vehicles. The ocean imposes severe limitations on wireless communications; light is quickly absorbed by seawater, and tradeoffs between frequency, bitrate and environmental effects result in data rates for acoustic modems that are routinely as low as tens of bits per second. These data rates usually limit telemetry to state and health information, to the exclusion of mission-specific science data.
In this thesis, I present a system designed for communicating and presenting science telemetry from untethered underwater vehicles to surface observers. The system's goals are threefold: to aid human operators in understanding oceanographic processes, to enable human operators to play a role in adaptively responding to mission-specific data, and to accelerate mission planning from one vehicle dive to the next. The system uses standard lossy compression techniques to lower required data rates to those supported by commercially available acoustic modems (O(10) - O(100) bits per second).
As part of the system, a method for compressing time-series science data based upon the Discrete Wavelet Transform (DWT) is explained, a number of low-bitrate image com- pression techniques are compared, and a novel user interface for reviewing transmitted telemetry is presented. Each component is motivated by science data from a variety of actual Autonomous Underwater Vehicle (AUV) missions performed in the last year.

A Computational Tool for the Rapid Design and Prototyping of Propellers for Underwater Vehicles
Kathryn D'Epagnier, S.M., 2007
Richard Kimball, Advisor

An open source, MATLAB™-based propeller design code MPVL was improved to include rapid prototyping capabilities as well as other upgrades as part of this effort. The resulting code, OpenPVL is described in this thesis. In addition, results from the development code BasicPVL are presented. An intermediate code, BasicPVL, was created by the author while OpenPVL was under development, and it provides guidance for initial propeller designs and propeller efficiency analysis. OpenPVL is part of the open source software suite of propeller design codes, OpenProp. OpenPVL is in the form of a Graphical User Interface (GUI) which features both a parametric design technique and a single propeller geometry generator. This code combines a user-friendly interface with a highly modifiable platform for advanced users. This tool offers graphical propeller design feedback while recording propeller input, output, geometry, and performance. OpenPVL features the ability to translate the propeller design geometry into a file readable by a Computer Aided Design (CAD) program and converted into a 3D-printable file. Efficient propellers reduce the overall power requirements for Autonomous Underwater Vehicles (AUVs), and other propulsion-powered vehicles. The focus of this study is based on the need of propeller users to have an open source computer-based engineering tool for the rapid design of propellers suited to a wide range of underwater vehicles. Propeller vortex lattice lifting line (PVL) code in combination with 2D foil theory optimizes propeller design for AUVs. Several case studies demonstrate the functionality of OpenPVL, and serve as guides for future propeller designs. The first study analyzes propeller thruster performance characteristics for an off-the-shelf propeller, while the second study demonstrates the process for propeller optimization—from the initial design to the final file that can be read by a 3D printer. The third study reviews the complete process of the design and production of an AUV propeller. Thus, OpenPVL performs a variety of operations as a propeller lifting line code in streamlining the propeller optimization and prototyping process.

Setup in the Surfzone
Alex Apotsos, Ph.D., 2007
Britt Raubenheimer, Advisor

Surfzone wave height transformation and wave-breaking-driven increases in the mean sea level (setup) are examined on alongshore-uniform beaches with alongshore homogeneous and inhomogeneous wave forcing. While previously derived models predict wave heights adequately (root-mean-square errors typically less than 20%), the models can be improved by tuning a free parameter or by using a new parameterization based on the deep-water wave height. Based on a sensitivity analysis of the cross-shore momentum balance used to predict setup, a one-dimensional (1-D) model is developed that includes wave rollers and bottom stress owing to the mean offshore-directed flow. The model predicts setup accurately at three alongshore homogeneous field sites, as well as at a site where the incident wave field is alongshore non-uniform, suggesting that setup is driven primarily by the cross-shore (1-D) forcing. Furthermore, alongshore gradients of setup can be important to driving alongshore flows in the surfzone, and the 1-D setup model predicts these gradients accurately enough to simulate the observed flows.

