Keywords: Acoustic telemetry and communication, Signal coherence and fluctuation, New vehicles, methods, and technologies for midwater exploration
Abstract: To this day the underwater world remains a largely unexplored terrain due to its vast size and extreme conditions like high pressure, temperature change and darkness. Furthermore, the high conductivity of saltwater severely limits conventional communication methods used in the air. The reliance of all localization frameworks on communication is a significant factor contributing to the lack of a global underwater positioning system. Therefore, swarm systems have gained increasing attention, as they can establish local reference frames, thereby enabling higher-level applications such as surveillance systems. These underwater swarm systems typically consist of a connected network of autonomous units, namely autonomous underwater vehicles (AUVs) and autonomous surface vehicles (ASVs), which serve as the interfaces for communication above water. A key requirement for such swarms is reliable online localization and navigation of the vehicles. This paper presents an analysis of the trade-offs between optimum signal configurations for delay-spread and Doppler-shift resistance and formulates a corresponding optimization framework, establishing a benchmark for reliable transmission of localization and navigation data over the underwater channel (UWC) at distances up to 150m.

Keywords: Acoustic telemetry and communication
Abstract: Orthogonal Chirp Division Multiplexing with Index Modulation (OCDM-IM) is a promising multicarrier modulation scheme for Underwater Acoustic (UWA) communications, offering improved robustness to frequency-selective fading and synchronization errors while exploiting both symbol and subchirp indices to enhance spectral efficiency. However, as a multicarriers mordulations, its practical deployment is challenged by the high Peak-to-Average Power Ratio (PAPR) resulting from the coherent superposition of subchirps. High PAPR can induce nonlinear distortion in power amplifiers and reduce energy efficiency, which is particularly critical for energy-constrained underwater platforms. In this paper, we propose a novel PAPR reduction technique for OCDM-IM based on a dynamically selected set of inactive subchirps. These subchirps generate a cancellation signal that suppresses large signal peaks without requiring a priori reservation of subchirp indices, thereby preserving the spectral efficiency of OCDM-IM. Simulation results demonstrate that the proposed method achieves substantial PAPR reduction while maintaining robust BER performance and a computational complexity comparable to conventional OCDM-IM. Our simulation also show a tradeoff between PAPR suppression, energy efficiency, and detection performance.

Keywords: Acoustic telemetry and communication
Abstract: Orthogonal Time Frequency Space (OTFS) mod- ulation has emerged as a robust waveform for highly time- varying channels such as underwater acoustic (UWA) environ- ments. However, conventional formulations treat the channel as a linear operator in the delay–Doppler domain and do not explicitly capture the non-commutative structure induced by multipath propagation and Doppler effects. In this work, we propose a novel braid-theoretic framework for OTFS channels, where propagation paths are modeled as braid generators and Doppler shifts correspond to topological twisting operations. This representation leads to a non-commutative algebraic model of the channel and enables a new notion of channel equivalence using braid group relations. Furthermore, we propose a detection framework based on braid invariants that is inherently robust to Doppler variations. The proposed formulation introduces a new algebraic-topological perspective for OTFS-based underwater communication systems. I. INTRODUCTION Underwater acoustic communication systems are known for their challenging propagation conditions, including severe multipath spread, strong Doppler effects, and rapid time vari- ation. These characteristics significantly degrade the perfor- mance of conventional modulation schemes such as OFDM due to inter-carrier interference. Orthogonal Time Frequency Space (OTFS) modulation has recently been proposed as a promising alternative because it represents information symbols in the delay–Doppler domain, where the channel exhibits sparsity and structured behavior. This representation improves robustness against time variation and Doppler shifts. Despite these advantages, existing OTFS formulations rely on linear convolution models in the delay–Doppler domain. These models do not explicitly account for the ordering effects of multipath propagation, where delay and Doppler interactions are inherently non-commutative. In this paper, we introduce a braid-theoretic interpretation of OTFS channels. By representing propagation paths as braid strands and Doppler shifts as twisting operations, we develop a non-commutative algebraic model that captures the intrinsic structure of underwater channels.

Keywords: Signal coherence and fluctuation, Sound propagation and scattering, Ocean noise
Abstract: The underwater acoustic channel presents strongly frequency-dependent absorption and non-white ambient noise, creating opportunities to trade communication reliability against low probability of detection. Prior work has focused primarily on waveform design, while frequency-dependent power allocation remains comparatively underexplored. This paper presents a waveform-level simulation-based analysis of uniform and non-uniform power allocation strategies across the 10--60 kHz acoustic frequency spectrum using a multiband OFDM waveform. Performance is evaluated in terms of bit error rate and probability of detection by a passive warden employing energy detection, whose distance from the transmitter is known. The results indicate that the preferred power spectral allocation is heavily dependent on the relative distance between the transmitter, receiver, and warden. Higher-frequency allocations can substantially reduce detectability when the receiver is sufficiently close relative to the warden, while longer receiver links incur larger penalties from high-frequency absorption, favoring lower-frequency allocations.

Keywords: Sound propagation and scattering, Signal coherence and fluctuation, Ocean energy
Abstract: Driven by applications in environmental monitoring, defense systems, and autonomous vehicles, research in underwater acoustic communications (UAC) has grown significantly in recent years. UAC systems must contend with severe channel impairments, such as attenuation, multipath, and Doppler shifts, while operating under severe power constraints. Software-defined radios (SDR) are becoming increasingly popular for UAC due to their versatility. While there is abundant research on the power consumption of various SDR hardware architectures, their energy efficiency in power-constrained underwater environments remains poorly understood. This paper presents a system-level experimental evaluation of software-defined UAC platforms, focusing on power–performance trade-offs. We compare a general-purpose processor (GPP)-based platform with a hybrid GPP–field-programmable gate array (FPGA)-based platform for transmission and reception. We find that the GPP-based platform is more energy-efficient while meeting the functional requirements for UAC. Furthermore, we implement BFSK and a WiFi-like protocol to further analyze the power-performance trade-offs of different modulation schemes. Through in-water testing, we show that the greater resilience to channel impairments provided by the more computationally complex WiFi-like protocol enables significant power savings through reduced transmit gain compared to the simpler non-coherent BFSK scheme. Despite its greater computational complexity, the WiFi-like protocol achieves a comparable bit error rate (BER) with minimal additional CPU overhead. Results from laboratory and aquatic experiments demonstrate how hardware and software architectural choices influence energy consumption and communication reliability, providing practical design guidance for energy-efficient software-defined UAC.

Keywords: Autonomous underwater vehicles, Low-cost, scalable oceanographic sensors and platforms, AI/ML for oceanographic and environmental data analysis
Abstract: Accurate underwater navigation is commonly performed using inertial sensors and a Doppler velocity log. Recently, a physics-informed inertial dead-reckoning approach offered an externally-derived inertial-based update to the navigation filter, creating a correlation between the measurement and process noise covariances. In this work, a physics-informed extended Kalman filter (EKF) with cross-covariance awareness framework is proposed for navigation of autonomous underwater vehicles (AUVs). Unlike conventional EKF that assumes independence between process and measurement errors, the proposed method explicitly incorporates the cross-covariance within the filter. This enables improved uncertainty modeling and more consistent state estimation in underwater environments. The framework is evaluated using real-world AUV trajectories. Experimental results demonstrate that the proposed cross-covariance-aware formulation consistently improves navigation accuracy. For unseen test data, the proposed method achieves an improvement of 34% over the baseline. The proposed framework provides a theoretically grounded and physically consistent solution for robust underwater inertial navigation.

Keywords: Vehicle navigation, Calibration of acoustic systems and metrology
Abstract: Accurate initialization of Doppler velocity log (DVL) systems is critical for reliable autonomous underwater vehicle (AUV) navigation. Conventional initialization pipelines comprise two sequential stages — alignment, which estimates the rotation between the inertial navigation system (INS) and DVL frames, and calibration, which corrects DVL error terms — both relying on model-based algorithms that demand complex vehicle maneuvers, external infrastructure, and significant operational overhead. This paper presents a unified, fully data-driven DVL initialization pipeline that replaces both stages with two complementary neural networks. ResAlignNet, a 1D ResNet-based model, estimates the three Euler alignment angles directly from synchronized INS and DVL measurements without requiring prescribed maneuvers. DCNet, a multi-head neural network, subsequently estimates DVL scale factor and bias terms from DVL and GNSS-RTK velocity pairs. The proposed pipeline operates on simple, nearly constant-velocity trajectories, substantially reducing initialization complexity compared to conventional approaches. The method is validated on the A-KIT dataset — a real-world AUV navigation dataset collected by the University of Haifa AUV Snapir, comprising 276 minutes of sea-trial recordings. A controlled noising pipeline simulates diverse sensor grades and error conditions. Results are evaluated using velocity root mean squared error (VRMSE) across four independent test trajectories. The proposed approach achieves an average improvement of 70% over the model-based baseline, with a minimum improvement of 65%, while accepting the same inputs as conventional methods. This work offers a practical, end-to-end solution for AUV DVL initialization that enhances mission readiness and navigation accuracy.

Keywords: Data fusion, uncertainty quantification, and predictive modeling, Autonomous underwater vehicles, MEMS-based sensing technologies and miniaturization
Abstract: Accurate process-noise covariance estimation is essential for reliable autonomous underwater vehicle navigation. This paper investigates how different training dataset strategies affect neural-aided process-noise estimation in a right-invariant extended Kalman filter (RI-IKF). A one-dimensional convolutional neural network estimates inertial sensor noise levels from inertial measurement windows. Three otherwise identical networks are trained using Sim2Real, Real2Real, and SimReal2Sim dataset strategies. Their navigation performance is evaluated on four held-out real autonomous underwater vehicle (AUV) trajectories using three independent noise realizations per trajectory. Fixed process-noise covariance and innovation-based adaptive covariance estimation are included as non-neural baselines. All three neural-aided RI-IKF variants outperform both baselines. Sim2Real achieves the lowest average position root-mean-square error of (7.70)~m, representing reductions of (55.8%) relative to the fixed-covariance baseline and (33.0%) relative to innovation-based adaptation. The similar performance of the three training dataset strategies demonstrates that a network trained exclusively on simulated inertial data can generalize effectively to real AUV navigation without requiring real data for training.

Keywords: Data fusion, uncertainty quantification, and predictive modeling
Abstract: Accurate navigation is vital for autonomous underwater vehicles (AUVs). While the fusion of an inertial navigation system and a Doppler velocity log (DVL) within an extended Kalman filter (EKF) is the standard navigation framework, performance severely degrades during DVL measurement outages. In these intervals, unconstrained inertial error propagation leads to rapid drift in velocity and position estimates. To address this challenge, we present an inequality-constrained navigation framework that embeds the physical kinematic limits of the AUV into the state estimation process. During complete DVL outages, the unconstrained EKF state estimate is projected onto the feasible velocity region by solving a quadratic programming optimization problem. When DVL measurements are available, a Kalman gain optimization approach refines the state update while respecting the physical boundaries. The proposed method was validated on real-world sea-trial data collected by an AUV across 11 trajectories. Experimental results demonstrate that applying inequality constraints significantly reduces velocity drift in pure inertial scenarios, achieving a 89% reduction in mean velocity RMSE. Furthermore, during single and dual DVL outage windows, the proposed constrained filter achieves up to a 23.0% improvement in horizontal velocity RMSE compared to the standard error-state EKF.

Keywords: Vehicle navigation, Data fusion, uncertainty quantification, and predictive modeling
Abstract: Accurate state estimation is fundamental to the reliable autonomous navigation of autonomous surface vehicles (ASVs). Sensor fusion of IMU and GPS readings is a common approach for robust localization. However, yaw is fundamentally unobservable in pure inertial navigation because rotation about the gravity axis is indistinguishable from accelerometer measurements. Common workarounds include dual-antenna GNSS baselines, heavy exteroceptive sensors (LiDAR/Radar), or magnetometer-aided IMUs. On small-scale ASVs, magnetometers suffer from electromagnetic interference caused by onboard systems, making heading estimates highly susceptible to steady-state drift. Furthermore, standard Extended Kalman Filter (EKF) sensor fusion suffers from linearization error accumulation under the marine environment's nonlinear dynamics, leading to filter inconsistency.

In this work, the Dead Reckoning in Field Time (DRIFT) framework, originally developed for legged and wheeled robots, is extended through the fusion of a single GPS unit with an onboard IMU. Unlike standard GPS/IMU fusion pipelines that treat heading as externally measured or magnetometer-aided, GPS-DRIFT introduces a left-invariant correction that injects GPS position updates into consistent orientation corrections within the extended special Euclidean matrix Lie group text{SE}_2(3). By continuously correcting yaw drift without relying on magnetometer readings, our solution yields accurate, long-term, and stable heading corrections for marine ASVs. The approach was validated quantitatively in simulation and demonstrated in closed-loop field deployment on a real-world ASV.

Keywords: Real-time decision systems for ocean robotics and autonomous operations, AI/ML for oceanographic and environmental data analysis, Data fusion, uncertainty quantification, and predictive modeling
Abstract: This paper presents the simulation results of a localization framework for maintaining acoustic links between a surface vessel and a mobile autonomous underwater vehicle (AUV). The framework fuses low-cost inertial measurement unit acceleration with modem-side acoustic observations: signal-to- noise ratio (SNR), one-way travel time and Doppler-derived radial velocity. These measurements are integrated through a multi-modal sensing-assisted particle filter. Its process covariance is scaled by real-time SNR so that unreliable acoustic condi-tions expand the particle distribution and strong links produce tighter prediction uncertainty. The resulting AUV position can be transformed into yaw and pitch commands for adaptive surface-modem pointing.

Keywords: Numerical modeling and simulation, Benchmark datasets, reproducibility, and open-source frameworks, Data visualization
Abstract: To strengthen coastal resilience and disaster mitigation workflows, the National Oceanic and Atmospheric Administration (NOAA) is migrating coastal forecast systems from traditional on-premises High-Performance Computing (HPC) platforms to cloud-native architectures. This paper presents the engineering and community aspects, as well as the resulting insights and challenges, of NOAA’s Integrated Ocean Observing System (IOOS) NOS Coastal Modeling Cloud Sandbox. Developed on Amazon Web Services (AWS), the Sandbox platform delivers flexible compute nodes, high-performance data pipelines, and advanced workflow capabilities for cloud-based coastal model development. While building a collaborative workspace is complex, the Sandbox establishes a foundation by providing shared resources and provisional multi-user access for both federal and external partners. User engagement is sustained through regular office hours and a comprehensive learning hub featuring curated community case studies, step-by-step tutorials, and recorded presentations. This paper evaluates the Sandbox platform’s performance through several case studies, including a 12-year regional hydrodynamic hindcast utilizing the LiveOcean ROMS model, and a continental-scale, 43-year Total Water Level retrospective reanalysis driven by the SCHISM component of the National Water Model (NWM v3.0). Notably, the cloud-native SCHISM configuration achieved a 90% reduction in direct computing costs compared to the NCAR Derecho cluster. Finally, we explore how the Sandbox supports a Research-to-Operations-to-Research (R2O2R) cycle. Ultimately, this platform functions as a proof of concept providing valuable lessons and architectural benchmarks, highlighting both the potential of cloud-native community modeling and the remaining requirements to achieve full operational utility.

Keywords: Numerical modeling and simulation, Oceanography: physical, geological, chemical, biological
Abstract: We present a high-order hybridizable discontinuous Galerkin (HDG) finite element solver for the two-dimensional quasi-geostrophic equations in vorticity-streamfunction form. The HDG methodology offers a substantial reduction of the globally coupled degrees of freedom compared with standard discontinuous Galerkin methods while retaining high-order accuracy and geometric flexibility. Spatial convergence is verified using an analytical Taylor-Green vortex benchmark on a geometrically deformed, unstructured mesh. Boundary conditions relevant to coastal flows, including free-slip and no-slip walls, are discussed. Forced-dissipative turbulence simulations reproduce the expected kinetic-energy scaling for two-dimensional turbulence. Finally, simulations in an idealized coastal domain demonstrate the method's ability to represent QG flows in the presence of irregular boundaries. These results establish the proposed HDG solver as an efficient and flexible framework for high-order simulation of quasi-geostrophic mesoscale ocean dynamics with idealized boundary layers in complex domains.

Keywords: Numerical modeling and simulation, Current measurement technology, Autonomous underwater vehicles
Abstract: The real-time estimation of a spatiotemporally evolving underwater contaminant plume is investigated using concentration measurements collected by a single autonomous underwater vehicle (AUV). The synthetic plume field and its real-time reconstruction are computed in three dimensions using a domain-decomposition finite-volume method. A single-sensor method estimates the local spatial gradient of the plume from a sequence of spatiotemporally distributed concentration measurements collected along the AUV trajectory. The method employs a first-order Taylor expansion and is integrated with the AUV guidance and real-time plume reconstruction framework. The AUV trajectory is modified to provide the measurement distribution required for gradient estimation. Numerical simulations are performed for stationary- and moving-source plume scenarios to evaluate the integrated approach. Plume reconstruction results obtained using the estimated gradient for AUV guidance are compared with those obtained using the true gradient. The results show that using the single-sensor gradient estimate has only a minor effect on plume reconstruction accuracy, demonstrating its effective integration with the AUV guidance and estimation framework.

Keywords: Ocean color and hyperspectral measurements, Oceanography: physical, geological, chemical, biological, Marine life and ecosystems
Abstract: Monitoring coastal ecosystems requires long, spatially consistent satellite records of sea surface temperature (SST) and chlorophyll-a concentration (Chl-a), but individual satellite missions differ in duration, resolution, calibration, and retrieval performance. This study developed a Regional Observational Portal System (ROPS) integrating multi-sensor harmonization, regional product generation, validation, visualization, and trend analysis. Satellite observations from SeaWiFS, MERIS, MODIS/Aqua, VIIRS/SNPP, and GCOM-C/SGLI were reprojected to a common 1 km grid and harmonized to the SGLI reference scale using robust regression over sensor-overlap periods. Daily regional products were generated through priority-based spatial merging, producing a 1 km SST record for 2002–2025 and a 1 km Chl-a record for 1998–2025. The products were evaluated against in situ observations and other global merged datasets such as ESA SST-CCI and OC-CCI. The regional merged SST product showed strong agreement with in situ observations (R = 0.99, bias = - 0.42 °C) and retained finer coastal structure than the approximately 5 km SST-CCI product. For Chl-a, the regional merged product reduced log-space bias, whereas OC-CCI showed stronger correlation with observed variability. Whole-scene comparisons showed close agreement between regional and ESA products, with R=0.98 for SST and R=0.88 for Chl-a. ROPS provides web-based data access for original and research products, spatiotemporal visualization, and point- and pixel-window extraction for validation and trend analysis. The results show that regional and global merged products are complementary for coastal monitoring.

Keywords: Hydrography / Seafloor mapping / geodesy, Data fusion, uncertainty quantification, and predictive modeling, Benchmark datasets, reproducibility, and open-source frameworks
Abstract: Quantifying bathymetric interpolation uncertainty remains a key challenge in ocean mapping, with direct implications for navigation, science, and industry. Although modern multibeam systems have improved coverage efficiency, interpolation is still required to fill data gaps, particularly in setline spacing and skunk-striping surveys where unsampled regions remain common. However, the uncertainty associated with these interpolated regions is rarely formally quantified, and existing approaches do not provide a widely adopted, computationally efficient solution for operational use. Like all scientific measurements, bathymetry requires an associated uncertainty estimate for each reported depth. While uncertainty in directly measured depths is well defined through total propagated uncertainty (TPU), interpolated bathymetry must additionally account for likely departures from the interpolated surface between known observations. In this work, we develop spectral and spatial statistical estimators that model these departures as a stochastic process and characterize them using nearby observations under the assumption that seafloor statistical properties are locally consistent over short spatial scales. The validity of this approach is assessed by evaluating the extent to which bathymetry statistics exhibit local second-order stationarity and limited directional dependence. Estimators are evaluated using pass coverage and error metrics that assess safety of navigation and statistical robustness. Testing across both complex and flat seafloor morphologies shows that estimated interpolation uncertainty reliably contains the interpolation error in approximately 90–95% of locations. The spatial statistical estimator outperforms the spectral estimator in complex terrain, while both perform comparably in flat regions. These results enable the propagation of interpolation uncertainty into bathymetric products, providing a practical solution to a long-standing challenge in ocean mapping and improving confidence in datasets for navigation and scientific applications.