Laboratory Evaluation of Laser-Induced Breakdown Spectroscopy (LIBS) as a new in situ Chemical Sensing Technique for the Deep Ocean
Anna P.M. Michel, Ph.D., 2007
Alan Chave, Advisor

Laser-induced breakdown spectroscopy (LIBS) possesses many of the characteristics required for in situ chemical sensing, and is a promising technique for field measurements in extreme environments. In this work, laboratory experiments validate the LIBS technique in a simulated deep ocean environment to pressures up to 2.76 x 107 Pa. A key focus of this work is the validation that select elements important for understanding hydrothermal vent fluid chemistry (Na, Ca, Mn, Mg, K, and Li) are detectable using LIBS. A data processing scheme that accurately deals with the extreme nature of laser-induced plasma formation was developed that allows for statistically accurate comparisons of spectra. The use of both single and double pulse LIBS for high pressure bulk aqueous solutions is explored and the system parameters needed for the detection of the key analytes are optimized. Using both single and double pulse LIBS, the limits of detection were found to be higher than expected as a result of the spectrometer used in this experimentation. However, the results of this validation show that LIBS possesses the characteristics to be a viable chemical sensing method for in situ analyte detection in high pressure environments like the deep ocean.

Effects of Internal Waves on Low Frequency, Long Range Acoustic Propogation in the Deep Ocean
Jinshan Xu, Ph.D., 2007
Tim Duda, John Colosi, Advisors

This thesis covers a comprehensive analysis of long-range, deep-ocean, low-frequency, sound propagation experimental results from the North Pacific Ocean. The statistics of acoustic fields after propagation through internal-wave-induced sound-speed fluctuations are explored experimentally and theoretically. The thesis investigated the environmental data by exploring the space-time scales of ocean sound speed variability and the contributions from different frequency bands. The results impose hard bounds on the characteristic scales of sound speed fluctuations in this region for both internal-wave and mesoscale band fluctuations. The thesis then analyzed the low frequency, broadband sound arrivals obtained in the North Pacific Ocean. The observed acoustic variability is compared with predictions based on the weak fluctuation theory, and direct parabolic equation (PE) Monte Carlo simulations. The comparisons show that a resonance condition exists between the local acoustic ray and the internal wave field such that only the internal-waves whose crests are parallel to the local ray path will contribute to acoustic scattering. This is the first observational evidence for the acoustic ray and internal wave resonance. Finally, the thesis examined the evolution with distance, of the acoustic arrival pattern of the off-axis sound source transmissions in the Long-range Ocean Acoustic EXperiment. The observations of mean intensity time-fronts are compared to the deterministic ray, PE and normal mode calculations. It is found the diffraction effect is dominant in the shorter range. In the longer range, the scattering effect smears the energy in both the spatial and temporal scales, and thus has a dominant role in the finale region.

Three-Dimensional Propagation and Scattering around a Conical Seamount
Wenyu Luo, Ph.D., 2007
Henrik Schmidt, Advisor

In this thesis, a numerically efficient three-dimensional propagation and scattering model is developed based on the three-dimensional coupled mode theory for axisymmetric bathymetry. The three-dimensional coupled mode approach applied in this thesis is fundamentally identical to the one applied in earlier models, such as the one presented by Taroudakis. Thus, it is based on a Fourier expansion of the acoustic field around a seamount, with each azimuthal expansion coefficient being represented by a two-way coupled mode formulation. However, earlier formulations were severely limited in terms of frequency, size and geometry of the seamount, the seabed composition, and the distance between the source and the seamount, and are totally inadequate for modeling high-frequency, large-scale seamount problems. By introducing a number of changes in the numerical formulation and using a standard normal mode model (C-SNAP) for determining the fundamental modal solutions and coupling coefficients, orders of magnitude improvement in efficiency and fidelity has been achieved, allowing for realistic propagation and scattering scenarios to be modeled, including effects of seamount roughness and realistic sedimentary structure. Also, by the simple superposition principle, the computational requirements are made independent of the distance between the seamount and the source and receivers, and dependent only on the geometry of the seamount and the frequency of the source.

Design Considerations for Engineering Autonomous Underwater Vehicles
Vikrant P. Shah, S.M., 2007
Hanumant Singh, Henrik Schmidt, Advisors

Autonomous Underwater Vehicles (AUVs) have been established as a viable tool for Oceanographic Sciences. Being untethered and independent, AUVs fill the gap in Ocean Exploration left by the existing manned submersible and remotely operated vehicles (ROV) technology. AUVs are attractive as cheaper and efficient alternatives to the older technologies and are breaking new ground in many applications. Designing an autonomous vehicle to work in the harsh environment of the deep ocean comes with its set of challenges. This paper discusses how the current engineering technologies can be adapted to the design of AUVs.
Recently, as the AUV technology has matured, we see AUVs being used in a variety of applications ranging from sub-surface sensing to sea-floor mapping. The design of the AUV, with its tight constraints, is very sensitive to the target application. Keeping this in mind, the goal of this thesis is to understand how some of the major issues affect the design of the AUV. This paper also addresses the mechanical and materials issues, power system design, computer architecture, navigation and communication systems, sensor considerations and long term docking aspects that affect AUV design.
With time, as the engineering sciences progress, the AUV design will have to change in order to optimize its performance. Thus, the fundamental issues discussed in this paper can assist in meeting the challenge of maintaining AUV design on par with modern technology.

Geoacoustic Inversion by Mode Amplitude Perturbation
Travis L. Poole, Ph.D., 2007
George Frisk, James Lynch, Advisors

This thesis introduces an algorithm for inverting for the geoacoustic properties of the seafloor in shallow water. The input data required by the algorithm are estimates of the amplitudes of the normal modes excited by a low-frequency pure-tone sound source, and estimates of the water column sound speed profiles at the source and receiver positions. The algorithm makes use of perturbation results, and computes the small correction to an estimated background profile that is necessary to reproduce the measured mode amplitudes. Range-dependent waveguide properties can be inverted for so long as they vary slowly enough in range that the adiabatic approximation is valid. The thesis also presents an estimator which can be used to obtain the input data for the inversion algorithm from pressure measurements made on a vertical line array (VLA). The estimator is an Extended Kalman Filter (EKF), which treats the mode amplitudes and eigenvalues as state variables. Numerous synthetic and real-data examples of both the inversion algorithm and the EKF estimator are provided. The inversion algorithm is similar to eigenvalue perturbation methods, and the thesis also presents a combination mode amplitude/eigenvalue inversion algorithm, which combines the advantages of the two techniques.

Applied Stochastic Eigen-Analysis
Rajesh Rao Nadakuditi, Ph.D., 2007
Alan Edelman, Advisor

The first part of the dissertation investigates the application of the theory of large random matrices to high-dimensional inference problems when the samples are drawn from a multivariate normal distribution. New parametric techniques for testing and estimation of signals in white noise are developed that exhibit robustness to high-dimensionality, sample size constraints and eigenvector misspecification. By interpreting the eigenvalues of the sample covariance matrix as an interacting particle system, the existence of a phase transition phenomenon in the largest (``signal'') eigenvalue is derived using heuristic arguments. This exposes a fundamental limit on the identifiability of low-level signals due to sample size constraints when using the sample eigenvalues alone.
The analysis is extended to address a problem in sensor array processing, posed by Baggeroer and Cox, on the distribution of the outputs of the Capon-MVDR beamformer when the sample covariance matrix is diagonally loaded.
The second part of the dissertation investigates the limiting distribution of the eigenvalues and eigenvectors of a broader class of random matrices. A powerful method is proposed that expands the reach of the theory beyond the special cases of matrices with Gaussian entries; this simultaneously establishes a framework for computational (non-commutative) ``free probability'' theory.

Stochastic Mapping for Chemical Plume Source Localization with Application to Autonomous Hydrothermal Vent Discovery
Michael V. Jakuba, Ph.D., 2007
Dana Yoerger, Advisor

This thesis presents a stochastic mapping framework for autonomous robotic chemical plume source localization in environments with multiple sources. Turbulent flows make the spatial relationship between the detectable manifestation of a chemical plume source, the plume itself, and the location of its source inherently uncertain. Search domains with multiple sources compound this uncertainty because the number of sources as well as their locations is unknown a priori.
Our framework for stochastic mapping is an adaptation of occupancy grid mapping where the binary state of map nodes is redefined to denote either the presence (occupancy) or absence of an active plume source. Occupancy grid maps explicitly represent explored but empty portions of the domain as well as probable source locations.
Application to hydrothermal plume data collected by the autonomous underwater vehicle ABE during vent prospecting operations in both the Pacific and Atlantic oceans verifies the utility of the approach. The resulting maps enable nested surveys for homing-in on seafloor vent sites to be carried out autonomously. This eliminates inter-dive processing, recharging of batteries, and time spent deploying and recovering the vehicle that would otherwise be necessary with survey design directed by human operators.