Keywords: Marine GIS and data fusion, Systems and observatories, Surveillance, detection, and threat-assessment techniques
Abstract: AIS-based anomaly detection can identify deviations in maritime traffic, but the operational meaning of a detected anomaly may differ substantially across maritime environments. This study applies a common maritime-state anomaly detection framework to the Strait of Hormuz, the Strait of Malacca, the Bashi Channel, and the Bosporus Strait. Hourly maritime-state images representing vessel locations and four common vessel categories were processed using ResNet50 and independently trained Isolation Forest models. Fifteen sustained anomaly episodes were extracted using a common temporal criterion and compared through their frequency, duration, vessel-count direction, vessel-type breadth, spatial change, recovery pattern, and external context. Representative episodes included security-related traffic suppression, repeated anomalies with limited changes in total traffic volume, weather-related traffic concentration in a sparse area, and temporary traffic decline followed by recovery. The results show that a common framework supports comparison at the episode-descriptor level, but not through direct comparison of area-specific anomaly scores. They also demonstrate that anomaly-score magnitude does not necessarily indicate the significance of an external risk. AIS-derived anomalies therefore require context-dependent interpretation using both internal traffic signatures and external evidence.

Keywords: Array signal processing and array design, Sonar signal processing, Fiber-optic distributed sensing (DAS, DTS, DSS)
Abstract: The minimum variance distortionless response (MVDR) beamformer serves as an essential reference method for evaluating novel adaptive beamformers, including reduced-rank methods and learned approaches, the latter of which may require labeled training and validation data. However, full-rank MVDR is computationally prohibitive at large channel counts N in part because the per-frequency-bin sample covariance matrix factorization scales as O(N^3). Another computationally demanding operation is the application of the inverted covariance to dense steering matrices, which are composed of B steering vectors, each associated with a direction of arrival. This scales as O(N^2 B) per frequency bin, where often B > N. Modern high-performance computing (HPC) systems accelerated by graphics processing units (GPUs) are well suited to these operations. However, successful application of GPUs to MVDR processing of large archived datasets requires a distributed approach. We use state-of-the-art HPC libraries to make full-rank MVDR tractable for large-scale processing tasks. We use message-passing interface (MPI) for distributed-memory parallelism, OCCA for portable on-device kernels, and Matrix Algebra on GPU and Multicore Architectures (MAGMA) for per-bin dense linear algebra. Both OCCA and MAGMA expose vendor-neutral interfaces that allow the same source code to target NVIDIA and AMD GPUs. The computational cost of solving the complex-valued systems of linear equations is characterized as a function of the number of simulated channels, beams, and frequency bins. The performance of these operations on the NVIDIA A100 and AMD MI300A GPUs is presented. Parallel scaling results obtained using MPI are demonstrated on up to eight NVIDIA A100 GPUs. The performance-portable architecture introduced in this work can be extended to support numerous other beamforming or direction-of-arrival estimation techniques.

Keywords: Synthetic aperture (active and passive), Sonar signal processing, Array signal processing and array design
Abstract: Synthetic Aperture Sonar (SAS) has been widely used to acquire high-resolution underwater imagery. To achieve the necessary image precision, the accurate estimation of the sonar array platform's motion is important. While motion estimation techniques relying solely on the correlation of sonar data (such as displaced phase center analysis) are common, they frequently suffer from the critical issue of drift, where estimation errors accumulate over time. To mitigate this limitation, micronavigation—which integrates sonar-derived displacements with underwater navigation sensor data—is typically employed. This study focuses on enhancing the accuracy of platform motion estimation by implementing a fusion strategy that applies variable weighting factors to both sonar and navigation datasets. By adjusting the weights based on the reliability of each data source, this research aims to optimize the estimation process and provide a more robust trajectory for SAS signal processing. The proposed approach contributes to the reduction of motion-induced errors, ultimately ensuring higher image quality in complex underwater environments.

Keywords: Array signal processing and array design, Ocean noise, Sound propagation and scattering
Abstract: Passive acoustic monitoring at marine research stations is degraded by vessel traffic and other anthropogenic noise, which can reduce marine mammal call detection by up to 80% near operational sites. We present a low-cost directional active noise cancellation (ANC) system built on a four-element uniform linear array (ULA) of hydrophones interfaced with Teensy 4.1 microcontrollers. Given the bearing of an interfering vessel, the system combines self-interference correction, wideband delay-and-sum beamforming, and normalized least-mean-squares (NLMS) adaptive filtering to estimate the ship signal and drive an anti-noise transducer, while whale direction of arrival (DoA) is estimated from the residual array signal. The system is validated first in air and then in a controlled lake deployment using recorded North Atlantic right whale and vessel signals. Terrestrial testing achieved ship-estimate and whale-preservation signal-to-distortion ratios (SDR) of 18.9 dB and 16.1 dB, respectively. Underwater, across two source-angle configurations, the system achieved ship SDRs of 12.1 and 10.6 dB, whale SDRs of 9.1 and 8.7 dB, whale DoA errors within 3.1°, and a measured cancellation depth of 7.48 dB at the error microphone. These results provide a proof of concept that commercial off-the-shelf embedded hardware can actively cancel directional interference underwater while preserving signal fidelity from a protected direction, motivating future open-water validation with live vessel traffic and unknown source bearings.

Keywords: Oceanographic instrumentation and sensors, Sonar and transducers, Rapid-deployment systems and modular instrumentation
Abstract: Acoustic data measured by a hydrophone with a pure solid encasing tends to suffer from high mechanical vibration and flow-induced noise, especially at low frequencies, because the hydrophone transducer is not adequately isolated from the solid exterior. We investigate a hybrid design that uses castor oil as a boundary layer separating the hydrophone transducer from a solid exterior. Each hydrophone is housed inside a two-stage protective casing: an oil-filled inner shell (3D-printed flexible filament) and a rigid epoxy outer shell. The optimal inner-shell material and geometry are identified through a proof-of-concept in-lab experiment. Sea trial results demonstrate the noise reduction achieved by the hybrid design relative to a solid-casing baseline.

Keywords: Oceanographic instrumentation and sensors, Sonar and transducers, Passive and active acoustic monitoring
Abstract: Modern multichannel data acquisition systems require a balance between channel count and the simplification of cabling, as well as reliable signal delivery between sealed enclosures while minimizing the number of interconnects. This paper presents a multichannel acquisition and signal preconditioning module designed to streamline communication between large-scale sensor arrays and a centralized processing unit. Using serialization and a serial transmission protocol, the module aggregates multiple digital streams into a single high-speed link. Key features of the architecture include: Adaptive Preconditioning: real-time onboard gain adjustment to maximize dynamic range utilization and increase the useful information content of transmitted samples. Flexible Physical Layer: compatibility with standard 50-ohm coaxial cables and category-rated twisted-pair cables. Low Latency: optimized serialization and deserialization (SerDes) paths for latency-sensitive applications. System performance was evaluated through stress testing over long cable runs, confirming resilience to common-mode noise and attenuation effects.

Keywords: Acoustic noise measurement, Room acoustics, Calibration of acoustic systems and metrology, Sonar and transducers
Abstract: This paper describes a method for the experimental determination of the reverberant field transfer function of an acoustic test tank in the frequency domain. Once obtained, the transfer function enables the calibration of transducers and marine recorders in a reverberant field. The method relies on a room acoustics framework for mathematically separating the direct and reverberant fields in a point inside the tank, but employs a transducer-centric model that dispenses with extensive characterization of the tank itself. A calibration experiment in the 1 to 24 kHz band of an omnidirectional SX147 transducer is conducted in the reverberant field, resulting in mean absolute errors of 0.78 and 1.00 dB, respectively for the receive and transmit voltage responses, relative to the free-field calibration. The expanded uncertainty (coverage factor of 2) in the calibration arising from the spatial averaging of the transfer function ranges from 4.63 to 8.57%. The model is extended to include a transfer function that models the effect of an acoustic recorder's body on the field near the hydrophone, presenting a feasible method for obtaining the sensitivity of the recorder. An experiment shows that the free-field sensitivity of the standalone hydrophone can be recovered after compensating the measurements for the tank's and recorder's transfer functions with absolute errors smaller than 0.92 dB. The expanded uncertainty due to spatial averaging of 20 measurements in determining the recorder's transfer function ranges from 6.31 to 13.23%.

Keywords: Soundscape ecology and marine organism acoustic behavior, Passive and active acoustic monitoring, Acoustical oceanography
Abstract: Temporally varying geographic distributions of whale populations are difficult to visually monitor across large marine ecosystems. Acoustic vector sensing (AVS) of the powerful calls produced by baleen whales enables continuous monitoring over large areas with spatial information contained in the bearing estimates to whales. Using three years of AVS data in Monterey Bay National Marine Sanctuary, we examine the habitat occupancy of endangered blue and fin whale populations. The calls of fin whales persistently originated primarily in offshore deep-water (ODW) habitat, shifting into shelf-slope habitat only occasionally. The calls of blue whales originated primarily in shelf-slope habitat. However, detection of blue whales in ODW habitat increased significantly across the study period with stronger wind-driven upwelling that increased krill abundance and expanded foraging habitat. This expansion increased both the spatial overlap of blue and fin whales and the exposure of blue whales to ship strike risk. AVS is an effective method of passive acoustic monitoring for informing marine ecological research and conservation.

Keywords: Machine-learning classification of acoustic signals, Passive and active acoustic monitoring, Systems and observatories
Abstract: Man-made marine noise pollution has had an estimated 32-fold increase of the past 50 years. This has been associated with a drastic increase in whale strandings and general deterioration of marine habitats. Passive Acoustic Monitoring (PAM) systems have been identified as means to protect marine wildlife; however, existing solutions tend not generalise easily and are typically focused on detection only. In this paper, we propose an approach based on deep learning for the combined task of detection, classification and spatial localisation of a broad range of marine mammal species from hydrophone audio data.

Our approach leverages transfer learning by using a pre-trained neural network to extract a semantically rich embedding vector from marine audio. These embedding vectors are used as input to a neural network to detect for the presence of a marine mammal call and classify the species of the source. Localisation of the source is achieved by estimating Time Difference of Arrival (TDOA) using embedding vectors from pairs of hydrophones. The models used by our approach were trained using labelled data from DCLDE and SanctSound datasets. The combined dataset covered a broad range of marine species (fish, dolphins and whales) as well as number of man-made sound sources.

The detection and classification capability was evaluated against a held-out test set whilst the localisation was evaluated using simulated sound emission events for a range of hydrophone array configurations. The detection and classification stages were assessed using precision, recall and F1 scores. The localization stage was assess using localisation distance error.

Our results demonstrate the capability of a deep-learning algorithm for detection, classification and localisation of marine mammals. Our approach can be used to support enforcement of protective measures by integrating with an early warning system or by monitoring protect areas for non-compliant activities or vessels.

Keywords: Passive and active acoustic monitoring, Soundscape ecology and marine organism acoustic behavior, Anthropogenic noise characterization and mitigation
Abstract: Passive acoustic monitoring (PAM) data provide an integrated view of the underwater soundscape and its spatio-temporal variability. These recordings can be interpreted as a compressed representation of multiple physical, anthropogenic, and biological processes, some of which may be linked through cause-and-effect relationships. However, the observed timing, strength, and apparent overlap of acoustic signatures in PAM data may not be a true reflection of actual relationships between drivers and responses in the ocean. Path-dependent acoustic propagation and uncertain source-receiver geometry can shape observed interconnections and make their interpretation challenging. To address this challenge, we developed an event-based approach that combines PAM data, Automatic Identification System (AIS) vessel information, automatic detection of baleen whale calls, and acoustic propagation modeling. The developed approach was tested by assessing the vocal activity of fin and blue whales in the presence of ship-induced noise. The application of the approach showed its ability to identify quantifiable patterns in the vocal activity of fin and blue whales associated with ship passages. Limitations and scalability of the method to other species are also discussed.

Keywords: Passive and active acoustic monitoring, Benchmark datasets, reproducibility, and open-source frameworks, Marine GIS and data fusion
Abstract: Long-duration passive acoustic deployments produce large archives of recordings that are not linked to vessel tracks or encounter structure, leaving range and contact conditions unavailable as variables and requiring manual selection for analysis. To address this limitation, we propose a database-native workflow that aligns hydrophone recordings with Automatic Identification System (AIS) position reports to produce distance-resolved data. Fixed-duration recording windows and AIS messages are stored as persistent geospatial tables and associated through an indexed spatiotemporal join, replacing in-memory nested iteration with a single scalable set-based database process capable of handling continuous, multi-year, million-window archival deployments without exhausting available memory. In this study, the approach processes approximately 9.5x10e5 recording windows and 6.9x10e6 AIS position reports, producing a structured table that separates no-contact, single-contact, and two-contact windows, with the closest point of approach computed directly where applicable and background conditions characterized via deterministic spectral ranking. This formulation enables a GeoAI framework in which spatially indexed, queryable data become directly usable for machine learning. The resulting data product reveals predominantly noise-dominated conditions, with vessel contributions emerging mainly at shorter ranges, indicating that the task lies in extracting structure under background-limited regimes. Spectrogram and quantitative analyses show weak tonal signatures embedded in noise and a consistent decay of signal-to-noise ratio with distance, supporting the use of this representation for scalable machine learning, similarity analysis, and predictive acoustic modelling in real maritime environments.

Keywords: Fiber-optic distributed sensing (DAS, DTS, DSS), Passive and active acoustic monitoring, Bioacoustics
Abstract: Previous Distributed Acoustic Sensing (DAS) studies have successfully detected baleen whale vocalizations < 100 Hz on existing telecommunication infrastructure. While DAS can theoretically detect higher frequencies, its response and detectable bandwidth depend on interrogation settings and cable configuration (e.g., burial depth, coupling medium). Here, we investigate DAS capabilities to monitor baleen whales in the 80– 300 Hz frequency range, including bowhead whales (Balaena mysticetus) and critically endangered North Atlantic right whales (Eubalaena glacialis).

We collected DAS data on telecommunication fibers in Northern Alaska, U.S., and the St. Lawrence Estuary, Canada, with georeferenced playbacks of synthetic signals and vocalizations of target species. Beamformed DAS measurements successfully detected playbacks of North Atlantic right whale up-calls, up to ∼180 Hz in the St. Lawrence Estuary and bowhead whale calls up to 125 Hz in Northern Alaska. We attribute this bandwidth difference to changes in the cable configuration: the cable is buried under ∼5 m of sediments at our Alaska site. Despite reduced performance in buried configurations, this work establishes the foundation for expanding DAS marine mammal monitoring on existing infrastructure beyond the currently demonstrated < 100 Hz range.

Keywords: Oceanography: physical, geological, chemical, biological, Marine geology and geophysics, Hydrography / Seafloor mapping / geodesy
Abstract: Deep seawater temperature and salinity are essential and fundamental parameters data for analyses, such as for understanding the dynamics of deep ocean circulation. However, near-seafloor environments in hadal zones exceeding 6,000 m depth remain data gaps due to sensor pressure limits and operational constraints. In this study, an autonomous underwater vehicle (AUV) observational was conducted in the Japan Trench while maintaining its altitude about 100 m above the seafloor, utilizing the cruising-type AUV "URASHIMA 8000." The vehicle was modified for a maximum operational depth of 8,000 m. Over a cruising distance of more than 130 km at a speed of 3 knots, conductivity-temperature-depth (CTD) and inertial navigation data were collected using onboard sensors. A vertical profile of potential temperature obtained during the dive indicates a vertically homogeneous environment at depths below 6,500 dbar. Further, we confirmed that potential temperature with a reference pressure of 7,000 dbar and salinity are fairly uniform along a terrain-following surface about 100 m above the 7,500 m deep seafloor in the Japan Trench.

Keywords: Current measurement technology, Oceanographic instrumentation and sensors, Systems and observatories
Abstract: Kelvin–Helmholtz (KH) instabilities play a critical role in ocean mixing processes, particularly in regions of strong shear associated with western boundary currents and complex bathymetry. This work presents direct observations of KH billows captured using a multi-instrument mooring array deployed downstream of a seamount in the Kuroshio east of Taiwan. The array included Nortek Signature500 and Signature100 Acoustic Doppler Current Profilers (ADCPs) equipped with broadband echosounder capabilities, complemented by additional environmental sensors distributed across six mooring locations (Fig. 1). The deployment captured persistent shear-driven instabilities characterized by distinct, periodic billow structures visible in both velocity and acoustic backscatter data. High-resolution echograms from the Signature systems reveal multiple trains of KH billows occurring at different depths, with vertical scales on the order of 10–100 m and temporal scales of a few minutes. These observations are consistent with strong vertical velocity gradients induced by flow-topography interaction as the Kuroshio current passes over and downstream of the seamount at the deployment site. The combination of current profiling and co-located acoustic backscatter measurements provides a unique capability to resolve both the dynamic and structural characteristics of these instabilities. These data highlight the evolution, propagation, and layering of KH billows, including simultaneous occurrences at multiple depths, suggesting complex stratification and shear profiles in the water column. This work demonstrates the effectiveness of modern ADCP system with embedded scientific echosounder capabilities for resolving turbulent processes in energetic ocean environments. The results contribute to improved understanding of mixing mechanisms in western boundary current systems and underscore the importance of high-resolution, co-located velocity and acoustic observations for studying submesoscale and turbulent ocean dynamics.

Keywords: Passive observing sensors, Hydrography / Seafloor mapping / geodesy, Oceanographic instrumentation and sensors
Abstract: In March 2026, a custom deep-sea gravity sensor was mounted to AUV Sentry and deployed to a 5600-meter depth, operating continuously at constant near-bottom altitude. This integration presents the first AUV-mounted system capable of routine dense near-source gravity mapping of the deep ocean floor. We discuss the technologies developed which enable this, and describe individual deployments. We describe preliminary sensor findings, and outline more comprehensive analysis and engineering for future work.

Keywords: Air / sea interaction, Vehicle design, Vehicle performance
Abstract: This paper presents the activities, technologies, and experimental results of the Naval Robotic Exercise 2025 (NAVAL-REX25), a operational demonstration focused on the development, integration, and validation of heterogeneous unmanned maritime and aerial systems for ocean monitoring, surveillance, and preservation. Multiple academic and industrial teams participated, contributing autonomous surface vessels, autonomous sailing platforms, unmanned aerial vehicles, smart maritime buoys, and supporting digital infrastructures. The exercise enabled real world testing of key technological vectors, including autonomous guidance and control, cooperative multi vehicle operations, renewable energy propulsion, digital twins, edge and cloud based artificial intelligence, and resilient communications. The results demonstrate the robustness of the deployed architectures, the feasibility of simultaneous multi vehicle mission management, and the effectiveness of integrating sensing, communication, and autonomy in realistic maritime environments. NAVAL-REX25 confirms the relevance of experimental naval exercises as a critical step toward operational readiness of autonomous systems, contributing to enhanced maritime security, environmental protection, and future ocean centric digital twin ecosystems. More importantly, it allows universities, research institutions, and companies to interact with military personnel and with each other in a realistic environment.

Keywords: Ocean energy, Hydrodynamics
Abstract: The floating oscillating surge wave energy converter (FOSWEC) comprises two pitching flaps mounted to a central floating platform. The device has been tested in multiple experimental campaigns at the Oregon State University (OSU) O.H. Hinsdale Wave Research Laboratory (HWRL) Directional Wave Basin (DWB), and a WEC-Sim numerical model has been developed and validated against these data. Building on this foundation, further improvements are planned to demonstrate pathways to reduce levelized cost of energy (LCOE).

WEC-Sim enables high-fidelity evaluation of the existing design, but rapid design iteration benefits from co-design tools that optimize both device parameters and control. WecOptTool uses a pseudo-spectral method to optimize control trajectories subject to system dynamics and constraints; however, modeling complex multibody systems in WecOptTool has historically required manual handling of kinematic transformations and coordinate couplings. The FOSWEC is a three-body system with 18 rigid-body degrees of freedom (DOFs) in WEC-Sim. To streamline efficient model development, we integrate WecOptTool with the Python multibody dynamics toolkit Multibody for Everybody (M4E), which assembles reduced-coordinate equations of motion in joint coordinates from user-defined body and joint locations.

We present a reduced-coordinate FOSWEC model for WecOptTool generated using M4E that reduces the planar model to 5 joint coordinates while retaining kinematic coupling. Code-to-code verification against WEC-Sim shows normalized root-mean-square error of approximately 0.02% for the flap pitch response in the small-amplitude wave case, with errors increasing to about 5% for the largest verification wave condition considered, supporting the model’s use in co-design and optimization studies. Using this framework, we optimize control trajectory across regular wave conditions to evaluate the effect of top-mounted ballast on flap hydrostatic stiffness and electrical power generation.