Acoustic Scattering of Broadband Echolocation Signals from Prey of Blainville's Beaked Whales: Modeling and Analysis
Benjamin A. Jones, SM, 2006
Timothy Stanton, Andone Lavery, Advisors

Blainville’s beaked whales (Mesoplodon densirostris) use broadband, ultrasonic echolocation signals (27 to 57 kHz) to search for, localize, and approach prey that generally consist of mid-water and deep-water ?shes and squid. Although it is well known that the spectral characteristics of broadband echoes from marine organisms are a strong function of size, shape, orientation and anatomical group, little is known as to whether or not these or other toothed whales use spectral cues in discriminating between prey and non-prey. In order to study the prey-classi?cation process, a stereo acoustic tag was mounted on a Blainville’s beaked whale so that emitted clicks and corresponding echoes from prey could be recorded. A comparison of echoes from prey selected by the whale and those from randomly chosen scatterers suggests that the whale may have, indeed, discriminated between echoes using spectral features and target strengths. Speci?cally, the whale appears to have favored prey with one or more deep nulls in the echo spectra as well as ones with higher target strength.
A three-dimensional, acoustic scattering model is also developed to simulate broadband scattering from squid, a likely prey of the beaked whale. This model applies the distorted wave Born approximation (DWBA) to a weakly-scattering, inhomogeneous body using a combined ray trace and volume integration approach. Scatterer features are represented with volume elements that are small (less than 1/12th of the wavelength) for the frequency range of interest (0 to 120 kHz). Ranges of validity with respect to material properties and numerical considerations are explored using benchmark computations with simpler geometries such as ?uid-?lled spherical and cylindrical ?uid shells. Modeling predictions are compared with published data from live, freely swimming squid. These results, as well as previously published studies, are used in the analysis of the echo spectra of the whale’s ensoni?ed targets.

Emulating the Fast-Start Swimming Performance of the Chain Pickerel (Esox niger) Using a Mechanical Fish Design
Matthew N. Watts, SM, 2006
Michael Triantafyllou, Advisor

Mean maximum start-up accelerations and velocities achieved by the fast-start specialist, northern pike, are reported at 120 ms^-2 and 4 ms^-1, respectively (Harper and Blake, 1990). In this thesis, a simple mechanical system was created to closely mimic the startle response that produces these extreme acceleration events. The system consisted of a thin metal beam covered by a urethane rubber fish body. The mechanical fish was held in curvature by a restraining line and released by a pneumatic cutting mechanism. The potential energy in the beam was transferred into the fluid, thereby accelerating the fish. The fish motion was recorded and the kinematics analyzed while using a number of different tail shapes and materials.

Performance of the mechanical fish was determined by maximum acceleration, peak and averaged maximum velocity, and hydrodynamic efficiency. Maximum start-up acceleration was calculated at 48 ms^-2. Peak and averaged maximum velocity was calculated at 0.96 ms^-1 and 0.8 ms^-1, respectively. The hydrodynamic efficiency of the fish, calculated by the transfer of energy, was 11%. Flow visualization of the mechanical fast-start wake was also analyzed. The visualization uncovered two specific vortex-shedding patterns; a single and a double-vortex pattern are described.

Application of Statistical Learning Theory to Plankton Image Analysis
Qiao Hu, Ph.D., 2006
Cabell Davis, Hanu Singh, Advisors

A fundamental problem in limnology and oceanography is the inability to quickly identify and map distributions of plankton. This thesis addresses the problem by applying statistical machine learning to images collected by the Video Plankton Recorder. The research is focused on development of a real-time automatic plankton recognition system to estimate plankton abundance. The system includes four major components: pattern representation/feature measurement, feature extraction/selection, classification, and abundance estimation. After an extensive study on a traditional learning vector quantization (LVQ) neural network (NN) classifier built on shape-based features and different pattern representation methods, I developed a classification system combined multi-scale co-occurrence matrices feature with a support vector machine classifier. This new method outperforms the traditional shape-based NN classifier by 12% in classification accuracy. Subsequent plankton abundance estimates are improved in the regions of low relative abundance by more than 50%. Two rejection metrics were developed. One was based on the Euclidean distance in the feature space for NN classifier. The other was dual-classification system. Dual-classification method yields almost as good abundance estimation as human labeling on a very large real-world data. The distance rejection metric for NN classifier might be more useful to reject outliers.

Estimation and Tracking of Rapidly Time-Varying Broadband Acoustic Communication Channels
Weichang Li, Ph.D., 2006
James Preisig, Advisor

This thesis develops methods for estimating wideband shallow-water acoustic communication channels. The very shallow water wideband channel has three distinct features: large dimension caused by extensive delay spread; limited number of degrees of freedom (DOF) due to resolvable paths and inter-path correlations; and rapid fluctuations induced by scattering from the moving sea surface. Based on state-space channel modeling, the thesis first develops algorithms that jointly estimate the channel and its dynamics, based on the Extended Kalman Filter (EKF) and the Expectation Maximization (EM) approach respectively. Analysis shows conceptual parallels, including an identical second-order innovation form shared by the EKF modification and the suboptimal EM, and the issue of parameter identifiability due to channel structure, reflected as parameter unobservability in EKF and insufficient excitation in EM. A two-model based EKF and a subspace EM algorithm, which selectively track dominant taps and reduce prediction error, are proposed to overcome the identifiability issue. The second part of the thesis develops algorithms that explicitly find the sparse estimate of the delay-Doppler spread function. The study contributes to better understanding of the channel physical constraints on algorithm design and performance improvement. It may be generalized to other applications where dimensionality and variability collide.

Self Consistent Bathymetric Mapping from Robotic Vehicles in the Deep Ocean
Christopher N. Roman, Ph.D., 2005
Hanu Singh, Advisor

Obtaining accurate and repeatable navigation for robotic vehicles in the deep ocean is difficult and consequently a limiting factor when constructing vehicle-based bathymetric maps. Errors in vehicle position estimation will corrupt the otherwise accurate acoustic range data and reduce map integrity. This thesis presents a methodology to produce self-consistent maps and simultaneously improve vehicle position estimation by exploiting accurate local navigation and incorporating terrain relative measurements. Our technique utilizes small terrain ``sub-maps'' that can be pairwise registered and used to additionally constrain the vehicle position estimates in accordance with actual bottom topography. A delayed state Kalman filter is used to incorporate these sub-map registrations as relative position measurements between previously visited vehicle locations. Archiving previous positions in a filter state vector allows for continual adjustment of the sub-map locations. The terrain registration is accomplished using a two dimensional correlation and a six degree of freedom point cloud alignment tailored for bathymetric data. A complete bathymetric map is then created from the union of all sub-maps that have been aligned in a consistent manner. Experimental results from the fully automated processing of a multibeam survey over the TAG hydrothermal structure at the Mid-Atlantic ridge are presented to validate the proposed method.

Large-Area Visually Augmented Navigation for Autonomous Underwater Vehicles
Ryan M. Eustice, Ph.D., 2005
Hanu Singh, John Leonard, Advisors

This thesis describes a vision-based, large-area, simultaneous localization and mapping algorithm (SLAM) that respects the low-overlap imagery constraints typical of autonomous underwater vehicles while exploiting the inertial sensor information that is routinely available on such platforms. We adopt a systems-level approach exploiting the complementary aspects of inertial sensing and visual perception from a calibrated pose-instrumented platform. This systems-level strategy yields a robust solution to underwater imaging that overcomes many of the unique challenges of a marine environment (e.g., unstructured terrain, low-overlap imagery, moving light source).
Our large-area SLAM algorithm recursively incorporates relative-pose constraints using a view-based representation that exploits exact sparsity in the Gaussian canonical form. This sparsity allows for efficient O(n) update complexity in the number of images composing the view-based map by utilizing recent multilevel relaxation techniques. We show that our algorithmic formulation is inherently sparse unlike other feature-based canonical SLAM algorithms, which impose sparseness via pruning approximations. In particular, we investigate the sparsification methodology employed by SEIF and offer new insight as to why, and how, its approximation can lead to inconsistencies in the estimated state errors. Lastly, we present a novel algorithm for efficiently extracting consistent marginal covariances useful for data association from the information matrix.

Blooms of the Toxic Dinoflagellate Alexandrium fundyense in the Gulf of Maine: Investigations using a Physical-Biological Model
Charles A. Stock, Ph.D., 2005
Dennis McGillicuddy, Advisor

Blooms of the toxic dinoflagellate Alexandrium fundyense are annually recurrent in the western Gulf of Maine (WGOM) and pose a serious economic and public health threat. In this thesis, a biological model of the A. fundyense life cycle developed from laboratory and field data is combined with a circulation model to test hypotheses concerning the factors governing A. fundyense blooms in the springs of 1993 and 1994. Several biological model structures are tested against the 1993 observations using the method of maximum likelihood. Analysis suggests that either nitrogen limitation or mortality is needed to match observed patterns. Application of 1993 optimized values to 1994 shows similar fit is maintained only for those runs with nitrogen limitation. Model diagnosis suggests that cysts germinating offshore of Casco Bay provide a plausible source of cells for the blooms, although inputs from the eastern Gulf of Maine become significant as the season progresses. Net growth rates are generally low and have a limited influence on bloom magnitude. The model suggests that differences in shellfish toxicity during the two years result primarily from differences in the wind and its influence on along and cross-shore transport of cells.