SNL is managed and operated by NTESS under DOE NNSA contract DE-NA0003525.

Keywords: Wave energy, Dynamics, Control of marine renewable energy systems
Abstract: Oscillating Water Columns (OWC) offer a robust option for ocean wave energy conversion (WEC) due to their minimal underwater moving parts. The OWC devices harness wave energy by trapping air in a chamber, which

converts ocean wave motion into a pulsating airflow that drives an air turbine coupled to an electrical generator. Efficient energy extraction

necessitates a clear understanding of the complex interactions among hydrodynamic, pneumatic, turbine, and electrical subsystems. The power take-off (PTO) mechanism is particularly critical, as it governs the entire energy conversion chain from hydraulic to electrical power. Given the analytical challenges, experimental studies are essential for

model validation, with small-scale experiments offering a cost-effective approach before full-scale prototype deployment. This study decouples the hydrodynamic and pneumatic subsystems for efficient testing. A piston rig, constructed from 6-inch cylindrical pipes and driven by a reciprocating actuator, generates controlled oscillatory pneumatic flow. The downstream end of the cylindrical pipe is fitted with the power generation device or its proxy - miniature Wells turbine, orifice plate or porous plug. Key performance metrics for all the above devices subject to different levels of oscillatory flow are determined and compared. The turbine achieved the highest pneumatic power output (~28 watts) due to elevated chamber pressures, while an orifice plate yielded approximately 9 watts and a porous plug 0.8 watts. Notably, the porous plug exhibited qualitative flow behavior

similar to that of the miniature Wells turbine. These experiments provided valuable insights into the effectiveness of oscillatory pneumatic systems. Such knowledge is critical for developing scaled damping proxies for PTO systems in oscillatory flows, ultimately contributing to the advancement of OWC converters.

Keywords: Energy-harvesting and long-duration monitoring solutions, Ocean energy, Benchmark datasets, reproducibility, and open-source frameworks
Abstract: A pressure-coupled wave-to-wire model is developed for the small oscillating-water-column (OWC) power take-off considered for a spar-buoy oceanographic platform. Analytical linear hydrodynamics, air compressibility, a quasi-steady impulse turbine performance map, shaft dynamics, and a parameterized generator map are assembled in WecOptTool. Periodic, constrained gradient-based optimal solutions define shaft-speed and generator-torque trajectories that maximize mean electrical power. Two selected impulse-turbine and direct-drive-generator configurations are evaluated under the same 0.8 m, 5.5 s sea state. The tested Design A produces 1.05 W, whereas the 90 mm Design B (smaller, better-matched rotor) produces 1.49 W. Design A passes more volumetric flow at lower chamber pressure but develops less mean pneumatic power than Design B. The framework connects experimental turbine data to a coupled wave-to-wire model suitable for future control co-design.

Keywords: Ocean energy
Abstract: Development of a speed controller for a power takeoff (PTO) device equipped with a ball-type continuously variable transmission (B-CVT) for a small marine hydrokinetic (MHK) turbine deployed from a floating unmanned autonomous mobile catamaran platform is described. The unmanned surface vehicle (USV) is a 16ft Wave Adaptive Modular Vehicle (WAM-V) equipped with a flight deck for launch, recovery, and recharging of aerial drones, and with self-anchoring capabilities [1]. The MHK turbine consists of a free-stream waterwheel (FSWW) with 7, 9 or 11 curved blades. The PTO consists of an input shaft driven by the FSWW to a 35:1 gearbox, a ball continuously variable transmission (B-CVT) and a generator. The B-CVT ratio ranges from 0.5:1 to 1.9:1 and is controlled via a servo motor [2]. The turbine is a partially submerged multi-blade free stream waterwheel (FSWW) [3]. The derivation of an analytical torque model for the MHK turbine is detailed and utilized for initial implementation and tuning of the MHK speed controller via bench tests. The speed controller is an error-based speed setpoint controller with integral gain due to the low rate of change of fluid flow. Field tests were carried out on Florida’s Intracoastal Waterway, where the MHK turbine onboard the USV is field tested, and the turbine speed controller’s initial effectiveness validated. The controller is subsequently upgraded to account for a range of varying flow speeds, utilizing three distinctive control stages to maintain control of the turbine speed, leveraging both the B-CVT’s ratio and charge controller’s electrical current demand. The system being developed is in support of potential self-powered autonomous mobile recharge stations for unmanned aerial vehicles (UAVs) operating in coastal zones [3]. Further field tests of the MHK platform were carried out and successfully demonstrated the validation of the upgraded MHK speed controller. The relationship between flow speeds and B-CVT ratios is derived from field test data. The overall optimization of the MHK system via speed control around its optimal operating speed setpoint was successfully demonstrated.

Keywords: Ocean energy, Energy-harvesting and long-duration monitoring solutions
Abstract: Marine energy, encompassing power harvested from ocean waves, tides, currents, and thermal or salinity gradients, represents an increasingly viable source of energy. Among these, tidal energy offers a predictable and reliable resource, with locations selected based on factors such as peak and average tidal inflow velocities, bathymetric conditions, and grid access. Solutions for small-scale tidal applications are in need of new approaches because commercial-scale technologies are optimized for higher tidal velocities and exhibit reduced efficiencies at moderate inflow tidal speeds, often requiring higher cut-in speeds which may exclude the potential for energy generation in many locations with moderate tidal inflow speeds. This paper explores the use of electric bicycle power electronics as a cost-effective and efficient alternative to traditional power takeoff systems for small-scale tidal turbines. Simulations built using MATLAB-Simulink modeled various power takeoff systems, including the UW Lander and OSTEC turbines, alongside electric bicycle-inspired solutions with static and variable gear ratio gearboxes. Simulation results indicate the benefits of e-bike power electronics, particularly their reduced hotel loads and compatibility with moderate tidal inflow velocities, unlocking opportunities for low-power tidal technologies. These findings provide a foundation for advancing small-scale tidal applications, with implications for sustainable energy harvesting and cost reductions in marine energy systems.

Keywords: Cybersecurity for marine robotic and sensing systems, Security and resilience of AUV/USV/ROV platforms, Maritime domain awareness and autonomous monitoring
Abstract: A Common Approach to Maritime Cybersecurity Optimizing cybersecurity investments for afloat platforms requires standardizing security measures, prioritizing threats, and improving modeling & simulation tools. Standardized protocols give fleets a consistent defensive baseline, allowing lessons learned to be applied enterprise-wide, including commercial supply vessels, which share ports and communications infrastructure with naval platforms but have far fewer resources. Prioritization directs resources at the greatest risks. Strengthening authentication limits access to authorized users, reducing the attack surface. Complementary investments in real-time monitoring, commercial security mechanisms, and risk management further reinforce posture. Context-aware decision-making recognizing shifting threat priorities between routine transit and contested engagements tie these into an adaptive strategy. Prospective Defense Through OT and IT Functional Domains Cyber defense spans Operational Technology (OT) and Information Technology (IT). OT defense centers on industrial control systems governing propulsion, power, and damage control, requiring robust authentication, anomaly detection, and secure communications. IT defense focuses on encryption, network segmentation, and least-privilege access. Autonomous systems reinforce both domains. Emerging Technologies and the Evolving Threat Landscape Adversarial AI conducts reconnaissance and executes multi-stage intrusions at speeds beyond conventional defenses, driving urgent need for automated countermeasures. Electromagnetic attack can sever communications and falsify GPS signals, requiring multi-source navigation validation. Autonomous drone swarms offer near-term protection for merchant vessels lacking dedicated systems. Strategic Alignment These priorities mirror the Trump Administration's National Cyber Strategy, which calls for hardening critical infrastructure, deploying AI-enabled cyber tools, and securing OT supply chains. Maritime platforms are exactly the infrastructure the strategy targets for prioritized defense. A layered, adaptive architecture integrating technical controls, human expertise, and organizational procedures remains essential to resilience.

Keywords: Security and resilience of AUV/USV/ROV platforms, Cybersecurity for marine robotic and sensing systems
Abstract: Unmanned Surface Vessels (USVs) are increasingly adopted in the maritime domain. There are risks of cyber-attacks through communication networks. There also exist threats of cyber-physical attacks, where the vessel is hampered and sensors blinded by adversaries. Current maritime cybersecurity heavily focuses on IoT and network defense. There is a lack of research on autonomous, non-lethal cyber-physical defense against physical vandalism. This paper introduces a framework to use runtime monitoring to detect physical tampering and trigger mechatronic defenses. Coupling ship dynamics with ship performance, this could form a baseline model based on design performance. This paper proposes to use anomalies in the propulsion system and kinematic state to draw inferences of physical vandalism and tampering. Anomalies in the propulsion system e.g. mismatches in thrust-to-RPM ratio suggests a fouled propeller or an entangled net. These effects cause drag which alters the expected propulsion output and causes the vessel to be reducing speed. Anomalies in the kinematic state could suggest environmental interference e.g. heavy seas altering pitch and roll or physical interference e.g. unauthorized boarding. When the runtime monitoring system assesses deviations beyond environmental influence and mechanical failure, the system would classify these as deliberate tampering and trigger zero-trust network lockdown i.e. erasing encryption keys and sensitive information. The USV would cease communication with the control system ensuring integrity of the overall network. Importantly, the vessel should be able to physically protect itself. The USV could decouple the drivetrain to prevent motor burn-out. The steering gear could be locked at a hard angle deterring efficient tow. Any external data ports could be physically shuttered to prevent actors from plugging into the local network hence protecting information security and preventing localized control. This paper aims to establish thresholds for anomalies to differentiate between environmental influences and adversarial actions which would trigger cyber-physical and mechatronic responses to secure the USV and the overall network.

Keywords: Security and resilience of AUV/USV/ROV platforms, Autonomous underwater vehicles, Vehicle performance
Abstract: Scenario-based validation is increasingly recognized as essential for assuring Maritime Autonomous Surface Ships, yet existing maritime approaches remain fragmented across hazard analysis, rule-based synthesis, and data-driven scenario extraction, without a unified theory of what constitutes the underlying scenario space. This paper introduces the Maritime Scenario Space Theory (MSST), a domain-specific and interaction-centric ontology that formalizes the Maritime Scenario Space (MSS) as a six-dimensional semantic coordinate system for maritime traffic situations. MSS factorizes scenario semantics along the axes of geometric encounter structure, rule applicability, kinematic feasibility, environmental modulation, behavior policy, and perception and communication. Building on this representation, the paper demonstrates how MSST enables axis-structured characterization and comparison of scenario sets and generation methods, supports comparable criticality reasoning across heterogeneous scenario sources, and provides a structural account of why automotive scenario ontologies are not fully applicable to maritime validation. Overall, MSST provides a principled semantic foundation for transparent coverage reasoning, criticality assessment, and future standardization of scenario-based assurance for autonomous maritime systems.

Keywords: Marine law and policy, Trust, ethics, and transparency in marine AI systems, Real-time decision systems for ocean robotics and autonomous operations
Abstract: Maritime operations are governed by a broad body of regulation. Two of these domains are the concern of this paper: navigation rules that determine how vessels avoid collision, and Rules of Engagement (ROE) that determine when and how force may be employed. Both are issued as natural-language directives. This paper reports two formalization efforts conducted at the Naval Postgraduate School using the Suggested Upper Merged Ontology (SUMO) and the Standard Upper Ontology Knowledge Interchange Format (SUO-KIF), and shows that the two domains admit a common formal treatment. A previously validated formalization of the International Regulations for Preventing Collisions at Sea (COLREGs) provides the foundation; we extend that approach into ROE, where authorization depends on command authority, mission context, force posture, and the legal classification of a contact. The resulting ROE ontology derives authorization through ontological subsumption over defined action classes rather than through lookup against enumerated measures. Eleven theorem-proving queries, seven expecting entailment and four expecting non-entailment, evaluate whether subsumption, authorization, prohibition, permissibility, the restrictive default, and closure consistency behave as specified. Nine returned the expected result, while two negative queries remained inconclusive against the full knowledge base under the applied resource bound. Each proved determination is accompanied by an inspectable derivation from input facts through applied rules to conclusion.

Keywords: Protection of offshore energy and critical maritime infrastructure, Autonomous underwater vehicles, Vehicle design
Abstract: This paper presents an underwater manipulator designed for non-destructive grappling of conventional AUVs, Lagrangian profilers and gliders. The proposed system is intended to function as a subsystem integrated into an underwater interceptor vehicle. Manipulation is achieved by two or more arms, each composed of multiple rigid links connected through hinge joints. A custom hydrophilic passive suction cup, inspired by the adhesion mechanism of clingfish, is mounted beneath each link via a compliant spherical joint which enables a novel adaptive surface contact when interacting with curved, irregular surfaces such as those of torpedo-shaped bodies. Furthermore, the mechanism can grapple the target AUV from a finite standoff distance without prior physical contact or parallel alignment, accommodating a range of relative positions and orientations. To evaluate the design, computational fluid dynamics (CFD) is used to assess the hydrodynamic behavior of the system. A prototype of the proposed manipulator is fabricated and deployed to experimentally validate the design concept.

Keywords: Vehicle design, Marine GIS and data fusion
Abstract: Abstract

This paper presents the design and development of a standalone, ROS-based topside software system for command and control of the National Deep Submergence Facility’s (NDSF) Liquid Robotics Waveglider surface vehicle. The Waveglider serves as a critical communication relay between ship-based operators and subsea vehicles such as the Sentry AUV, bridging satellite communications on the surface with acoustic communications underwater via WHOI Micro-Modem. This relay capability enables the research vessel to maneuver independently, eliminating the need to remain within direct acoustic range any subsea vehicle. The Waveglider runs on solar power and, under operator guidance, either station-keeps or tracks the subsea vehicle's position. Previously, operating the Waveglider topside required a fragmented collection of scripts, manual port configurations, and tight coupling to the Sentry topside computing environment. This new software effort consolidates Waveglider control functionality into a unified, portable system that can be launched with a single desktop shortcut, significantly reducing operator workload and configuration complexity during at-sea operations.

Keywords: Vehicle design, Vehicle performance, Remotely operated vehicles
Abstract: Marine robotic systems are routinely assembled from components never designed to share a system context. Control frameworks including ArduSub, ROS, and ROS2 were developed for aerial and terrestrial applications and impose structural liability in subsea deployments. Each ROS2 distribution is pinned to a specific Ubuntu LTS release, with compatibility determined by core system library versions that differ across releases. A sensor developer on ROS2 Humble (Ubuntu 22.04) and a vehicle integrator on ROS2 Jazzy (Ubuntu 24.04) are on incompatible stacks, components may function independently but will not integrate without overhead that delivers no new capability. In multi-vendor marine development, this is a structural source of integration failure, not a theoretical one.

Shared power domains with no fault isolation allow a single shorted peripheral to collapse compute, control, and payload subsystems simultaneously. Absent time-synchronized logging forces post-hoc correlation that slows diagnostics and delays deliverables.

This paper presents NadirOS and Mission Dock, a unified marine vehicle operating architecture built natively on RTI Connext DDS, eliminating OS version dependencies entirely. A marine-specific Error-State Kalman Filter delivers commercial-grade navigation from a standard AHRS, removing the cost and complexity of a full INS. Mission Dock consolidates fault-isolated power distribution, per-channel telemetry, and integrated NVIDIA Jetson Orin edge compute onto a single validated board.

The architecture has been validated across 24/7 over-the-horizon remote operations, multi-kilometer pipeline inspections, aquaculture intervention, and offshore surveys completed eight days ahead of schedule across more than six distinct vehicle configurations on a single software stack.

Keywords: Open-source hardware, software, and community innovation
Abstract: Underwater vehicle programs at WHOI have historically maintained independent software stacks with a lot of duplicated effort. With ROS1 Noetic reaching end-of-life, that model became unsustainable. WHOI-ROS is a publicly available, plugin-based ROS2 ecosystem co-developed by the Oceanographic Systems Laboratory, the Deep Submergence Laboratory, and the Acoustic Communications Group to consolidate the previously independent codebases into a shared foundation for underwater vehicles.

WHOI-ROS is organized in three tiers. Foundational libraries provide shared message definitions, transport abstractions, and a common asynchronous sensor-driver framework. Drivers inherit these foundations so per-sensor packages primarily implement wire-protocol parsing and mapping to ROS messages. Vehicle-specific hardware interfaces are loaded as plugins, allowing new vehicles to be integrated through thin interface layers without modifying shared packages.

A central architectural challenge has been defining the boundary between reusable and vehicle-specific software across very different platforms, including small hybrid ROVs, full-ocean-depth AUVs, and human-occupied deep submergence vehicles. WHOI-ROS addresses this by concentrating shared logic in communication, sensing, and support infrastructure while preserving platform-specific behavior behind explicit interfaces. Acoustic communications are integrated through ros_acomms, extended with a ROS1/ROS2 interoperability shim.

WHOI-ROS is already deployed across three vehicle classes and is being introduced into operational vehicles through a companion migration effort. Twenty sensor drivers are currently shared across two or more vehicles, covering core deep-submergence instruments including depth, DVL, INS, magnetometer, sound velocity, and GPS.

For the OCEANS community, WHOI-ROS demonstrates a practical model for consolidating marine robotics software across laboratories and vehicle classes while reducing duplicated development, accelerating integration of new instrumentation, and establishing a shared open foundation for future underwater vehicle systems.

Keywords: Low-cost, scalable oceanographic sensors and platforms, Vehicle design, Open-source hardware, software, and community innovation
Abstract: This paper presents an open-source hardware and software architecture that extends the capabilities of the commercial Blue Robotics BlueBoat autonomous surface vehicle (ASV) for adaptive marine monitoring applications. The proposed framework integrates a distributed computing architecture based on a frontseat/backseat paradigm, enabling the deployment of ROS2-based autonomy while maintaining compatibility with the native BlueOS and ArduPilot ecosystems. The platform incorporates heterogeneous communication technologies, including Wi-Fi, 4G/LTE, LoRaWAN, and underwater acoustic communications, together with support for multi-modal sensing payloads. These developments transform the BlueBoat into a flexible and cost-effective research platform capable of adapting its behaviour based on in situ observations. Initial field experiments validate the integration of the hardware, software, and communication subsystems, demonstrating the suitability of the proposed architecture for ocean observation, biodiversity monitoring, and other marine robotic applications.

Keywords: Remotely operated vehicles, Vehicle design, Autonomous underwater vehicles
Abstract: This work presents the design, integration, and field evaluation of a modular sediment- and water-sampling retrofit for an observation-class remotely operated vehicle (ROV) intended to support per- and polyfluoroalkyl substances (PFAS) studies. The design is intended for sediment and water sampling in turbid, near-bottom environments with limited camera visibility, while reducing the need for active downward thrust during sampling. The system integrates a passive attitude-biasing mechanism with a gripper-based sediment collector and a removable-bottle water sampler. The water sampler uses inlet and vent/displacement tubing that are simultaneously closed by a waterproof, servo-actuated dual-tube pinch valve to isolate the sample during recovery. A quick-release bottle holder supports rapid bottle removal and exchange after deployment. Field tests confirm the system’s capability for effective sediment collection, water sampling, and bottle exchange. Analytical comparison of both manual and robotic samples shows that the integrated design can support ROV-based water and sediment sampling for PFAS-focused environmental field studies.

Keywords: Autonomous underwater vehicles, Vehicle design, Vehicle performance
Abstract: Heterogeneous uncrewed surface–subsea systems pairing a USV with one or more AUVs must coordinate command, state, and status information over communication links with different latency, availability, and failure characteristics. Although an Interface Control Document (ICD) typically specifies message contents and update rates, a loosely coupled USV–AUV system (UAS)—comprising a Mission Control System (MCS), Launch and Recovery System (LARS), two power-controlled AUVs (PCAs), and shore-side remote-control stations—creates operational dependencies that a document-only ICD cannot fully capture, especially during safety-critical launch and recovery. This paper presents an ICD-based verification framework with three layers: (1) an interface-specification layer defining nodes, topics, and command–acknowledgment contracts; (2) an operational-semantics layer defining valid state transitions and the launch–dock–recovery sequence; and (3) a verification layer that derives topic- and message-level key performance indicators, cross-node consistency, command–acknowledgment completion, and subsystem health from ROS2 bag recordings. The framework was evaluated using 25 datasets collected during pier-side and nearshore trials of a 12 m-class USV, a 2,500 m-rated PCA, and a shore-based remote-control station over five days in November–December 2025. Since the datasets include USV transit, stationary LARS checks, and PCA docking/undocking, evaluation combines corpus-level topic/node coverage with event-level PCA–LARS docking-cycle observability for 14 docking-related sessions. Onboard interfaces were broadly observable; four sessions exercised a complete docking cycle, and one of six reconstructed cycles fully conformed to the ICD-defined sequence. The results demonstrate the feasibility of ICD-to-verification using operational logs and identify practical refinements for future integrated sea trials.