Inversion for Subbottom Sound Velocity Profiles in the Deep and Shallow Ocean
Luiz L. Souza, Ph.D., 2005
George Frisk, Advisor

We investigate the application of acoustic measurements in the ocean to measure the sound velocity profile (svp) in the subbottom. For the deep water ocean, an exact method based on the Gelfand-Levitan integral equation is evaluated. The input data is the complex plane-wave reflection coefficient. We apply the method to experimental acoustic data and estimate the reflection coefficient and the svp in the seabed.
For the shallow ocean, an inverse eigenvalue technique is developed. The input data are range-varying eigenvalues associated with propagating modes. We investigate the estimation of eigenvalues from fields measured in laterally varying shallow environments, in particular the errors associated with autoregressive (AR) methods. We propose and analyze two sequential estimators, one for AR coefficients, another for the zeros of the AR characteristic polynomial. The decimation of pressure fields is analyzed as a tool to improve eigenvalue estimation.
The solution to the inverse eigenvalue problem is proposed in the form of a Kalman filter. The resolution and variance of the inferred svp are analyzed in terms of the Cramer--Rao lower bound and the Backus--Gilbert (BG) resolution theory. A method is developed to compensate for the Doppler deviation observed in experiments with moving sources.

Advances in the Visualization and Analysis of Boundary Layer Flow in the Swimming Fish
Erik Anderson, Ph.D., 2005
Mark Grosenbaugh, Wade McGillis, Advisors

In biology, the importance of fluid drag, diffusion, and heat transfer suggest the boundary layer as an important subject of investigation, however, the complexities of biological systems present significant and unique challenges to analysis by experimental fluid dynamics. In this investigation, a system for automatically profiling the boundary layer was developed and the boundary layer over mackerel, bluefish, scup and eel was profiled. The profiling system combined robotics, particle imaging velocimetry, a custom particle tracking code, and an automatic boundary layer analysis code. Over 100,000 image pairs of flow in the boundary layer were acquired in swimming fish alone.
Friction drag in rigid and swimming fish was determined by integrating wall shear stress, to, over the surface of the fish. In general, there was no significant difference in friction drag between the rigid-body and swimming cases. In swimming, separation was, on average, delayed. Therefore, pressure drag was estimated on the basis of thickness ratio and used to calculate an upper-bound total drag on a swimming fish. Total drag was used to determine the required muscle power output during swimming. to and boundary layer thickness oscillated with undulatory phase. The magnitude of oscillation appears to be linked to body wave amplitude.

Large Scale Structure from Motion for Autonomous Underwater Vehicle Surveys
Oscar Pizarro, Ph.D., 2004
Hanumant Singh, Advisor

Our ability to image extended underwater scenes is severely limited by attenuation and backscatter. Generating a composite view from multiple overlapping images is usually the most practical and flexible way around this limitation. In this thesis we look at the general constraints associated with imaging from underwater vehicles for scientific applications -- low overlap, non-uniform lighting and unstructured motion -- and present a methodology for dealing with these constraints toward a solution of the problem of large area 3D reconstruction. Our approach assumes navigation data is available to constrain the structure from motion problem. We take a hierarchical approach where the temporal image sequence is broken into subsequences that are processed into 3D reconstructions independently. These submaps are then registered to infer their overall layout in a global frame. From this point a bundle adjustment refines camera and structure estimates. We demonstrate the utility of our techniques using real data obtained during a SeaBED AUV coral reef survey. Test tank results with ground truth are also presented to validate the methodology.

Performance Analysis of Subperture Processing Using a Large Aperture Planar Towed Array
Jennifer Watson, Ph.D., 2004
Arthur Baggeroer, Advisor

In recent years the focus of passive detection and localization of submarines has moved from the deep ocean into the littoral regions. The problem of passive detection in these regions is complicated by multipath propagation with high transmission loss. Large aperture planar arrays have the potential to improve detection performance with their high resolution and high gain, but are susceptible to two performance degradation mechanisms: limited spatial coherence of signals and nonstationarity of high bearing rate interference sources common in littoral regions of strategic importance. This thesis presents subarray processing as a method of improving passive detection using large arrays. This thesis develops statistical models for the detection performance of three adaptive, sample-covariance-based subarray processing algorithms which incorporate the effects of limited spatial coherence and snapshot support. The performance of the optimum processor conditioned on known data covariances is derived as well for comparison. These models are used to compare subarray algorithms and partitioning schemes in a variety of environments using several propagation models. The analysis shows a tradeoff between the required adaptive degrees of freedom, snapshot support, and adaptive resolution. This thesis shows that for both plane-wave and matched-field processing, the Conventional-Then-Adaptive (CTA) algorithm optimizes this tradeoff most efficiently.