Keywords: Vehicle navigation, Autonomous underwater vehicles, Numerical modeling and simulation
Abstract: This paper presents a unified guidance law for a class of marine environmentally-propelled vehicles that includes sailboats and Lagrangian profiling floats, which navigate exclusively by harnessing disturbances such as wind or layered ocean currents. Since propulsion is dictated by the environment, the guidance problem reduces to the selection, at each decision instant, of a particular course from a finite set of environmentally admissible directions such that the vehicle progresses toward the goal, producing the characteristic zig-zag maneuvers well-known to sailors. The selection mechanism is event-driven and finite-state-machine-like, and is modeled as a hybrid dynamical system so that the continuous plant dynamics can be integrated easily. Within this framework, uniform global pre-asymptotic stability of the waypoint capture region is established via a Lyapunov function that strictly decreases during flow and is invariant across jumps, with finite-time capture of the closed set around the target waypoint. The guidance law is solved on a purely kinematic level and intended to be agnostic to the detailed platform-specific dynamics. Simulations of a sailboat under time-varying wind and a profiling float exploiting depth-dependent current layers illustrate the potential of the approach.

Keywords: Vehicle navigation, Remotely operated vehicles, Autonomous underwater vehicles
Abstract: Model-based controllers can provide superior Marine Surface Vehicle (MSV) trajectory-tracking performance when plant and actuator parameters are precisely known, but their performance may degrade when these parameters are unknown a priori or change with mission configuration, environmental conditions, or component degradation. This paper reports a nullspace model-based Surface Adaptive Trajectory Tracking Controller (S-ATTC) for an MSV represented by a nonlinear, second-order 3-degree of freedom (DOF) model. The proposed S-ATTC is designed to provide asymptotically stable position and heading trajectory tracking while adaptively identifying plant and actuator parameters online, without requiring acceleration measurements. We report a numerical simulation of an overactuated four-thruster MSV with asymmetric propeller thrust and compare the trajectory tracking performance of the proposed S-ATTC to its nonadaptive Surface Trajectory Tracking Controller (S-TTC) counterpart, and a proportional-integral-derivative (PID) controller. Despite substantially inaccurate initial parameter estimates, the numerical simulation demonstrates asymptotically exact pose and velocity trajectory tracking and convergence of the estimated

Keywords: Autonomous underwater vehicles, Remotely operated vehicles
Abstract: This paper reports a preliminary simulation evaluation of an adaptive approach for experimentally estimating both the plant model parameters and the actuator model parameters for the dynamical model of an underactuated underwater vehicle (UV) that operates both fully submerged and surfaced, i.e. in contact with the free surface. This approach employs a hybrid dynamical plant model which (i) exactly represents the conventional dynamical plant model when the vehicle is fully submerged and (ii) approximates the buoyancy and gravitational effects of a low-speed UV when interacting with the free surface. We report a simulation study of the performance of an adaptive parameter identification algorithm for identifying both plant and actuator model parameters, comparing the performance of the conventional fully-submerged plant model and the hybrid plant model. This is the first reported simulation study of adaptive parameter identification using the hybrid model approach.

Keywords: Autonomous underwater vehicles, Dynamic positioning, Vehicle performance
Abstract: Fujitsu aims to realize an “Ocean Digital Twin” that recreates the state of the ocean in a digital space for the purpose of global environmental conservation and biodiversity conservation. As a first step, we are focusing on utilizing seaweed beds for blue carbon credits. We are developing automated measurement of seaweed beds using underwater drones and seaweed recognition using image processing. Underwater drones are greatly affected by disturbances from ocean currents and waves, making precise position control and accurate path tracking difficult. In this report, we formulate adaptive ocean current estimation for underwater drones using Lyapunov functions, conduct experiments on disturbance reduction control using proposed method in both calm-water and real-sea environments. We also compare the proposed method with a disturbance observer.

Keywords: Vehicle navigation, Vehicle performance, Dynamic positioning
Abstract: As a catamaran type Unmanned Surface Vehicle (USV), ARAGON CAT40 has been developed under the financial support of KOREA Ministry of Science and ICT. ARAGON CAT40 is the carrier of two autonomous underwater vehicles (AUVs), which can dive up to 2,500 meters for the sea bottom survey. ARAGON CAT40 is also the mothership in which two AUVs can be docked and retrieved for the battery charging. It has the length of 11 meters and the displacement of about 10 tons with electric twin motors and sail-drives. AUV is about 5 meters and it has the weight of about 500kg. For the launch and recovery of AUV, it is necessary of USV to navigate to the position of docking with AUV and keep the position in wave. For this, the autonomous navigation system of USV ARAGON CAT40 has been developed with the path following and dynamic positioning algorithms. The path following algorithm is implemented to control the reduction of speed considering the significant wave height and direction for avoiding the risk of wave. The dynamic positioning algorithm is implemented to control the four actuators of two bow thrusters and two stern thrusters as a type of sail-drive for the compensation of the position error and velocity error of surge, sway and yaw. In this study, the criteria of path following is the track-off distance under 3L and the criteria of dynamic positioning is the position error under 3meters. Now, we would like to evaluate the performance of path following and dynamic positioning in the autonomous navigation system of USV ARAGON CAT40 in the actual sea area. Actual sea tests are carried out in order to ensure the robustness of autonomous navigation of USV during the launch and recovery of AUV. Scenarios are tested on the launch and recovery of AUV in USV such as the path following of USV, the dynamic positioning at the docking position, and the launch and recovery of AUV in USV. All of scenarios are tested in the actual sea area with wave. In this paper, the main features of USV ARAGON CAT40 and main results of field tests on autonomous navigation of USV for the launch and recovery of AUV are described.

Keywords: Vehicle performance, Numerical modeling and simulation, Hydrodynamics
Abstract: This paper presents a physically constrained adaptive-dynamics framework for identifying three-degree-of-freedom unmanned surface vehicle (USV) maneuvering models from field data. The model retains the standard marine craft equations for surge, sway, and yaw. These are rearranged into an inverse-dynamics affine model whose 19-dimensional parameter basis enter linearly, while velocity and acceleration states define nonlinear regressor entries. This formulation enables constrained moving-horizon parameter updates with symmetric, positive-definite inertia, and nonpositive-damping constraints. Field-log results from a Blue Robotics BlueBoat differential-thruster surface vehicle show that the method can estimate a coupled model from Global Positioning System (GPS), inertial measurement unit (IMU), and actuator data and use the identified coefficients for dynamic simulation and forward validation. The framework is intended to support field model calibration, adaptive control, uncertainty-aware simulation, and digital-twin construction for small autonomous surface vehicles.

Keywords: AI/ML for oceanographic and environmental data analysis, Data visualization, Access, custody, and retrieval of data
Abstract: Ocean data is distributed across agency archives, project portals, local files, and live observing systems. Mature services make large holdings discoverable or computable, but project teams must still reconcile formats, specialized retrieval systems (such as APIs), spatial and temporal references and resolutions, access rules, provenance, and decision products across multiple systems. This paper presents the logical architecture and a pilot implementation of the Big Blue Cloud, a cloud-based integration workspace for heterogeneous large environmental data. The architecture separates intake and metadata registration, storage and curation, analytics and derived products, visualization, and dissemination, with cross-cutting access control and versioning while integrating AI to expedite some of these processes. The platform is positioned relative to NASA Earthdata Search, ERDDAP, Google Earth Engine, and the Copernicus Data Store rather than attempting to replace them. In a Ross Sea Region Marine Protected Area pilot, the workflow combines multiple data sources, such as CCAMLR zone boundaries, OSCAR surface currents, GFS winds, and NASA PACE ocean-color products, in a shared common operating picture. The paper also presents the necessity of openness and interoperability of a platform such as this: source datasets should retain their original licenses, public data and derived products can be exported in reusable formats, and project-restricted data must remain governed. The contribution is therefore a configurable integration and decision-support layer, together with a commercial stewardship model designed to sustain maintenance without requiring exclusive custody of users' data.

Keywords: Data fusion, uncertainty quantification, and predictive modeling, Marine GIS and data fusion, AI/ML for oceanographic and environmental data analysis
Abstract: As marine ecosystems face increasing stressors from climate change and anthropogenic activity, the ability to synthesize disparate oceanic datasets remains a critical challenge. Traditional data silos and inconsistent analysis workflows often hinder comprehensive situational awareness, preventing stakeholders from making rapid, data-driven decisions for coastal resilience and resource management. Through a NOAA and Esri collaboration, the Ocean and Coastal Information System (OCIS) presents an integrated framework designed to harmonize multi-dimensional marine data to improve the delivery of actionable knowledge. Central to this initiative is the implementation of a standardized spatial indexing system utilizing a hexagon-based analysis strategy, specifically leveraging the Uber H3 hexagon system. This advanced framework offers multiple spatial scales, enabling users to analyze and visualize data at varying levels of detail—whether at the global, national, regional, local, or even hyper-local level—providing the flexibility to tailor insights to the needs of different applications and stakeholders. Unlike traditional rectangular grids, the hexagon tessellation provides more uniform neighbor connectivity and reduces sampling bias, offering a superior scalable geometry for integrating spatial data. We leveraged this strategy to enrich the hexagons with a variety of data from different disciplines, including: • predictive oceanographic climate models • biodiversity data • bathymetric surveys • socio-cultural indicators • benthic habitats These aggregations resulted in characterizing each discrete ocean hexagon with over 100 spatially explicit variables. By synthesizing these diverse inputs, OCIS delivers multi-theme knowledge through a single, intuitive application, moving beyond fragmented maps to provide a holistic view of the maritime domain. Analysis results are accessed through maps with dynamic visualizations and reported via plain-language interfaces, providing users with easily interpretable and actionable information. By refining metadata and optimizing data assets for success in AI-readability, OCIS establishes a new standard in best practices that ensure machine-learning accuracy and efficiency.

Keywords: Data visualization, Marine GIS and data fusion, Access, custody, and retrieval of data
Abstract: Through collaboration with scientists and multiple organizations, MIT Sea Grant GIS+VIZ (MITSG GIS+VIZ) has developed the Seaglass geospatial data analysis and visualization system.

Seaglass provides a comprehensive, mobile-friendly environment for data management, visualization, analysis, modeling, and download. It utilizes a dynamic and flexible ensemble of data sources, including its own internal database to which users can upload their data, that responds to the needs of the user. External data source APIs include both Eulerian platforms and Lagrangian platforms, including buoys, drifters, ship tracks, fixed sensors, as well as mapserver layers and raster data from satellites. Data may be visualized and analyzed via the interface form and/or map. In addition to visualizing data, the system also supports index calculation, modeling visualization, using either the host server or external, high-performance servers. Calculated input can be visualized in the system real-time. All visualized data and modeled output can be downloaded from the system on-the-fly as tabular or graphical files.

The Seaglass system has developed through three unique uses cases, water quality management with the Town of Nantucket, habitat assessment with NOAA Fisheries, and shellfishing growing area management with several Massachusetts towns and the MA Division of Marine fisheries, and more specific applications with the Massachusetts Water Resources Authority and Stellwagen Bank National Marine Sanctuary. In all cases, we collaborated with staff, funded researchers, outreach professionals, and artists to identify the most effective features, utilized popular, open-source web technology, and flexible, scalable storage, and ensured cross-platform / cross-device compatibility and customizability.

Keywords: Maritime domain awareness and autonomous monitoring, Information management, Access, custody, and retrieval of data
Abstract: In recent years, Maritime Domain Awareness (MDA) has become essential for navigational safety, economic security, and efficient maritime operations, and there is a growing demand for communications infrastructure that can continuously collect shipborne data and utilize it onshore. The VHF Data Exchange System (VDES), a next-generation AIS, provides bidirectional links via shore stations and satellites and is expected to form a wide-area maritime communications backbone once carriage becomes mandatory under the SOLAS Convention. However, most studies address physical-layer schemes or isolated algorithms and do not assess the end-to-end (E2E) operational feasibility of integrated systems from ship terminals to data platforms. This paper presents the system architecture of the VDES Satellite Constellation System for Oceanographic Information Sharing (VOIS) and proposes a system-level evaluation framework for its operational feasibility. VOIS comprises satellites, shore stations, ship terminals, and a data platform that together form a VDES-based maritime information-sharing system for wide-area data collection and information distribution. Rather than evaluating communication and data processing separately, the proposed framework enables quantitative assessment of system-level E2E performance. In this initial phase, we focus on the maritime/terrestrial segment and conduct subsystem integration tests with ship terminals, shore stations, and a data platform. For the evaluation of E2E operational feasibility, we define three metrics: E2E delivery ratio, E2E latency, and measurement-point delivery ratio, and place measurement points along the data path. In a terrestrial communications environment, we confirm E2E connectivity and evaluate system performance based on these metrics. By observing behavior under steady-state and degraded conditions, we evaluate the fault tolerance of VOIS. This study thereby contributes to the development of future maritime data communications infrastructure, including satellite segments, by providing a practical architecture and evaluation framework for VDES-based maritime information-sharing systems.

Keywords: Hydrography / Seafloor mapping / geodesy, Marine GIS and data fusion
Abstract: BathyGlobe GapFiller is a software tool that helps hydrographers plan transits and surveys while optimizing coverage of unmapped areas. It was developed to support the Seabed 2030 project, the goal of which is to completely map the world's seafloor by 2030. GapFiller 2.0 is a new web-based version that improves accessibility and ease of use compared to the original. This version includes new implementations of two planning tools: one for modifying long transits to opportunistically map gaps along the way, and one for planning surveys of polygonal regions. These improvements increase the utility of GapFiller as a planning tool for Seabed 2030.

Keywords: Low-cost, scalable oceanographic sensors and platforms, Citizen-science networks and participatory ocean observations, Oceanographic instrumentation and sensors
Abstract: Water monitoring observations play an important role in benchmarking accuracy of inundation forecasting systems in a variety of ways. As the length of record extends for each observing medium, the confidence in that data set for present validation purposes increases. This study examines the use of a combination of automated water level sensors operated by federal, state, and local (StormSense) governments and organizations in Coastal Virginia. Water level anomalies recorded by these sensors over the past decade (2016-2026) during the highest astronomical tides of the year (king tides) were compared across key river transects in major tributaries leading to Chesapeake Bay and out to the Atlantic Ocean (Figure 1). These time series anomalies were used to heuristically construct a record of flooding impacts based upon compounding wind speeds and directions affecting the amplitude of the king tide on opposing banks of major estuaries (Figure 2). These automated sensor observations were then validated by hydrodynamic models (Tidewatch Map/SCHISM), community science collected GPS maximum flooding extent contours (Catch the King), and aerial surveys (drones and Planet satellite data).

Catch the King is a participatory tidal flood mapping program in the Tidewater region of Virginia that mobilizes local volunteers to map the peak extent of inundation during annual king tide events. Its core purpose is to generate empirical observations that support calibration and validation of hydrodynamic forecast tools, particularly the Virginia Institute of Marine Science’s (VIMS) Tidewatch Map. The Tidewatch Map produces 36-hour storm tide forecasts using the open-source SCHISM model, operated by the Center for Coastal Resources Management (CCRM), with twice-daily runs driven by National Weather Service atmospheric inputs. Since its inception in 2017, Catch the King has collected over 250,000 GPS reported high water marks, and has functionally served as a public annual tidal calibration of the SCHISM hydrodynamic model underpinning the Tidewatch Map’s automated GIS inundation mapping framework. The Tidewatch Map has validated near-field predictions of the Tidewatch Map annually, and has regularly aided in improving the model.

Keywords: Fiber-optic distributed sensing (DAS, DTS, DSS), Sound propagation and scattering, Large-aperture and persistent seabed sensing arrays
Abstract: Distributed acoustic sensing (DAS) is a powerful modality of active and passive acoustic sensing of the ocean. It has been also successfully applied to measure near-bottom temperature variations caused by internal waves, mesoscale eddies, and upslope bores. Building on the existing seafloor infrastructure, DAS can provide real-time ocean measurements with a combination of spatial and temporal resolution that is unprecedented for subsurface measurements in the ocean. Taking the full advantage of DAS capabilities and transitioning the DAS technology from an exploration phase to operational use in the ocean is impossible without knowledge of the transfer functions from the DAS measurand - the time-resolved variation of phase of the Rayleigh-scattered coherent light propagating in an optical fiber - and the physical quantities of interest, such as acoustic pressure, temperature variations, and seabed pressure perturbations due to surface gravity waves or hydrodynamic flows. Currently employed legacy models of pressure and temperature sensitivity of optical fibers imply a local relation between variations of the optical phase and, respectively, pressure or temperature variations. These models are not supported by oceanic measurements. Moreover, the legacy models directly contradict available laboratory and oceanic measurements when it comes to directionality of DAS as an acoustic sensor. This paper presents new, physics-based models of pressure and temperature sensitivity of DAS. DAS transfer functions are rigorously derived from first principles for a simplified model of fiber-optic cable as a straight, jacketed or unjacketed fiber immersed in fluid. In addition to magnitude of wave-induced pressure and temperature perturbations and material parameters of the fiber, the jacket and surrounding medium, DAS response is found to depend on the speed of the wave propagation along the fiber. Extensions of the DAS transfer function model to a more realistic cable structure and to buried cables will be briefly discussed.

Keywords: Fiber-optic distributed sensing (DAS, DTS, DSS), Systems and observatories, Marine geology and geophysics
Abstract: Distributed acoustic sensing (DAS) measurements on seafloor cables are advantageous for seismic observations, since they provide a dense linear one-dimensional seismic array. For seismic research purposes it is desirable to observe seismic waves spatially in two-dimension. Therefore, DAS measurements using a multiple marine cables simultaneously allow two-dimensional seismic observation on the seafloor. There are two seafloor seismic and tsunami observation systems are currently operating off Sanriku coast in northeastern Japan. Each of those cabled systems has seismometers and pressure gauges, and both cables originate from the same landing station. One cable system installed in 1996 (1996 system) has spare/dark fibers in the seafloor cable, however, another system deployed in 2015 (2015 system) utilizes all fibers for transmission of data. Therefore, the 1996 system is frequently used for DAS observation on its dark fiber. However, two-dimensional DAS observations with multiple seafloor cables increases accuracy of earthquake locations. The 2015 system transmits scientific data using ICT with WDM and bidirectional communication via SFP optical transceivers. When carrying out DAS measurements on an active systems, any negative effects on data transmission must be avoided. The optical receiver module in the SFP has a narrow optical filter to increase signal-to-noise ratio. Careful choice of wave length and strength of the transmitted laser light, in combination with an optical splitter including a narrow WDM filter and a FPGA-based coherent optical frequency domain reflectometry (OFDR) system enables simultaneous DAS observation and TCP/IP communication. Utilizing constant and low power laser light provides continuous DAS measurements without data communication interference. In February 2026, we installed the new DAS system to the actively operating 2015 system off Sanriku. Since this DAS system can interrogate two fibers independently, we configured a simultaneous DAS observation on an active/lit fiber of the 2015 system and on a spare/dark fiber of the 1996 system are performed. We successfully obtained records of many earthquakes including micro-earthquakes due to low noise environment.