A Distributed Approach to Underwater Acoustic Communications
Christopher Bohner, S.M., 2003
Arthur Baggeroer, Advisor

A novel distributed underwater acoustic networking (UAN) protocol suitable for ad-hoc deployments of both stationary and mobil nodes dispersed across a relatively wide coverage area is presented. Nodes are dynamically clustered in a distributed manner based on the estimated position of one-hop neighbor nodes within a shallow water environment. The spatial dynamic cellular clustering scheme allows scalable communication resource allocation and channel reuse similar in design to land-based cellular architrectures, except devoid of the need for a centralized controlling infrastructure. Simulation results demonstrate that relatively high degrees of interference immunity, network connectivity, and network stability can be achieved despite the severe limitations of the underwater acoustic channel.

Mixing Processes and Hydraulic Control in a Highly Stratified Estuary
Daniel MacDonald, Ph.D., 2003
W. Rockwell Geyer, Advisor

This thesis utilizes field data from the Fraser River Estuary (British Columbia, Canada) to investigate the nature of mixing processes in a highly stratified environment. During the late ebb, a stationary front exists at the Fraser mouth. Although streamwise densimetric Froude numbers at the front are supercritical, the front is oriented such that Froude numbers are equal to one in a reference frame perpendicular to the front. This represents a robust extension of established two-layer hydraulic theory to three dimensions, and implies similarity with trans-sonic flows. Mixing processes were evaluated at the mouth using a control volume approach to isolate mean vertical entrainment processes from turbulent processes. Observed turbulent dissipation is on the order of 10^-3 m²s^-3. Flux Richardson numbers are confined between 0.15 to 0.2, with gradient Richardson numbers between 0.2 and 0.25. These results are consistent with previous laboratory studies, but represent energetic conditions several orders of magnitude higher. In the estuarine channel, mixing variability was investigated using control volume and overturn scale methods. Spatially, mixing was observed to be more intense near a width constriction on the order of 25%. Temporally, more dominant mixing was observed during ebbs, due to increases in both vertical shear and stratification.

Geoacoustic Inversion in Laterally Varying Shallow-Water Environments Using High-Resolution Wavenumber Estimation
Kyle Becker, Ph.D., 2002
George Frisk, Advisor

An experimental method for extracting horizontal wavenumber spectra for point-source acoustic fields in laterally varying shallow water waveguides is discussed. The experiment was designed to provide input data for a perturbative inversion method for inferring the geoacoustic properties of the bottom. The input data are discrete values of horizontal wavenumbers that correspond the propagating modes in a shallow-water waveguide. Based on the asymptotic Hankel transform pair relationship between the complex pressure field and the depth-dependent Green's function, a high-resolution autoregressive (AR) spectral estimator is applied to extract wavenumber content. The estimator is characterized in terms of it performance on short-aperture noisy data. The estimator is then used on short-aperture synthetic acoustic data for extracting local horizontal wavenumber content for doing range-dependent geoacoustic inversion. Results are discussed in terms of discrete and continuous changes in the waveguide environment. The estimator is then applied to real data for a fixed receiver and towed source. For a source speed of 2 m/s, a shift was observed in the measured wavenumbers. Based on this observation, a method for measuring modal group velocity is presented from measurements of the shifted wavenumbers for a source towed out and back along the same track.

Enhancement of Fine Particle Deposition to Permeable Sediments
J. Stephen Fries, Ph.D., 2002
John Trowbridge, Advisor

Generally, fine particle deposition rates are assumed to be equivalent to the suspension settling velocity, therefore, deposition rates in excess of settling are considered enhanced. Flume observations of deposition were made using treatments that covered a wide range of flow, particle, and bed conditions. Specific treatments demonstrated large enhancements (up to eight times settling). Delivery of particles to the interface is important, but models based on delivery alone failed to predict the observed enhancement.