Keywords: Fiber-optic distributed sensing (DAS, DTS, DSS), Distributed hydrophone arrays and networked acoustic systems
Abstract: Distributed Acoustic Sensing (DAS) often employs commercial telecommunication cables (dark cables) for sensing and monitoring of ocean environments. The tens of hundreds of kilometers of these cable limits the interrogator's sampling frequencies to a few Hz to a couple of kHz depending on the cable length. For example, a standard length of 50km cable allows one to sample at a maximum frequency of 2kHz thus limiting the detection of usable acoustic frequencies to 1kHz or less. Custom experiments allow use of much shorter cables and a way to measure the sensitivity of DAS at higher frequencies than otherwise would have been possible. Apart from tailoring the setup to specific test scenarios, a custom experimental design also allows use of different cable types and layouts and without it being buried under the seabed. The Acoustic Research Laboratory (ARL) under the Tropical Marine Science Institute (TMSI) and at the National University of Singapore has setup an underwater DAS experiment using COTS fiber optic cables. Experiments were conducted successfully to test the custom DAS cable detection performance using many acoustic events such as surface vessels, divers and acoustic transmissions. Knowledge of cable geometry is critical when detecting and tracking sound sources at sea with respect to the cable position. During the custom experimental setup, the deployment of cable is influenced by currents, wind and waves and the actual geometry may differ from intended one. Hence, estimating the cable geometry after the cable is laid out is important. Another important factor to consider is the directionality of the DAS cable array. The gauge length introduces additional directivity to the DAS cable though it helps to reduce noise through spatial averaging. We estimate the shape of the custom laid DAS cable with reasonable accuracy using a cost minimization algorithm and we also demonstrate that the DAS cable directivity is influenced by the gauge length and the estimated results are consistent with our qualitative assessment of DAS directionality

Keywords: Fiber-optic distributed sensing (DAS, DTS, DSS), Calibration of acoustic systems and metrology, Ocean noise
Abstract: This work presents a comparative study of single-frequency intensity- and phase-based distributed acoustic sensing (DAS) over subsea cables. The study highlights key differences in sensitivity, spatial resolution, and signal-to-noise ratio (SNR), revealing practical considerations and insights of using them for underwater asset monitoring and environmental monitoring. The intensity-based DAS exhibited stronger high-frequency components in both background measurements and vessel crossing events and required a higher effective low cut-off frequency when applying filtering to suppress the background noise. The phase-based DAS achieved higher SNR across the investigated frequency bands.

Keywords: Fiber-optic distributed sensing (DAS, DTS, DSS), Smart cables and telecom-enabled ocean observation, Calibration of acoustic systems and metrology
Abstract: Distributed Acoustic Sensing (DAS) on submarine telecommunications cables could enable large-scale ocean acoustic monitoring using existing infrastructure. Although previous studies have detected low-frequency baleen whale vocalizations on long-haul fibers, performance at higher frequencies remains poorly characterized. We evaluated detection of 100-200 Hz signals using controlled playback experiments on three telecommunications-cable DAS systems with different configurations. Synthetic tones and frequency sweeps were transmitted at known source levels and source-cable distances. Both 100 and 200 Hz tones were detected on all three systems. Using a 3 dB mean signal-to-noise ratio threshold in band-pass-filtered distance-time plots, conservative 100 Hz detection ranges across systems were 101.29-651.53 m. The 200 Hz signals had mean source levels approximately 7 dB re 1 μPa at 1 m lower than the 100 Hz tones and were detected at shorter ranges of 17.91-144.52 m. Single-channel time-frequency plots confirmed the expected playback structure but often lacked clarity. Delay-and-sum beamforming substantially improved signal-to-noise ratio and resolved signals that were not distinguishable in single-channel data. These results demonstrate DAS detection capability in the 100-200 Hz band and provide a framework for evaluating the effects of source characteristics, cable configuration, and array processing. Extending DAS capability into this band broadens the potential use of existing telecommunications cables for marine-mammal and other undersea acoustic monitoring.

Keywords: Low-cost, scalable oceanographic sensors and platforms, Open-source hardware, software, and community innovation, Oceanographic instrumentation and sensors
Abstract: Miniature Open-Source CTD for Marine Animal Tagging Kevin D’Angelo and Salvador J. Jorgensen

Department of Marine Science, California State University Monterey Bay, 100 Campus Cr, Seaside CA, 93955

Abstract: Low-cost miniaturization in ocean instrumentation can increase oceanographic and ecological data streams by broadening use cases and removing financial barriers. We present an open-source conductivity–temperature–depth (CTD) sensor optimized for animal tagging. The system fits within a 20 mm diameter × 50 mm length cylindrical carbon fiber housing, including onboard control electronics and a lithium-polymer battery. Conductivity sensing architecture is based on the AD5933 12-bit impedance analyzer IC paired with 1/8” cylindrical graphite electrodes directly epoxied to a compact 11 × 27 mm PCB. Depth is measured using an integrated MS5837 micro pressure sensor with a rating to 300 m, while temperature is obtained from an onboard semiconductor sensor. A 4-wire I²C interface provides power and communication from an Arduino-based data logger. The system emphasizes affordability, with an estimated bill of materials near 100 in low-volume production and a reliance on 3D-printed components. Calibration is performed against a reference CTD using a temperature-varying home-made potassium chloride (KCl) solution to derive a linear correction that is stored in an onboard EEPROM. The design is optimized for seawater salinity (~33 PSU) however, the calibration approach and sensing range can be adapted for other conductivity regimes. This work demonstrates a pathway toward distributed, animal-borne ocean sensing using accessible, open-source hardware.

Keywords: Open-source hardware, software, and community innovation, Oceanographic instrumentation and sensors, Citizen science and community involvement
Abstract: Oceanographic research can be prohibitively expensive. Open source systems built with consumer-off-the-shelf parts, like the SeaLabMOS family of systems, can help make hardware more accessible. However, the physical system is only part of the solution; it must also be reasonably accurate for the given scientific objective(s). To validate the performance of SeaLabMOS systems during a real field campaign, two deployment methods were used to resolve a transient salinity minimum zone (i.e., baroclinic convergence zone) last observed twenty-eight years ago in Mare Island Strait, part of San Francisco's North Bay. Two five-day deployments were completed with systems arranged vertically on a bottom frame, then spatially distributed from the mouth of the strait northward. SeaLabMOS systems resolved the salinity minimum zone, showing it recurs twenty-eight years on, enabling analysis of its salinity and temperature anomaly, along-channel propagation, and tidal phasing. We also encountered and demonstrate potential drawbacks and pitfalls from using open source systems.

Keywords: Low-cost, scalable oceanographic sensors and platforms, Oceanographic instrumentation and sensors, Integration with autonomous platforms and observatories
Abstract: Current oceanographic salinity sensors primarily rely on conductivity measurements, which provide high accuracy but are relatively expensive and can be affected by changes in seawater ion ratios resulting from biogeochemical processes. Here, we demonstrate the use of Salino, a solid-state chloride ion-selective electrode (ISE) with a miniaturized reference electrode, as an alternative approach for measuring salinity. Chloride has historically been used as a primary indicator of salinity because of its conservative behavior and abundance in seawater, prior to the widespread adoption of stable and precise conductivity sensors. We test Salino in an estuarine environment, where salinity varies strongly in space and time. Salino was previously characterized in laboratory studies presented at OCEANS (Taylor et al., 2025; Nawaz et al., 2023). In this new study, Salino was deployed in situ for five days in a local estuary alongside a conductivity-based reference sensor. Water samples collected from the estuary and surrounding waters were analyzed using ion chromatography to quantify chloride and sulfate concentrations. These measurements revealed deviations in ion ratios from those expected in open-ocean seawater. To evaluate the influence of non-chloride ions, estuarine samples and NaCl solutions prepared at equivalent chloride concentrations were compared using Salino and a commercial conductivity-based sensor and the apparent salinity difference associated with deviations in sulfate-to-chloride ratio was characterized, demonstrating differences between chloride ISE and conductivity-based measurements under altered ionic compositions. These results demonstrate that direct chloride measurement can provide a more chemically specific measurement of salinity compared to conductivity in environments where freshwater mixing and biogeochemical processes alter seawater ion ratios. Salino therefore provides a promising low-cost approach for characterizing the mixing of seawater and freshwater in estuarine environments. In the open ocean where ion ratios are nearly constant, Salino may provide a low-cost alternative to traditional conductivity based salinity sensors.

Keywords: Low-cost, scalable oceanographic sensors and platforms, Oceanographic instrumentation and sensors, Real-Time Data Quality Control
Abstract: Nearshore wave conditions at coastal structures and shorelines remain among the least-observed parameters in the U.S. ocean observing enterprise. No operational system provides simultaneous, real-time wave and water level (WWL) measurements at the coast: offshore buoys and nearshore models supply wave information and sparse tide-gauge networks supply water level, but the two are rarely co-located and referenced to a common datum. We present a low-cost, structure-mounted gauge that measures sea-surface elevation with a single downward-looking acoustic sensor. The device samples at 5 Hz with centimeter vertical precision, yielding datum-referenced water level and wave parameters from one deployment on existing infrastructure. Because the high-frequency wave signal and still water level share a common datum, the gauge produces a continuous, wave-resolved water level record at the point of impact. To date, a prototype deployed at the Santa Cruz Municipal Wharf (~7 m depth) since December 2025 has produced over seven months of continuous data and nearly 2.5 million individually resolved waves. Bulk parameter data correlate with regional buoy and model references, and a structure-referenced overtopping analysis yields empirical critical-elevation thresholds previously unavailable to wharf operators.

Keywords: Low-cost, scalable oceanographic sensors and platforms, Open-source hardware, software, and community innovation, Rapid-deployment systems and modular instrumentation
Abstract: Environmental monitoring by Autonomous Underwater Vehicles (AUVs) is a highly promising application area, with examples including pollution monitoring, biological mapping, and radiation-related inspection.

However, developing and evaluating exploration algorithms in experiment remains challenging, since real-world fields are difficult and expensive to reproduce safely and repeatably in controlled underwater environments. As a result, many existing studies rely on simulated measurements and therefore miss important real-world effects like sensor noise and robot-environment interaction.

We address this gap by proposing a low-cost, reproducible experimental testbed for scalar field exploration in confined environments. The setup uses high-power infrared LEDs to generate stationary irradiance fields in a water tank, which are measured by a custom photodiode sensor array mounted on a mobile underwater robot. We characterize the system experimentally, including attenuation over distance and angular sensitivity, and demonstrate robot-in-the-loop field mapping with a BlueROV2.

The resulting testbed is open-source, easy to reproduce, and supports the controlled benchmarking of underwater exploration methods under realistic sensing conditions. In particular, it lowers the barrier for research groups that have access to small underwater test tanks but not to large experimental facilities or immediate field deployments, thereby supporting the growing underwater research community working on information gathering and exploration.

Keywords: Under-ice robotics, sensing, and operations, Remote sensing and field campaigns in polar regions, Long-term observing systems and international collaborations
Abstract: Operations involving autonomous operations in ice-covered waters present significant challenges to be addressed. This work outlines the methodology for developing resilient under-ice navigation for unmanned underwater vehicles (UUVs), and documents preliminary field testing results in a relevant arctic environment.

Keywords: Under-ice robotics, sensing, and operations, Rapid-deployment systems and modular instrumentation
Abstract: The icy moons Europa and Enceladus are prime targets in the search for extraterrestrial life due to strong evidence of subsurface oceans beneath their ice shells. Upcoming missions such as JUICE and Europa Clipper aim to characterize their habitability, while future lander missions may directly access and sample these hidden oceans. To advance the required technologies, the German Aerospace Center initiated the TRIPLE project (“Technologies for Rapid Ice Penetration and Subglacial Lake Exploration”) in 2018, focusing on the development and validation of systems for ice penetration, subglacial exploration, and in situ sampling under terrestrial analogue conditions in Antarctica. In parallel to space exploration objectives, these technologies provide new opportunities for studying ice-covered marine environments on Earth, particularly beneath Antarctic ice shelves, where access remains extremely limited. This work presents the development of a compact, modular science payload integrated into a nano-scale autonomous underwater vehicle (nanoAUV), designed for deployment through narrow boreholes using the TRIPLE IceCraft melting probe system. The nanoAUV is equipped with a Multi Science Probe (MSP) featuring interchangeable sensor heads (including chlorophyll-a, turbidity, and dissolved oxygen), a CT sensor for oceanographic measurements, a side-scan sonar for spatial mapping, and a miniaturized in situ filtration system for sample acquisition. Collected samples are analyzed ex situ using the field-portable AstroBioLab, which integrates automated Sterivex processing, fluorescence spectroscopy, and microscopy, complemented by DNA sequencing and mass spectrometry. The system is scheduled for deployment during a field campaign at the Neumayer III Station in Antarctica, where the nanoAUV will operate beneath shelf ice to investigate environmental conditions within a ~200 m radius of the borehole. The final paper will describe the details of the scientific payload and its application as a modular system within the upcoming field test at NM3 on the shelf as well as the potential future applications as modular system e.g. beneath sea ice with alternative carriers, such as BlueROVs.

Keywords: AI/ML for oceanographic and environmental data analysis, Under-ice robotics, sensing, and operations, Surveillance, detection, and threat-assessment techniques
Abstract: Detection of oil spills in ice-covered freshwater environments poses a significant challenge, as traditional optical and radar-based sensing methods are often ineffective when water surfaces are obscured by ice. This study proposes a novel approach using unmanned aerial vehicle (UAV)-based thermal infrared (IR) imaging to detect oil spills beneath ice by identifying anomalies arising from differences in thermal properties and heat transfer behavior. Oil beneath ice alters system heat flux, producing measurable deviations in ice surface temperature relative to unaffected regions, particularly during periods of rapid atmospheric temperature change. Field observations using a simulant to replicate the thermal properties of oil show that subsurface layers with oil-like thermal properties produce consistent and distinguishable surface temperature anomalies detectable by UAV-mounted thermal sensors. Results indicate that while steady-state temperature differences are minimal, the change in heat flux produces detectable thermal signatures during transient events under certain environmental conditions. To interpret and generalize these findings, a one-dimensional transient heat transfer model was developed to characterize the underlying thermal behavior and evaluate detectability across varying environmental conditions. Using our heat transfer model, we are able to identify the conditions under which this method would be useful in a real-world oil spill response. These findings demonstrate that temporal thermal analysis provides a viable approach for detecting oil beneath ice using UAV-based IR imaging. This work addresses a key limitation of conventional thermal sensing and offers a scalable, non-invasive method for oil spill monitoring in ice-covered environments.

Keywords: Sonar signal processing, Signal coherence and fluctuation, Sound propagation and scattering
Abstract: In conventional open-ocean environments, underwater acoustic uncertainty used in sonar detection probability calculations is typically modeled as a normal distribution with a standard deviation of 8–9 dB. However, such assumptions may not be valid in Arctic environments, where sea ice and the unique sound speed structure significantly influence acoustic propagation. Variations in ice thickness, rough ice undersides, and changes in sound speed profiles introduce additional complexity, making field-based assessments essential for reliable sonar performance prediction.

This study investigates the transmission characteristics and uncertainty of mid-frequency acoustic signals using measurements collected in August 2025 in an ice-covered region north of the Chukchi Sea. Acoustic instruments were deployed beneath the ice through drilled holes, ensuring stable and controlled measurement conditions. The acoustic source was positioned at five depths (25–125 m), while receivers were located at ranges of 2.3–4.2 km and depths of 25, 75, and 125 m. A vertical array was installed at one location to examine depth-dependent variability. Continuous wave (CW) and linear frequency modulated (LFM) signals in the 2–3.8 kHz band were transmitted.

Numerical simulations incorporating in situ environmental data, including sound speed profiles and ice conditions, were performed to model acoustic propagation. Acoustic uncertainty was quantified by statistically analyzing the differences between modeled transmission loss and measured signals.

Results show that mid-frequency acoustic propagation in the Arctic is strongly affected by both sea ice and sound speed structure. In particular, propagation paths involving significant surface–ice interactions exhibit much greater variability than refracted or bottom-interacting paths. These findings indicate that conventional open-ocean assumptions are insufficient to represent acoustic uncertainty in ice-covered Arctic environments.

This study provides improved understanding of Arctic acoustic variability and supports more accurate sonar performance prediction and future acoustic operations in polar regions. This work was supported by the Korean Government (912A12701).

Keywords: Under-ice robotics, sensing, and operations, Remote sensing and field campaigns in polar regions, E - M sensing
Abstract: The Arctic Ocean is a region of interest both scientifically and geo-politically, however its most salient feature, year-round and seasonal ice-cover that can reach over 5 m thick, presents a significant obstacle to instrumenting and understanding this important region. The primary impediments are limited access opportunities, dynamic ice conditions, and an inability to access or communicate into the deep Arctic ocean beneath the ice without humans drilling a hole. This work aims to address the problem of through-ice communication by operationalizing a Magneto-Inductive communication system (MI-COMMS) that would allow communication through solid sea ice and a small amount of salt water to connect under-ice sensors to the rest of the world. If successfully deployed, such a system could foster a proliferation of sensors and capabilities that would enable a much denser sampling of the under-ice environment. In this paper, a prototype MI-COMMS system is introduced, and a test campaign that spans from Vermont, USA to the North Slope of Alaska is described.

Keywords: Offshore structures, Ocean energy, Offshore aquaculture structures
Abstract: The co-location of floating offshore wind turbines (FOWTs) and aquaculture systems is increasingly recognized as an effective strategy for maximizing ocean space utilization. Despite this potential, the hydrodynamic interactions between semi-submersible FOWTs and aquaculture net cages are not yet fully understood. This study provides an experimental investigation of the motion responses of a 1/80 scaled semi-submersible platform, modeled on the VolturnUS-S concept, co-located with aquaculture net cages under regular wave conditions. A series of water tank experiments are performed for wave periods ranging from 0.6 s to 2.0 s. The FOWT is evaluated with one to three net cages positioned upstream of wave propagation, with varying separation distances. Motion responses including surge, heave, and pitch are quantified using response amplitude operators (RAOs) for both the FOWT and net cages. The results indicate that the surge RAO of the FOWT remains largely unaffected by the presence of net cages across all tested conditions. Conversely, the heave RAO shows configuration-dependent variations when two or three cages are present, with both increases and decreases observed depending on wave period and separation distance. The pitch RAO is generally reduced in the co-located configurations, with the reduction becoming more evident at longer wave periods. For the net cages, RAO analysis reveals distinct trends: in the single-cage co-located configuration, the surge RAO generally increases at shorter wave periods and becomes comparable to or lower at longer periods, while the heave RAO is generally amplified. In multi-cage configurations, the surge response becomes more configuration-dependent and is frequently increased, whereas both heave and pitch RAOs generally decrease over the long-period ranges.

Keywords: Offshore structures
Abstract: In recent years, offshore data center and subsea loads have seen a significant high in demand due to fewer regulatory constraints, combined with the increase in availability of renewable energy sources promising a sustainable power solution for offshore energy systems. Previously, Diesel generator or long-range transmission line often used for offshore power system which are very expensive, less efficient and creates high carbon footprint. This paper proposed the hybrid offshore energy storage systems which combines of Hydrogen Energy Storage (HES) and Battery Energy Storage System (BESS) implementation in medium-voltage DC (MVDC) provide energy for offshore Data Center and Subsea loads with integrated of renewable energy sources such as offshore wind and floating solar. The original architecture proposed is depicted in Fig. 1 and is comprised of renewable generation, hybrid energy storage, and offshore loads all tied together through a common MVDC bus. The HESS consists of an electrolyzer to produce hydrogen, a storage tank for the hydrogen and a fuel cell for converting the hydrogen to electricity when needed to extended energy support for the micro grid. Alongside this the battery system provides fast response and short-term power balancing to help maintain stability during load or renewable transients. To connect various input energy sources to the MVDC 6kV bus, a solid-state transformer (SST) based on multilevel isolated DC-DC Dual Active Bridge (DAB) converter is used in the system. It provides voltage conversion in an efficient way, bidirectional power transfer, as well as galvanic isolation and modularity. The modularity feature permits isolating faulty units from the rest by keeping up the DC bus voltage at an adequate level through the coordination operation between healthy units. Simulation results have proven that our proposed system is able to stabilize MVDC bus voltage and properly govern power sharing of renewable sources and hybrid storage systems in different operating scenarios.

Keywords: Ocean energy
Abstract: This work develops a physics-informed neural network (PINN) tool for predicting bottomhole pressure (BHP) and temperature (BHT) in offshore CO2 pipeline–wellbore injection systems. We first develop a steady-state PIPESIM model to generate training data across variations in injection rate, pipeline geometry, well depth, and reservoir pressure. The PINN combines these data with momentum and energy conservation, the Peng–Robinson equation of state, and wellhead boundary conditions to capture coupled CO₂ pressure-temperature behavior. Over the simulated operating range, the model achieves a MAPE of 0.20% and RMSE of 0.54 °F for BHT, and a MAPE of 0.29% and RMSE of 20.43 psia for BHP. Ablation analysis shows that physics constraints improve BHP prediction, while enforcing boundary conditions further reduces errors in both BHP and BHT. The trained model is implemented in a Streamlit interface for rapid single- and multi-scenario evaluation without rerunning PIPESIM or retraining. The resulting framework provides an efficient surrogate for pressure–temperature prediction and scenario analysis in offshore CO₂ injection systems

Keywords: Sustainable offshore industries and the evolving blue economy, Numerical modeling and simulation, Ocean energy
Abstract: Reliable offshore energy is vital for environmental observation, desalination, aquaculture, and unmanned underwater vehicle (UUV) charging. Many of these systems currently rely on power solutions that are costly, logistically challenging, and maintenance intensive. In this context, offshore integrated energy systems with multiple energy sources offer a promising approach that can also help mitigate spatial conflicts. Floating photovoltaic (PV) systems can be an important part of such offshore energy platforms, but effective design requires understanding how platform motions affect PV performance and fixed module orientation. Addressing this challenge requires modeling tools that capture offshore platform motion, system interactions, and support design optimization.

This paper introduces PVlib.jl, an open-source photovoltaic modeling tool in Julia with utilities for offshore floating PV applications. PVlib.jl replicates core functionality of existing pvlib packages and adds features to account for floating offshore platforms and hybrid renewable energy systems, including 6 DOF motion, thermal averaging, ocean albedo, and shading. Because it is implemented in Julia, PVlib.jl interfaces efficiently with automatic differentiation tools, enabling gradient-based optimization for integrated energy models.

To demonstrate its capabilities, this paper models a PV module mounted on the Pioneer Array Coastal Surface Mooring system, an observation buoy in the Atlantic Ocean powered by solar, wind, and wave energy. A previously developed hydrodynamic model provides 6 DOF buoy motion inputs to PVlib.jl, which computes panel orientation over time, imports meteorological data, determines solar irradiance, and calculates module DC power using the Sandia Array Performance Model formulation. An optimization routine using automatic differentiation identifies an installation tilt of 40.4° and azimuth of 171.3° that maximize average DC power generation (28.4 W). The study serves as a proof of concept for PVlib.jl’s modeling and optimization capabilities, with future work prioritizing coupled hybrid renewable energy systems.

SNL is managed and operated by NTESS under DOE NNSA contract DE-NA0003525.

Keywords: Sustainable offshore industries and the evolving blue economy, Maritime decarbonization and eco-efficient operations, Marine GIS and data fusion
Abstract: This study outlines the need for an integrated and user-friendly marine spatial planning tool to support the co-location of marine energy systems with coastal and marine industries across U.S. Pacific states. Marine energy, which is derived from waves, tides, currents, and thermal, salinity, or pressure gradients, offers resilient power sources that can be paired with coastal industries like aquaculture, desalination, marine carbon dioxide removal, mineral recovery, and wastewater treatment. Co-location enables shared use of resources, improved planning, and streamlined infrastructure needs, but project planners currently lack a comprehensive decision-support tool that brings together geospatial, environmental, infrastructural, and oceanographic datasets. To scope such a siting tool, this study examines four regional use cases in Alaska, California, Hawaii, and Washington, leveraging extensive datasets across the following categories: marine energy resources, oceanography, bathymetry, natural and cultural resources, national security, existing infrastructure, and ocean uses. The analysis highlights major data access challenges, inconsistent repository availability, and critical data gaps such as missing infrastructure layers and uneven ecosystem data coverage. Improved, stable, and publicly accessible geospatial data systems are essential for informed siting decisions. A unified decision-support tool would significantly aid ocean industry stakeholders in advancing reliable and resilient coastal development and marine energy deployment.

Keywords: Real-time decision systems for ocean robotics and autonomous operations, Autonomous underwater vehicles
Abstract: Autonomous navigation in underwater cave en- vironments is essential for search-and-rescue operations, sci- entific exploration, and emergency egress. Traditional navi- gation systems commonly depend on dense visual features for localization and mapping. In underwater caves, however, visual degradation can undermine feature-based localization, sonar-based mapping may yield overly conservative obstacle representations, and communication constraints preclude real- time human guidance. To address these limitations, we propose an autonomous underwater cave navigation framework that leverages a Vision-Language Model (VLM) with Chain-of- Thought (CoT) reasoning to infer navigable directions from environmental cues, including light intensity gradients, passage morphology, and geometric complexity, captured through multi- modal inputs comprising RGB imagery, depth maps, and sonar- based vertical-clearance measurements, thereby supporting safe 3D navigation through confined cave passages. High-fidelity simulations across multiple cave topologies demonstrate that the proposed framework completes all evaluated end-to-end traversals without collisions while maintaining safe clearance from cave boundaries.

Keywords: Autonomous underwater vehicles, Security and resilience of AUV/USV/ROV platforms, Real-time decision systems for ocean robotics and autonomous operations
Abstract: The proliferation of low-cost uncrewed underwater vehicles (UUVs) poses an asymmetric defense problem for critical maritime infrastructure, ports, naval vessels, and high-value nearshore structures. Contemporary maritime defense architectures remain weighted toward situational awareness, with limited capability to deny or neutralize threats once detected.

We present KRAKEN, a multi-UUV architecture for the cooperative sensing and interception of underwater threats. Sensor-laden tethered UUVs are deployed to localize inbound targets, a central processing node fuses their observations, and a payload-carrying tethered UUV is continuously directed along the dynamic track until interdiction. A prototype tethered UUV was built to validate component integration, bench tether communications, and fly-by-wire control.

We conducted a 40,700-trial Monte Carlo campaign using a reduced-order model incorporating sonar detection, navigation uncertainty, communications latency, interceptor kinematics, tether drag, and multi-vehicle cable interference to evaluate the architecture against autonomous single-interceptor and acoustic-communications alternatives. First, cooperative target tracking substantially reduces sensitivity to initial cue error: the fused-scout architecture maintains capture probability Pcap ≥ 0.987 through 10◦ of cue-bearing error and Pcap = 0.860 at 20◦, while a cue-fired interceptor using its own onboard sensors degrades sharply as cue uncertainty increases. Second, low-latency communications are critical for reliable closed-loop interception. With three scouts and a 3.0 m/s threat, the modeled fiber architecture, achieves Pcap = 1.000, compared with Pcap = 0.375 for the best case acoustic link. Third, tether configuration strongly affects interception performance. Vehicle-side payout of a 0.9 mm fiber link maintains Pcap ≥ 0.96 through 4 m/s for straight inbound threats, whereas UUV-towed configurations incur significant hydrodynamic drag and risk cable-clearance failures. The results suggest cooperative interdiction of underwater threats depends not only on interceptor speed or onboard autonomy, but on cooperative localization, low-latency communications, engagement geometry, an

Keywords: Real-time decision systems for ocean robotics and autonomous operations, Vehicle navigation
Abstract: Autonomous surface vessels require collision-avoidance systems that can adapt to multi-vessel encounters while remaining traceable to the International Regulations for Preventing Collisions at Sea (COLREGs). Conventional rule-based planners encode these regulations as predetermined maneuvers, providing structured and interpretable behavior that becomes increasingly conservative in crowded waters. Conversely, deep reinforcement learning may reduce conservatism, but yields uninterpretable policies whose safety and rule compliance are correspondingly difficult to verify. To bridge this gap, this paper presents a hybrid architecture within the CORALL framework to restore decision-level traceability. A large language model (LLM) processes an environmental state against encoded COLREGs regulatory guidelines to classify encounters and issue discrete navigational intents. A low-level reinforcement learning policy then issues heading and speed commands for a single ownship and is directionally constrained by COLREGs-based LLM intent guidance. Across a range of encounter geometries, this pipeline enabled successful path-following, collision-free ownship trajectories. The results indicate that pairing language-based reasoning with deep-learning-based planning can improve collision-avoidance efficiency while keeping maneuvers attributable to COLREGs.

Keywords: Real-time decision systems for ocean robotics and autonomous operations, Imaging and vision, Vehicle navigation
Abstract: Precise obstacle detection is essential for autonomous navigation of Unmanned Surface Vehicles (USVs) in maritime environments. For USV autonomy, perception systems must detect, localize, and map obstacles in real time for path planning, obstacle avoidance, and mission execution. Many existing approaches stop at image detection using AI or traditional methods and do not progress to obstacle maps for navigation. This work introduces a real-time AI-assisted perception and semantic mapping framework for a Wave-Adaptive Modular Vehicle 16 (WAM-V 16) USV.

The vehicle perception system uses a ZED 2i stereo camera and a Velodyne VLP-16 Hi-Res LiDAR to detect, classify, and localize maritime obstacles. LiDAR point clouds are processed using k-means clustering to identify obstacle candidates in 3D space, while the stereo camera uses a YOLO-based object detection model to classify observed obstacles. Previous work relied on fixed geometric thresholds; the proposed framework uses XGBoost as a probabilistic fusion layer to associate asynchronous camera and LiDAR detections, correct their global positions, and maintain a persistent semantic obstacle map.

A major challenge in global-frame maritime mapping is that camera detections, LiDAR clusters, and GNSS measurements are not time-synchronized. To address this, XGBoost is used in a three-stage fusion and mapping approach: association, map-frame correction, and semantic map update. The system was evaluated through buoy-field mapping experiments with the WAM-V 16 platform and through autonomous navigation using the generated semantic map to traverse a green-red buoy gate. The test environment contained 20 ground-truth buoys. Compared with a previous geometric fusion and Kalman-filter-based tracking approach, which achieved a mean error of 0.55 m and RMSE of 0.66 m, the proposed XGBoost-based framework matched all 20 visible ground-truth buoys and achieved a mean error of 0.123 m and RMSE of 0.180 m. This corresponds to reductions of 77.5% in mean error and 72.8% in RMSE.

Keywords: Vehicle navigation
Abstract: Recent advances in reinforcement learning (RL) have enabled deep neural networks to autonomously control agents in complex, real-world environments. A prominent application is the navigation of Autonomous Surface Vessels (ASVs), where safety and trust are particularly important because of frequent interactions with human-operated vessels. Although RL-based models have demonstrated considerable potential for ASV navigation, understanding their behavior and decision-making in complex maritime environments remains essential. This paper presents a convolutional neural network (CNN)-based Double Deep Q-Network (DDQN) for controlling ASV movement in constrained maritime environments, including Traffic Separation Schemes (TSS). Through simulation-based experiments, we compare two explainable AI methods, Grad-CAM++ and Layer-wise Relevance Propagation (LRP), to analyze the trained model’s behavior at different stages of navigation. The results show that Grad-CAM++ produces relatively diffuse explanations, whereas LRP more precisely identifies the locations of nearby vessels as influential regions in the model’s decisions. Across multiple threshold values, LRP statistically outperforms Grad-CAM++ in identifying the image regions that most strongly influence the model’s decision-making. These findings provide practical guidance for selecting XAI methods in safety-critical applications and contribute to the development of more transparent and trustworthy human-AI systems for autonomous maritime operations.

Keywords: Vehicle navigation, Autonomous underwater vehicles
Abstract: Advances in underwater vehicle capability and payload efficiency are pushing Slocum gliders from multi-month to multi-year time horizons and across increasingly remote regions. However, shore-side piloting workflows have not kept pace with the evolution of the platform technology. At these larger operational scales, pilots must reconcile evolving ocean currents and weather windows with hard vehicle constraints such as energy budget, endurance, communication cadence, and safety margins. Existing mission planning workflows remain largely manual, requiring synthesis across forecast products and repeated time-consuming replans as conditions evolve. Here, we present the Glider Guidance System 3 (GGS3), a decision-support and route-planning framework designed for glider operations. The GGS3 ingests multiple ocean forecast products and converts them into actionable mission guidance by estimating drift and along-track progress under forecast currents, evaluating potential paths against vehicle and operational constraints, and generating routes/waypoints that balance mission objectives with risk and efficiency. The system produces pilot-facing products (e.g., depth-averaged current fields, drift summaries, and recommended paths) to support rapid iteration during near-real-time operations. We demonstrate GGS3 in support of the Teledyne Webb Research Sentinel Glider REDWING’s attempt at the first global circumnavigation mission, where long mission segments and evolving mesoscale conditions amplify the cost of suboptimal routing. Examples from navigational decisions made during key moments of the initial Atlantic crossing show how GGS3 quantifies predicted time and battery performance, while minimizing operational risks. We discuss how decision-making support was provided during operations, the iterative refinement of the GGS3 based on operational constraints and pilot requirements, and the future of the GGS3 framework to support future legs of the circumnavigation mission.

Keywords: AI/ML for oceanographic and environmental data analysis, Low-cost, scalable oceanographic sensors and platforms, Vehicle navigation
Abstract: Monitoring, reporting, and verification of marine carbon dioxide removal (mCDR) is critical for climate change mitigation, yet existing dissolved inorganic carbon (DIC) measurement methods rely on consumable reagents, prone-to-biofouling membranes, or manual sampling that do not allow persistent autonomous operations. In this paper, we present an integrated sensing and physics-informed navigation framework for persistent, high-resolution DIC monitoring. The sensing payload comprises a low-SWaP (Size, Weight, and Power) all-fiber Raman spectroscopy system integrated into an AUV-compatible towed casing for non-invasive, reagent-free in-situ chemical quantification. To maximize information gathering under vehicle energy constraints, we couple the sensor platform with a hydrodynamic transport model where 2D Shallow Water Equations (SWE) driven by public GEBCO bathymetric and meteorological data force an advection-diffusion-reaction model of coastal DIC plumes. We formulate plume tracking as a finite-horizon optimal control problem and solve it dynamically using Model Predictive Control (MPC) over extended deployment windows (T = 36,000–40,000 s). Across four distinct Florida coastal environments, simulation results demonstrate that our energy-aware navigation captures near-equivalent spatial DIC information compared to an energy-unconstrained pursuit strategy while drastically reducing thruster control effort.

Keywords: Maritime domain awareness and autonomous monitoring, Vehicle navigation, Security and resilience of AUV/USV/ROV platforms
Abstract: Under the context of SLAM, point and line features evidence a strong complementarity: point features offer accurate relative localization, while line features are less prone to detection failure. This work presents a two-dimensional sonar-based underwater SLAM framework, built on the Extended Kalman Filter formulation, that combines point and line features extracted from imaging sonar data. The full feature extraction layer is prepared to operate directly in polar acoustic images, further improving computation efficiency and perception robustness. A line feature descriptor and matching strategy are introduced, building on the geometric information extracted from the acoustic landmark representation. The proposed framework is evaluated in a confined underwater environment using a robotic platform equipped with a low-grade inertial sensor and an imaging sonar. Results illustrate the enhanced performance achieved.

Keywords: Real-time decision systems for ocean robotics and autonomous operations, Automatic control, Oceanographic instrumentation and sensors
Abstract: Quantifying, and eventually modelling, deep-ocean carbon storage requires detailed understanding of periods with unusually high activity in the biological carbon pump. “Pulse” events, when large amounts of biological material sink to the deep-sea over a short period of time, are increasing in magnitude, duration, and frequency in the California Current (USA). They can account for a significant fraction of annual deep-ocean carbon storage in this region but remain unpredictable and hard to study.

Here we present progress toward targeted study of unpredictable pulse events through monitoring, event detection and command communication by a low-power instrument that then wakes up and initiates physical sample collection by a separate instrument with limited collection capacity. We added on-board image analysis and adaptive event detection to the Sedimentation Event Sensor (SES), an optical sediment trap with acoustic communication capabilities (McGill et al. 2016). After each sample collection, the SES calculated image attenuance and checked for rapid increases in marine snow accumulation. Image attenuance has been shown to correlate with particulate organic carbon flux (Estapa et al. 2024). If the slope of running mean attenuance exceeded a predetermined value, then the SES sent an acoustic “trigger” signal to initiate sampling by a nearby acoustics-enabled conventional sediment trap. Two triggering approaches were tested: 1) slope-only thresholds, and 2) toggle between two different slope thresholds based on recent triggering rates. This workflow aimed to detect the onset of, and collect during, pulse events. Sampling capacity of the conventional sediment trap was 21 cups, so the overall goal was to develop algorithms that made efficient use of that number of triggers. In three field trials (2024–2025; outer reaches of Monterey Bay, California, USA; ~900 m deep), the SES successfully detected unpredictable carbon storage events and sent acoustic trigger commands. We present details of triggering performance, and rates of false positive and negative triggers.

Keywords: Automatic control, Open-source hardware, software, and community innovation, Dynamic positioning
Abstract: Ocean exploration with unmanned underwater vehicles (UUVs) has an increasing need for full maneuverability in cluttered, three-dimensional environments such as canyon walls, hydrothermal vents, and coral structures. Achieving this requires 6-degrees-of-freedom (DOF) control on agile robots, but most open-source controller libraries focus on terrestrial and aerial robots. The ones that support UUVs either assume pitch-roll-stable platforms, fixed thruster configurations, or specific communication middleware. In this paper, we present a control framework for thruster-based UUVs that addresses these limitations. The framework performs attitude control directly on the special orthogonal group (SO(3)), avoiding Euler-angle singularities, implements pose and twist loops through a cascaded controller structure, and allows thruster allocation that permits arbitrary thruster configuration. The communication layer is decoupled from the controller logic, allowing deployment across a variety of vehicles and pre-existing middleware. Simulation and field results on MOLA, a 6-DOF agile UUV, validate stable pose tracking across the full attitude space, including high-pitch and high-roll orientations.

Keywords: Sonar signal processing, Array signal processing and array design, Acoustical oceanography
Abstract: Matched-field processing (MFP) is widely used for underwater source localization owing to its ability to exploit the full acoustic-field structure in ocean waveguides. However, conventional MFP (conv-MFP) is highly sensitive to environmental mismatch, such as uncertainties in the sound-speed profile and seabed properties, which can significantly degrade localization performance. Although broadband conv-MFP improves robustness through frequency averaging, it does not resolve the underlying structural mismatch, particularly in dispersive shallow-water waveguides. Frequency-difference MFP (FD-MFP) further enhances robustness by exploiting low-frequency difference components with improved coherence. However, the modal structure of the measured difference field is generally inconsistent with that of the replica field, resulting in a mismatch in replica-field construction and, consequently, reduced localization accuracy. To address these limitations, this paper proposes the mode-difference MFP (MD-MFP). The method constructs replica fields directly in the mode domain, where nonlinear modal interactions are explicitly represented, ensuring improved structural consistency. In addition, a mode-selection strategy is introduced to suppress the mismatch-inducing components, which further enhances robustness, particularly for short-aperture vertical arrays. The theoretical formulation is developed based on the normal-mode representation of the acoustic field, and the ambiguity surface is derived within the mode-difference framework. The proposed method effectively reduces undesired interference while preserving constructive matching. Simulation and experimental results in mismatched shallow-water environments demonstrate that MD-MFP achieves sharper ambiguity surfaces, higher resolution, and more accurate localization compared with conv-MFP and FD-MFP. These results demonstrate that the proposed method provides a robust and physically consistent framework for underwater source localization in practical environments.