This necessitated the development of a new model based on a balance between delivery and filtration in the bed. Interfacial diffusion was chosen as a model for particle delivery. The model performed well in prediction of flow conditions, but there remained a discrepancy between predictions and observed deposition rate, especially for treatments with significant enhancement. Filtration of particles by the bed is a useful framework for retention, but the shear in the interstitial flow may introduce additional factors not included in traditional filtration experiments. The observation of enhanced deposition to flat sediment beds reinforces the importance of permeable sediments to the mediation of transport from the water column to the sediment bed.

Acoustic Scattering by Axisymmetric Finite-length Bodies with Application to Fish: Measurement and Modeling
D. Benjamin Reeder, Ph.D., 2002
Timothy Stanton, Advisor

This thesis investigates the complexities of acoustic scattering by finite bodies in general and by fish in particular through the development of an advanced acoustic scattering model and detailed laboratory acoustic measurements. A general acoustic scattering model is developed that is accurate and numerically efficient for a wide range of frequencies, angles of orientation, irregular axisymmetric shapes and boundary conditions. An extensive series of broadband acoustic backscattering measurements has been conducted involving alewife fish (Alosa pseudoharengus). A broadband (40-95 kHz) chirp was used to insonify live, adult alewife that were tethered while being rotated in 1-degree increments over all angles in two planes of rotation (lateral and dorsal/ventral). Spectral analysis correlates frequency dependencies to morphology and orientation. Pulse compression processing temporally resolves multiple returns from each individual which show good correlation with size and orientation, and demonstrate that there exists more than one significant scattering feature in the animal. Imaging technologies used to exactly measure the morphology of the scattering features of fish are used for morphological evaluation and incorporation into the scattering model. Studies such as this one, which combine scattering models with high-resolution morphological information and high-quality laboratory data, are crucial to the quantitative use of acoustics in the ocean.

Horizontal Linear Array Sensor Localization and Preliminary Coherence Measurements from the 2001 ASIAEX South China Sea Experiment
Theodore Schroeder, M.S., 2002
James Lynch, Advisor

Arrays obtain their gain by coherently adding the energy that impinges on them. To maximize the efficiency of an array, the size of the aperture over which the signal remains coherent is needed. Scattering of sound by the ocean environment reduces the coherence of acoustic signals, and thereby limits the useful aperture of an array. This thesis examines data collected in the South China Sea (SCS) component of the 2001 Asian Seas International Acoustic Experiment (ASIAEX), where a Horizontal Linear Array (HLA) was deployed on the continental shelf to study transverse array coherence in a coastal environment. 224 Hz and 400 Hz sources were placed on the continental slope and a 400 Hz source was placed on the shelf to provide up slope and along shelf propagation paths. In this thesis, sensor-to-sensor correlations of one day of transmissions from these three sources were completed to give a first look at coherence lengths of the HLA. The thesis also analyzes the data from the Long Base Line (LBL) navigation system and two days of light bulb drops to provide array sensor localization. Accurate sensor positions are needed to determine the correlation versus sensor separation distance and ultimately the array coherence length.

Performance Bounds on Matched-Field Methods for Source Localization and Estimation of Ocean Environmental Parameters
Wen Xu, Ph.D., 2001
Arthur Baggeroer, Advisor

Matched-field methods concern estimation of source location and/or ocean environmental parameters by exploiting full wave modeling of acoustic waveguide propagation. Typical estimation performance demonstrates two fundamental limitations. First, sidelobe ambiguities dominate the estimation at low signal-to-noise ratio (SNR), leading to a threshold performance behavior. Second, most matched-field algorithms show a strong sensitivity to environmental/system mismatch, introducing some biased estimates at high SNR.

In this thesis, a quantitative approach for ambiguity analysis is developed so that different mainlobe and sidelobe error contributions can be compared at different SNR levels. Two large-error performance bounds, the Weiss-Weinstein bound (WWB) and Ziv-Zakai bound (ZZB), are derived for the attainable accuracy of matched-field methods. To include mismatch effects, a modified version of the ZZB is proposed.

Performance analyses are implemented for source localization under a typical shallow water environment chosen from the Shallow Water Evaluation Cell Experiments (SWellEX). The performance predictions describe the simulations of the maximum likelihood estimator (MLE) well, including the mean square error in all SNR regions as well as the bias at high SNR. The threshold SNR and bias predictions are also verified by the SWellEX experimental data processing. These developments provide tools to better understand some fundamental behaviors in matched-field performance and provide benchmarks to which various ad hoc algorithms can be compared.