Keywords: Hydrography / Seafloor mapping / geodesy, Acoustical oceanography, Integration with autonomous platforms and observatories
Abstract: NOAA Ocean Exploration seeks to integrate uncrewed surface vessel-based bathymetric mapping products into its routine data collection. This “force augmentation” approach reserves increasingly expensive ship time for both high-resolution deep water remotely operated vehicle operations and water column data collection that is best suited to crewed vessels. Because of the significant and rapid developments in private sector bathymetry collection and hydrographic services deployment, there is now capacity for routine mapping from uncrewed surface vessels (USVs). NOAA has previously leveraged this capacity nearshore through the Office of Coast Survey for its mandated nautical chart production. The need to understand this evolving capability in a deep water context drove prior partnering between NOAA Ocean Exploration and the University of New Hampshire Center for Coastal and Ocean Mapping/NOAA-UNH Joint Hydrographic Center. This partnership, funded through the NOAA Ocean Exploration Cooperative Institute, sought to evaluate this “Data as a Service” mapping approach using Saildrone’s Surveyor USV in Alaska in 2023 With this expanded knowledge base, NOAA Ocean Exploration made a decision to begin broadening the routine acquisition of USV-based mapping data. In 2025, NOAA Ocean Exploration partnered with Woolpert and Saildrone to begin closing significant mapping gaps in the western Pacific, within U.S.waters near the Marianas Islands. This project was made possible through collaboration and support from NOAA’s Office of Marine and Aviation Operations and Office of Coast Survey, resulting in the successful execution of this deep water mapping effort on the southeast side for the Marianas Trench that began in early February 2026. Managed through NOAA’s Office of Coast Survey, this project leverages the Hydrographic Services Contract mechanism to operationalize this “Data as a Service” model with adherence to IHO Order 1 technical specifications. Saildrone is delivering bathymetric and oceanographic data, other ocean physics variable data, and derived hydrographic products. NOAA Ocean Exploration chose to partner with Saildrone for this mission because the Surveyor platform is one of the few shore deployable USVs

Keywords: Sonar and transducers, Acoustical oceanography, Low-cost, scalable oceanographic sensors and platforms
Abstract: A framework for bathymetric-only seafloor mapping and classification using an entry-level multibeam echosounder (MBES) and platform was developed. Bathymetric data were acquired using a Surveyor 240-16 MBES attached to a BlueBoat uncrewed surface vehicle (USV) with two GoPro HERO10 cameras for ground-truth imagery. Geometric features were calculated at different scales from the bathymetric point clouds using CloudCompare. Image-informed 0.5-meter radius patches were labeled on the point clouds in tandem with the ground truth images, and a hierarchical random forest classifier was trained to classify sand, seagrass, hardbottom and reef structure patches. The area under the receiver operating characteristic (ROC AUC) curve was used to inform the geometric feature set for the hierarchical classifier. Within the development dataset, the model presented an accuracy of 82.6%. The macro-F1 score was 81.9%, and the balanced accuracy was 86.8%. Applying the model to an independent survey site yielded different results. It produced an accuracy of 77.6%. Considering all the present classes, the balanced accuracy was 50.1%. The macro-F1 was 47.2%. Sand transferred strongly to the independent site, while none of the 14 reef-structure patches were correctly classified, indicating sensitivity to habitat morphology in the training data versus the independent site.

Keywords: Acoustical oceanography, Systems and observatories, Integration of acoustics with imaging, physical, and biological data
Abstract: This work highlights results from studies in underwater acoustics and acoustical oceanography conducted by external research groups in collaboration with Ocean Networks Canada (ONC, a University of Victoria initiative.) ONC offers ocean acousticians the rare opportunity to access multi-year time series that include both hydrophone data and a variety of co-located oceanographic and seismic sensors, in environments ranging from shallow coastal/urban-sea waters to the 2,600 m of depth of the Cascadia Basin, to Cambridge Bay in the Arctic. One of the nodes on ONC’s North-East Pacific Time-series Undersea Networked Experiments (NEPTUNE) large-scale observatory enjoys close proximity to a hydrothermal-vent field on the Endeavour segment of the global mid-ocean ridge system, where tectonic plates diverge forming new crust. Here, at a depth of approximately 2,200 m, data from a number of instruments, including a volumetric hydrophone array and a camera, are being used to explore passive acoustics as a technique for long-term monitoring of vent activity in a very dynamic and chemically aggressive environment. The NEPTUNE site of Barkley Canyon offers a range of depths and environments, from submarine canyons to oxygen minimum zones, and from methane seeps to unique gas hydrate mounds. Here, a four-element array of hydrophones located at a depth of 646 m (near the axis of the deep sound channel) receives the signal of the Kauai Beacon, a very-low-frequency source situated near Hawaii at a distance of approximately 4,100 km. As part of a larger effort, several institutions leverage the Beacon signal for basin-scale acoustic observations over long time scales.

Keywords: Imaging and vision, Sonar imaging, Integration of acoustics with imaging, physical, and biological data
Abstract: Many of the current fisheries-independent surveys for reef fishes in the southeastern United States rely on optical imaging systems for the collection of data. However, the efficacy of optical systems is reduced in areas of higher turbidity and/or lower ambient light, both of which are encountered frequently within the survey universes. A proposed alternative approach to optical sampling utilizes acoustic cameras (high-frequency multibeam sonar) to detect targeted species and/or the substrate within the ensonified area. The acoustic systems are less impacted by environmental parameters and can operate at ranges further than even the most advanced optical systems. A paired optical/acoustic imaging system was developed in order to better characterize the performance of the acoustic systems, and validate the observed species. This paired system was built onto the existing SphereCam platform, a 360° stereo camera developed and utilized by NOAA Fisheries and employed both a Sound Metrics ARIS Explorer and a Kongsberg Flexview sonar. Preliminary data products show the sonar systems’ ability to provide occurrence and length estimates, however further analysis is required to explore the utility for species identification.

Keywords: Sonar signal processing, Sonar imaging, Autonomous underwater vehicles
Abstract: This paper presents a digital signal processing pipeline for dock-centric relative localization of an approaching underwater vehicle. The proposed approach assumes that the docking station is equipped with acoustic and sonar sensing modalities and acts as an external observation node. Acoustic measurements are affected by noise, low update rate, and outliers, while sonar observations are disturbed by clutter and intermittent target visibility. To address these issues, the pipeline combines acoustic outlier rejection, temporal smoothing, sonar preprocessing, target detection, and Kalman-based state estimation in the docking-station frame. The method is evaluated under nominal and degraded conditions using simulated or experimental data. Results show that signal conditioning and temporal filtering improve trajectory continuity and reduce relative localization error compared with direct use of raw measurements. This suggests that a dock-centric signal processing pipeline is a practical basis for infrastructure-assisted underwater docking.

Keywords: Low-cost, scalable oceanographic sensors and platforms, Rapid-deployment systems and modular instrumentation, Sonar imaging
Abstract: Abstract— Accurate bathymetric surveys in shallow coastal waters are challenging to obtain using conventional crewed survey vessels, which are constrained by draft limitations, maneuverability, operating costs, and personnel safety risks. This extended abstract presents the integration and field deployment of a commercially available Autonomous Surface Vessel (ASV), the Blue Robotics BlueBoat, as a versatile and cost-effective platform for shallow-water mapping. The system supports this through the combined use of survey-grade or entry-level multibeam sonars and a compact inertial navigation sensor (INS) with multi-frequency real-time kinematic (RTK) global navigation satellite system (GNSS) for centimeter-level positioning. Preliminary results show field deployments conducted in Moss Landing Harbor before and after a dredging operation, illustrating the system's capability to measure accurate change in shallow environments.

Keywords: Holography and 3D imaging, Array signal processing and array design, Calibration of acoustic systems and metrology
Abstract: Near-Field Acoustic Holography (NAH) is widely used to localize and quantify acoustic sources from scanning measurements, providing access to far-field directivity, radiated power, and spatial distribution of acoustic hot spots. However, cylindrical NAH relies on simplified propagation models and idealized geometries, which limit its applicability to complex underwater structures and lead to inaccuracies in far-field predictions, particularly near the acoustic horizon. To overcome these limitations, a formulation based on a Boundary Element Method (BEM) of NAH combined with Bayesian regularization is proposed, enabling explicit modeling of complex geometries while improving robustness to measurement noise and stabilizing the ill-posed inverse problem. The approach is evaluated through numerical simulations on a stiffened model representative of underwater structures, under both radial and axial excitations. Controlled datasets are generated using a conventional cylindrical hologram with overscan, ensuring compatibility with standard NAH acquisition procedures. The proposed method is compared to classical cylindrical NAH in terms of reconstruction accuracy and far-field directivity estimation, and the limitations introduced by the measurement geometry are discussed.

Keywords: Optical communication, Optical sensors and adaptive optics
Abstract: Underwater communication remains a critical challenge for marine operations, as acoustic systems suffer from low bandwidth, high latency, and detectability, while radio frequency signals are severely attenuated in seawater. Optical wireless communication offers high data rates but is highly sensitive to turbidity and alignment constraints. Existing systems are mainly optimized for blue-green wavelengths, limiting performance in low-visibility environments.

This work proposes an adaptive near-infrared (NIR) laser-based underwater optical communication system for short- to medium-range applications (10–150 m) in turbid conditions. The system integrates a compact hemispherical photonic module with multiple emitters and receivers, enabling quasi-omnidirectional communication. Adaptive control algorithms dynamically adjust power and modulation based on environmental feedback.

The system targets data rates above 1 Mbps with BER ≤ 10⁻⁵ under varying conditions. Expected results include improved link stability and a systematic dataset on NIR propagation in turbid water.

This work contributes to underwater networks, diver communication, and autonomous systems while reducing acoustic impact on marine environments.

Keywords: AI/ML for oceanographic and environmental data analysis, Data fusion, uncertainty quantification, and predictive modeling
Abstract: Accurately characterising the spatial variability of storm surge drivers remains a key challenge for coastal flood forecasting, particularly regions characterised by spatial variability in tidal regime and coastal geometry, such as Ireland. Existing approaches are typically site-specific or rely on computationally intensive hydrodynamic models, limiting their scalability and interpretability at regional scales. In this work, we propose a novel, data-driven framework to identify and compare dominant surge drivers across a national tide gauge network. The approach integrates unsupervised clustering, Long Short-Term Memory (LSTM)–based surge prediction, sensitivity analysis, and explainable artificial intelligence (XAI) to jointly characterise both physical drivers and model behaviour. Tide gauge observations are decomposed into tidal, surge, and sea level rise components and combined with ERA5-derived atmospheric forcing. Location-specific LSTM models are trained using 24-hour input windows to predict surge at a 1-hour lead time, achieving root mean squared errors of 0.10–0.13 m across stations. The novelty of this work lies in the combined use of clustering and XAI to derive a physically interpretable, national-scale classification of surge drivers while validating these relationships through predictive modelling. Results reveal a consistent inverse relationship between surge and atmospheric pressure, alongside spatially varying wind-driven responses. Three distinct clusters of tide gauges are identified, corresponding to differences in tidal dominance, wind directionality, and coastal setting. XAI analysis further highlights non-linear sensitivities to wind stress that are not captured by traditional correlation-based approaches. These findings demonstrate a transferable and interpretable framework for analysing surge dynamics in data-rich coastal environments, with direct applications to regional flood forecasting and coastal hazard assessment.

Keywords: Data fusion, uncertainty quantification, and predictive modeling, Energy-harvesting and long-duration monitoring solutions, AI/ML for oceanographic and environmental data analysis
Abstract: Accurate energy demand prediction is critical for battery-electric ship design, yet small vessels, the most viable electrification candidates, often lack sufficient operational data. This paper proposes a physics-constrained machine learning framework incorporating hydrodynamic constraints (power--speed cubic law, wind and wave resistance monotonicity, minimum hotel load) as loss regularization terms. In an evaluation on the FuelCast benchmark with a coastal vessel possessing only 3 months of data, ablation reveals that, among neural networks, physics-constrained loss is the dominant factor, improving R^2 by 0.076, compared with gains of 0.001 from transfer learning and 0.011 from augmentation. XGBoost achieves the highest overall R^2 of 0.767. Physics constraints narrow the neural network gap but do not fully close it. The converted demand profiles drive a 10-year battery degradation simulation comparing nickel manganese cobalt (NMC) and lithium iron phosphate (LFP) under cold and tropical conditions. Among 48 scenarios, only the 60,000,kWh battery sustains continuous operation, and LFP retains over 97% state-of-health after 10 years in cold climates. Weather-adaptive charging extends NMC state-of-health by 3.7 percentage points under tropical conditions.

Keywords: AI/ML for oceanographic and environmental data analysis, Data fusion, uncertainty quantification, and predictive modeling, Marine life and ecosystems
Abstract: AUVs routinely measure properties (e.g., fluorescence, backscatter) that track phytoplankton biomass, but many ecosystem impacts depend on community composition. In Monterey Bay, bioluminescence-fluorescence proxies exploit the fact that many dinoflagellates are bioluminescent while diatoms are not. These proxies are restricted to nighttime and require a costly bioluminescence sensor. Here, we trained a machine-learning model using gradient-boosted trees to target a dinoflagellate-diatom ratio proxy using low-cost AUV sensor data. The model was developed together with conformalized quantile regression (CQR) and thresholded interval width models used to filter low-trust outputs making it operationally useful. Environmental DNA (eDNA) samples provided an in situ ratio for comparison; the training period subset (2016-2022) informed hyperparameter selection and interval-width threshold calibration. In the training period during nighttime, rank correlation vs eDNA increased from 0.61 (N=94) to 0.66 (N=47) under thresholding while the bioluminescence reference achieved 0.71 and 0.72 on the same samples. In the recent period (2023-2024), nighttime samples were insufficient for bioluminescence comparison, but the combined day/night correlation increased from 0.21 (N=69) to 0.37 (N=33). Applied to an example AUV survey, the model predictions provided daytime estimates where the bioluminescence proxy was unavailable and captured community shifts consistent with co-located eDNA samples while also highlighting cases where proxy labels and eDNA disagree. This work expands proxy coverage using low-cost sensors already deployed fleet-wide, with calibrated prediction intervals enabling selective use of higher-trust predictions for scientific interpretation and real-time onboard applications.

Keywords: Benchmark datasets, reproducibility, and open-source frameworks, AI/ML for oceanographic and environmental data analysis, Data fusion, uncertainty quantification, and predictive modeling
Abstract: Robust perception is essential for the deployment of autonomous underwater robots. However, optical cameras become unreliable under poor illumination and backscatter. Forward looking (2D) acoustic sensors remain effective under these conditions, but they measure range and bearing while leaving elevation unresolved, creating an ambiguity that prevents individual sonar returns from being localized in three dimensional (3D) space. This complicates the sensor use for 3D scene understanding and precise object detection. We introduce uScenes, a multimodal underwater dataset containing synchronized 3D multibeam sonar point clouds and RGB imagery. The dataset contains 110 scenes and 95,834 synchronized observation, representing 277.6 minutes of data collected across multiple field sessions. uScenes establishes a foundation for underwater sensor fusion, cross modal representation learning and 3D scene understanding. Code and datasets are given at https://github.com/era-research-lab/uScenes.

Keywords: Foundation models and large-scale training for marine applications, Benchmark datasets, reproducibility, and open-source frameworks, AI/ML for oceanographic and environmental data analysis
Abstract: Scalable machine learning in challenging underwater environments is strongly limited by the lack of labeled real-world training data. This data is often expensive and laborious to gather, making large-scale real-world data challenging to gather and curate. However, simulated data can help close the gap, enabling many learning-based tasks for underwater perception. In this work, we extend OceanSim, an IsaacSim-based underwater perception simulator, with a Synthetic Data Generation (SDG) pipeline for training models to be used in underwater scenarios. The proposed pipeline enables users to generate large, automatically labeled, photorealistic datasets with configurable scene appearance, structure, and sensor settings. We evaluate the pipeline on a real-world sea urchin detection task and study how different forms of synthetic scene variation affect sim-to-real performance. Based on these experiments, we discuss findings on our results, main limitations of the current pipeline and identify future directions for improving underwater rendering fidelity, scene diversity, and the evaluation of sim-to-real generalization. The open-source code can be found at https://github.com/umfieldrobotics/OceanSim.

Keywords: Foundation models and large-scale training for marine applications
Abstract: Computer vision has become an important tool for automated fish detection and tracking, supporting continuous ecological monitoring and fisheries management without the need for direct human observation underwater. However, underwater fish tracking remains challenging because of low illumination, light absorption and scattering, grayscale imagery, background camouflage, verical bars, turbidity, noise, and partial occlusion. These factors make it difficult to maintain consistent object identities across frames, even when the detector produces reasonable bounding boxes. In this study, underwater videos from the Gulf Fishery-Independent Survey of Habitat and Ecosystem Resources Dataset 2024 (GFISHERED24) were processed for fish detection and multi-object tracking using a YOLOX-based detector and a ByteTrack-based tracking framework. To improve tracking consistency, the standard Kalman Filter (KF) used in ByteTrack was replaced with an Interacting Multiple Model motion (IMM) estimator. Unlike the conventional constant-velocity assumption, the proposed IMM framework incorporates multiple fish movement patterns, including constant velocity, acceleration, deceleration, and turning or maneuvering behavior. In addition, the traditional IoU-based association strategy was replaced with a proposed FishIoU metric, which integrates standard overlap, center-region spatial emphasis, aspect-ratio consistency, area consistency, and center-distance penalty to improve bounding-box association under underwater degradation. Quantitative evaluation on 45 representative video sequences shows that the proposed IMM with FishIoU improves overall tracking performance compared with the baseline KF with standard IoU. The average MOTA increased from 73.75% to 76.50%, while the average false negative rate decreased from 23.90% to 20.20%. The proposed method also reduced average ID switches from 0.95% to 0.60%, indicating improved identity preservation. These results suggest that adaptive motion modeling and fish-specific association can improve underwater multi-fish tracking, particularly in challenging scenarios involving low contrast, occlusion, and irregular fish motion. ByteTrack, while also improving MOTA from 60.60% to 61.10%.

Keywords: Sensor synergy, Data fusion, uncertainty quantification, and predictive modeling, Marine GIS and data fusion
Abstract: Seafloor ecosystems like seaweed beds and coral reefs are vital for carbon sequestration and sustaining biodiversity, but they have been declining rapidly in recent years. Timely conservation policies require frequent, large-scale, and precise mapping to monitor these habitats. To balance survey efficiency with mapping accuracy, we previously proposed the Underwater-Assisted Aerial Clustering (UAAC) method, which integrates aerial and underwater observations. However, its robustness to environmental variability has not been fully established. This study evaluated aerial image clustering, a fundamental component of the UAAC method, to assess its generalizability. Results suggest that water surface waves and shadows introduce uncertainty into mosaic-blended RGB-based clustering. In contrast, average-blended DII-based processing mitigates wave-induced artifacts while effectively capturing underlying benthic distributions, highlighting its applicability across diverse environments.

Keywords: Marine GIS and data fusion
Abstract: Underwater simulation provides a safe and inexpensive method of developing and testing algorithms for unmanned underwater vehicles (UUV) autonomy and perception before full field deployment. However, these simulations are often limited by the availability of simulated environments. Current marine robotics simulation suites offer manual ways to edit or create environments, but they require time, data, or expert knowledge of the environment system. To address these issues, we introduce a novel Unreal Engine 5 level generation pipeline that enables automatic creation of coastal environments for HoloOcean. Our pipeline relies on a user-provided overhead image of a coastal scene. The pipeline then uses the image to generate height map data, as well as automatically select assets and place them in the environment.

Keywords: Oceanographic instrumentation and sensors, Low-cost, scalable oceanographic sensors and platforms, Marine GIS and data fusion
Abstract: This paper introduces a specialized data converter interface for the real-time integration of high-precision oceanographic sensors into institutional web-based GIS platforms. Deployed on the icebreaker Araon, the system achieves three primary engineering objectives: (1) a mechanical sensor mounting assembly using IP68 polycarbonate enclosures and industrial cable glands ensuring fluid-tight reliability against high-velocity flows; (2) an Arduino-based interface utilizing non-blocking tokenization algorithms to parse asynchronous 1-Hz ASCII streams from diverse sensors; and (3) a robust network gateway employing an RS232-to-LAN converter to convert legacy protocols for seamless Web-GIS integration. Field trials in polar sea ice zones demonstrated excellent operational robustness, delivering continuous and reliable data acquisition alongside stable underway observations under demanding environmental conditions.

Keywords: Oceanographic instrumentation and sensors, Data fusion, uncertainty quantification, and predictive modeling, Low-cost, scalable oceanographic sensors and platforms
Abstract: Marine ecosystems are increasingly affected by pollution, climate variability, and human pressures, and interpreting these changes requires observations linked to position. Biologging can collect long-term environmental and behavioral records, but compact loggers cannot obtain continuous GPS fixes underwater, so indirect geolocation is required. Existing methods using twilight timing, temperature, or bathymetry are each individually ambiguous under real marine conditions: light is affected by shading, weather, reflection, and depth, while temperature or bathymetry matching becomes unreliable under weak gradients or noisy measurements. This paper presents an integrated probabilistic geolocation framework for a marine biologger. Without GPS trajectories or prior route information, it combines biologger light, temperature, depth, and tilt records with NOAA OISST sea surface temperature and GEBCO bathymetry. It constructs an ocean-only state space, computes observation likelihoods from twilight, temperature, and depth-bathymetry consistency, and links daily candidates through distance-based transitions that exclude excessive movement and land crossings. The release point is used as an initial prior and recovery information as an optional final prior. Unlike conventional single-cue light-based methods, and building on hidden Markov model geolocation, the framework jointly fuses light, temperature, and depth-bathymetry likelihoods, enforces land-crossing exclusion, and applies a corridor decoder that reports explicit r50 and r95 posterior radii rather than a single per-day coordinate. The framework was evaluated on vessel-mounted biologger data from a coastal field test from Geoje Island to Mukho Port via Yeongil Bay, with estimated positions compared against a GPS ground-truth track only after estimation. The estimated trajectory reproduced the overall coastal movement, with a mean error of 60.9 km at five known port calls and 37.5 km for the daily estimates and a mean posterior r95 of 43.0 km, showing that biologger records can be converted into daily positions when continuous GPS tracking is unavailable.

Keywords: Airborne and satellite radar and SAR, Passive observing sensors, Oceanographic instrumentation and sensors
Abstract: Sea-surface information is important for ocean observation and safe water-landing-type UAV operation, but onboard sensors are constrained by payload mass, power, and space. This study investigates a lightweight passive sensing framework: camera image analysis and GNSS reflectometry (GNSS-R). Camera images are used to estimate wave direction and dominant period, while GNSS-R retrieves wave parameters from reflected carrier phase using a physics-based rough-surface scattering model. A modified Beckmann-Spizzichino model was implemented to coherently integrate reflected complex fields and simulate time-differenced carrier phase (TDCP). At the fixed GNSS-R station, GNSS-R-derived wave height showed a correlation coefficient of 0.77 with a radio-altimeter reference, although elevation-angle-dependent sensitivity was observed. UAV experiments at 80 and 130 m altitude also showed flight-heading-dependent carrier-phase tracking, providing a candidate wave-propagation direction, with preliminary period and wave-height estimates of 6.95–7.90 s and 0.70 m. These results demonstrate the feasibility of carrier-phase GNSS-R within a compact passive UAV-borne ocean-wave observation framework that also incorporates camera-based wave analysis.

Keywords: Coastal radars, Oceanography: physical, geological, chemical, biological, Current measurement technology
Abstract: Small-scale eddies detected by phased-array radar at high spatio-temporal resolution Thanh Huyen Tran, Roberto Gomez, Malcolm Heron, and Herman Peters Keywords: HF Radar, surface currents, eddies, shear. turbulence The use of HF radars has become widespread in coastal ocean applications for mapping surface currents over scales of hundreds of km, with time resolution of typically one hour and spatial resolution of a few km. Some applications exceed these general scales which are used for archiving data, and applications such as tsunami observations and port operations have been made. The present poster introduces an application of a VHF phased-array ocean radar system in a constricted estuary which includes a major shipping channel. Eddies in the Western Scheldt estuary are generated by the high tidal flows in deep channels adjacent to emerging sandbanks. The dynamic sandbanks make numerical modelling quite challenging. An eddy detection procedure is described which can run in near real-time. The procedure requires the high spatial resolution achieved with the phased-array VHF radar to accurately detect shear and vortices in the time-scale of minutes. Eddy detection methods applied to the surface current 2D field include (a) Okubo-Weiss criteria, (b) CURL operator, and (c) localised zero speed zones. The study in the Scheldt Estuary shows that the interplay of tides, winds and dynamic bathymetry play a dominant role in eddy formation. Eddies in the estuary fall into two groups. Group 1 is small eddies that form mainly during the outgoing tide. Group 2 covers the larger eddies, which occupied the full width of the river including the shipping channel, formed during the incoming tide when currents were strongest around high tide. On a monthly basis, eddy’s lifetime is prolonged in the presence of strong winds and eddies are detected more frequently during spring tide. The next steps in this work are to identify precursors to the formation of the large eddies with the aim of making early predictions of occurrence. The VHF phased-array radar can effective

Keywords: Coastal radars, Operational observation, Air / sea interaction
Abstract: Coastal marine waters, including open coasts, fjord systems, and estuaries can challenge area-extensive monitoring efforts and accurate numerical model forecasts. Complex hydrodynamics in these regions are a result of strong forcing by tides, freshwater outflow, and regional weather events, produce strong gradients in currents and stratification. These, in turn directly affect transport and mixing of nutrients and pollution, transport and exchange of oxygenated waters, and navigation and sensing in these regions. Yet, dense and persistent sampling of these phenomena needed for numerical model improvement, validation, initialization, and data assimilation from in situ sampling methods is prohibitive. Surveys from crewed and autonomous vessels are expensive and performed infrequently, drifting sensors are not able to maintain persistence in these dynamic flows, and logistics and costs to maintain large arrays of moored sensors is prohibitive. Many of the physical fluid processes along coasts and in estuaries produce ubiquitous and identifiable features. Fronts associated with the mixing of different water masses and river outflow, internal waves and solitons, surface gravity waves and breaking, and surface currents all have signatures in visible, thermal, and radar modalities, and remote sensing retrieval methods for quantifying these features from these sensors are already available and relatively robust. Thus, remote sensing can play an important role in providing persistent and area-extensive observations that, along with judicious in situ sampling, are needed for numerical model advancement.

Motivated by the need for observations for feeding to forecast modeling and general monitoring, we demonstrate coherent X-band radar as a remote sensing technique to observe surface roughness features, currents, and waves in an estuary. This effort is part of the Pacific Infrastructure Supporting Continuous Engineering and Science (PISCES) project, focused on improving forecasting efforts for autonomous marine vehicle operation. Our main goal for this part is to develop a continuous remote sensing-based monitoring system to gather geophysical data (currents, fronts, and waves) for model input and validation of the reg

Keywords: Data visualization, Strategies to broaden participation and promote inclusivity, Hydrography / Seafloor mapping / geodesy
Abstract: Oceanographic institutions around the world use the open source MB-System software to process underwater mapping data to produce maps and other useful products. Undergraduate computer science students at the California State University Monterey Bay have contributed to MB-System via their senior capstone class. The resulting tools: mbgrd2gltf and mbmesh (in development) convert seafloor measurements into 3D graphical formats that can be efficiently loaded and visualized in modern web browsers. A broad set of skills are required, including C++, 3D computer graphics, data manipulation, source code control, and most importantly good teamwork and communication.

These have been challenging yet appropriate software development assignments for the students. Proper scoping by the mentor of 16-week semester projects helps the teams of usually 4 students achieve some level of success at the end of the semester. The mentor also provides guidance and direction during the semester with regular meetings. Over the past six years, approximately 50 students from CSUMB’s Capstone classes have contributed to this project. Every student pursuing a bachelor’s degree in computer science at CSUMB completes a senior capstone course designed to support a smooth transition from college to career. Through real-world projects mentored by industry professionals, students gain practical experience and confidence. The collaboration between MBARI and CSUMB’s Computer Science program has created meaningful and enriching learning opportunities, fostering both the professional and personal growth of participating students.

This poster will showcase tools that the students have created to generate standards-based glTF (.glb) files in OGC 3D Tiles data structures that are optimized for viewing in web browsers using the X3DOM javascript library. These capabilities allow the global oceanographic community to more easily document and share important deep sea discoveries in a format that is both engaging and informative.

Keywords: Data visualization, Data assimilation, Meteorology
Abstract: Hurricanes represent a major hazard to Florida’s coastal systems, where storm surge and heavy precipitation are the primary drivers of coastal flooding and erosion. In October 2024, Hurricane Milton made landfall on Florida’s west coast as a late-season major hurricane, producing extensive storm surge, heavy rainfall, and inland flooding across the region. Although Milton weakened to a Category 3 hurricane at landfall, its impacts were amplified by antecedent conditions, including Hurricane Helene, which affected the region two weeks earlier. This study assesses the combined effects of storm surge and precipitation associated with Hurricane Milton, with emphasis on land–ocean interactions that influence coastal flooding and vulnerability along Florida’s west coast. The analysis integrates ERA5 surface-level reanalysis data with satellite imagery, surface observations, and official documentation from the National Hurricane Center and the National Oceanic and Atmospheric Administration. Key atmospheric and oceanic variables examined include mean sea-level pressure, surface wind fields, wind-wave swell, and accumulated precipitation. Results show that persistent onshore winds and reduced surface pressure enhanced storm surge along Florida’s shallow continental shelf, facilitating inland penetration of surge waters. Elevated precipitation totals further compounded flooding impacts, particularly in urbanized areas where impervious land surfaces limited infiltration and increased surface runoff. The interaction between storm surge and extreme rainfall resulted in prolonged coastal inundation and river flooding, highlighting the critical role of coastal bathymetry, topography, and land cover in modulating hurricane impacts. These findings demonstrate how land–ocean coupling processes can significantly amplify coastal flooding during late-season hurricanes. Improved understanding of these interactions is essential for hazard assessment, coastal planning, and mitigation strategies in vulnerable, low-lying coastal regions.

Keywords: Data visualization, Benchmark datasets, reproducibility, and open-source frameworks, Remote sensing and field campaigns in polar regions
Abstract: Abstract—The process of vertical advective flux of Salinity (VAFS) is due to a strict movement of mass via a salt gradient between the upper and lower layers from surface to deep. This paper explores a change in VAFS in the Norton Sound in 2014 as compared to 2017. I examine the differences in interannual-scale changes in both the sea surface state and in river discharge in-situ, remote sensing, inverse modelling, reanalysis datasets to suggest a possible mechanism for this change vertical advection.

Keywords: Data fusion, uncertainty quantification, and predictive modeling, Data visualization, Marine life and ecosystems
Abstract: Corals serve as essential contributors and fundamental engineers of Earth’s tropical ecosystems worldwide. The intrinsic symbiotic relationship with zooxanthellae, along with their complex communal biological interactions allow corals to function as keystone species providing necessary sustenance and structural frameworks. Experimental and observational studies done on modern corals suggest that they are highly vulnerable to acidification and warming ensued by the ocean’s rampant absorption of atmospheric carbon dioxide (CO2) along with indications of biological traits being strongly correlated with susceptibility to environmental disturbance; however, since these studies are short-term it is uncertain which traits indicate greater extinction risk. The fossil record can preserve past warming and acidification events to demonstrate how crises filter coral lineages and reveal which combination of traits consistently survive and which are more prone to extinction. Earth’s climate and oceanic pH have fluctuated over geologic time (Patra & Punekar, 2025) with specific hyperthermal intervals appearing more prominent due to their magnitude of change. During the Paleocene-Eocene Thermal Maximum (PETM), rapid carbon release initiated sharp global warming and pH decline, with widespread carbonate dissolution in marine sediments (Babila et al., 2018; Zachos et al., 2005). In contrast, the mid-Cretaceous greenhouse featured persistently warm conditions, little polar ice and broad shallow seas which supported high marine productivity and extensive reefs (Jones et al., 2022). Across these climate perturbations, coral assemblages exhibited repeatable shifts in composition and traits. Together, these traits provide a systematized link between elevated temperature and reduced pH observed in the fossil record. The Ancient Reef Traits database (ARTD) (Raja, N.B., et al., 2022) and Paleobiology Database (PBDB) function as valuable resources to locate these trait patterns. To investigate the resilience and survival of corals, we intend to employ logistic regression analysis to create a quantifiable framework between each trait and extinction risk, ultimately allowing for promising mitigation of contemporary coral extinction

Keywords: E - M sensing, Numerical modeling and simulation, Data fusion, uncertainty quantification, and predictive modeling
Abstract: Adding complexity to an equivalent-source model lowers the ship-magnetic calibration residual without necessarily adding physical information. This paper separates calibration, prediction, and identifiability in a controlled synthetic benchmark with known truth. Five primary model families, each with five sources, are compared: point dipoles (M1), free real multipoles through order two (M2) and order three (M3), a ridge-regularized order-three model (M3R), and exact uniformly magnetized finite ellipsoids (M4). All unregularized families calibrate between 7.83 and 8.05 percent, around the 8.00 percent truth-reference residual, so the training residual does not rank them. Held-out prediction does: M1 and M4 remain near 10.4 percent median leave-one-run-out error and 6 percent on a severe inward holdout, whereas M3 reaches 213 percent. A fixedgrid 25-dipole control with the same linear capacity as M3 fails inward as well, which places the instability with flexible, ill-conditioned representations rather than with high multipole order. One longitudinal gauge and one confocal shape null per ellipsoid leave 32 locally observable directions out of 38. Across 20 paired acquisition realizations, coherent navigation error displaces recovered M4 centers by a mean of 3.34 m with a standard deviation of 1.29 m, compared with a mean of 0.32 m and a standard deviation of 0.12 m for independent magnetic noise. Stable predicted fields therefore coexist with non-unique and acquisition-biased source geometry.

Keywords: Marine GIS and data fusion, Hydrography / Seafloor mapping / geodesy, Numerical modeling and simulation
Abstract: The seafloor plays a critical role in ocean ecosystem function, yet it remains one of the least consistently characterized parts of the marine environment. Ecological Benthic Units (EBUs) were developed to address this gap by providing a globally consistent, data‑driven way to describe benthic environments using integrated geospatial data. Rather than focusing on a single variable, the EBU approach brings together seafloor bathymetry, geomorphology, sediment properties, and key oceanographic conditions and uses unsupervised clustering to identify benthic regions that share similar ecological characteristics. By combining these datasets into a unified classification framework, EBUs offer a practical way to explore how physical structure and environmental drivers shape benthic habitats across different ocean regions. The project is designed to be collaborative and flexible, allowing new data and methods to be incorporated over time. EBUs are made accessible through interactive mapping applications and open data services, enabling researchers and resource managers to explore seafloor patterns, compare regions, and better understand benthic variability at both regional and global scales. Ecological Benthic Units provide a foundation for more holistic seafloor analysis, supporting applications ranging from marine research to conservation planning and spatial decision‑making. By translating complex, multidimensional seafloor data into intuitive ecological units, EBUs help bridge the gap between data availability and meaningful interpretation of benthic systems.

Keywords: Ocean color and hyperspectral measurements, Marine life and ecosystems, Pollution monitoring
Abstract: By nature of its geostationary orbit, the Geostationary Littoral Imaging and Monitoring Radiometer (GLIMR) is NASA’s first ocean color sensor designed to resolve short term biogeochemical processes in the surface ocean. Such processes include physiological changes in phytoplankton communities, vertical migration of phytoplankton and zooplankton, horizontal and vertical fluxes of land and ocean materials, and a host of other processes occurring at diurnal or semidiurnal tidal frequencies.

This presentation will describe an overview of the mission scheduled to launch in the 2027-2028 time frame, and relevant features of the GLIMR instrument. We will highlight how it is ideally suited to observe biogeochemical and physical processes in the surface ocean at hourly timescales and include a brief overview of its capabilities to observe phytoplankton dynamics, land-ocean interactions and hazards (e.g., oil spill, harmful algal blooms).

Keywords: Ocean color and hyperspectral measurements, Sensor synergy, Ocean stewardship, conservation strategies, and climate resilience
Abstract: Here we present a scheme, and its evaluation, for the tractable monitoring and verification of marine carbon dioxide removal. The proposed framework leverages current remote sensing platforms and evaluates three potential particle tracers, based on commercially available materials (quartz sand, recycled glass, hollow glass microspheres), that are observable across satellite, autonomous float, and in situ optical imaging platforms (PACE, BGC-Argo, UVP5), providing empirical constraints on both physical and biological export across spatial and temporal scales. Forward Mie scattering models evaluated material-size combinations across the three platforms, identifying feasible detection regimes. Candidate materials were selected to span trade-offs in optical properties, cost, and environmental impact, key criteria for scalable ocean carbon dioxide removal.

Keywords: AI/ML for oceanographic and environmental data analysis, Geoacoustic inversion
Abstract: One deep learning application in ocean acoustics is to obtain properties of the seabed. To this end, a one-dimensional regression convolutional neural network (1D CNN) is designed that uses relative sound levels from a towed seismic source as a function of range as input. The CNN learns the labels associated with acoustical parameters of the seabed. These parameters include sediment sound speed, sediment density, acoustic attenuation, and sediment layer thickness. Because access to large amounts of labeled data is limited, synthetic training data from a variety of seabed properties generated by a range-independent normal mode model for sound propagation are used to train the 1D CNN. The hyperparameters of the 1D CNN and the synthetic training data specifications are optimized in a series of studies. Hyperparameters varied include number of convolutional layers, inclusion of batch normalization and maxpooling, normalization of data and labels, and data augmentation with noise during training. Variations in synthetic data include offsets in water depth and source-receiver geometry and changes to the water sound speed profile. Additionally, the accuracy and efficiency of the 1D CNN in estimating parameters jointly are compared to those when estimating parameters individually. It is shown that each seabed parameter is affected differently by differences in the network hyperparameters and training data. This example illustrates the many choices that must be made when designing a deep learning application for ocean acoustics. [Work supported by Office of Naval Research]

Keywords: AI/ML for oceanographic and environmental data analysis, Benchmark datasets, reproducibility, and open-source frameworks
Abstract: In recent years, the need for monitoring the open ocean has increased significantly. The use of thermal payloads on remotely piloted aircraft has greatly expanded our sensing capabilities, enabling us to monitor vast stretches of ocean. We are leveraging a dataset of aerial thermal videos to develop a machine learning-driven solution for detecting whales in the vicinity of ships. The system focuses on identifying whales and their associated "fluke prints," which are the surface upwelling events generated by the whales’ swimming behavior, visible as smooth patches on the water surface in the visible spectrum and readily detected in the IR spectrum [1-3]. This tool could be used by ships as an early-warning system to minimize the risk of collisions and other harmful interactions with marine wildlife. This work advances the state of the art by (i) benchmarking baseline machine learning architectures against a novel dataset; (ii) diagnosing dataset deficiencies through novel architectures to further improve system accuracy. Preliminary findings are presented here, while updated results will be shared at the conference.

Keywords: Low-cost, scalable oceanographic sensors and platforms, Oceanographic instrumentation and sensors
Abstract: We developed two compact, scalable, and low-cost submersible sensors for measuring dissolved oxygen (DO) and carbon dioxide (pCO2) to address the need for accessible in situ measurements of these parameters. Significant spatial and temporal variability in these climate-relevant gases demands continuous, high-resolution observations to characterize dynamic ocean processes. The two sensors developed here integrate commercial off-the-shelf components with a passive gas extraction system, are compact, depth-rated to 2000 m, cost under 500 and can be integrated onto a wide range of deployment platforms. By reducing cost and complexity compared to existing technologies, these sensors enable broader access to ocean data collection. The sensor design, laboratory validation, and field demonstration are described.

Keywords: Low-cost, scalable oceanographic sensors and platforms, MEMS-based sensing technologies and miniaturization, Oceanographic instrumentation and sensors
Abstract: The ocean plays a central role in regulating Earth’s climate, sustaining biodiversity, and driving global biogeochemical cycles. Accurate, long-term measurements of ocean properties are essential for understanding these processes. However, existing oceanographic sensors remain constrained by their large size, high power consumption, and their reliance on wired connections.

This paper addresses these limitations through the development of a centimeter-scale, microfabricated marine sensing platform that combines multiple sensing modalities and utilizes wireless power and communication. The proposed device integrates up to seven key oceanographic sensors onto a single 1 cm² microchip. The microchip is oxide encapsulated to withstand pressures up to a target depth rating of 2000 meters. By eliminating bulky packaging and wired interfaces, this approach enables compact, low-power sensing nodes suitable for dense, distributed ocean monitoring networks.

We have fabricated the first version of the proposed device which includes temperature, conductivity, pH, turbidity, fluorescence, and irradiance sensors. A wireless power and communications system has been developed and verified in the lab for operation in saltwater. The temperature and conductivity sensors have been verified up to 1000 psi and across their target ranges. Additionally, the microchip remained operational after being subjected to pressures up to 3000 psi, or approximately 2000 meters depth, the operational depth of core biogeochemical Argo Floats. Our results prove that this method of sensor development is a highly feasible advancement for ocean sensing.

This work enables scalable, repeatable production of marine sensors at significantly lower cost than traditional systems. The relevance to the OCEANS community lies in the transformational potential of this approach towards how ocean data is collected. By leveraging semiconductor manufacturing techniques, this research is advancing a new paradigm for distributed, low-cost, and minimally invasive sensing platforms. The presented device lays the foundation for a new generation of scalable ocean observing systems capable of supporting long-term environmental monitoring.

Keywords: Oceanographic instrumentation and sensors
Abstract: We present three novel approaches for turbidity estimation using continuous wave (CW) lidar. The first uses the distance to the center of the volumetric backscatter, measured at 5 MHz with a custom Software-Defined Radio (SDR) based CW lidar. This approach achieved a mean absolute percent error (MAPE) of 1.715%. The second approach fits the amplitude and phase response of the volumetric backscatter to an a priori model using Nonlinear Least Squares (NLS) and achieved a MAPE of 2.368%. The third approach builds on the second by adding the ability to mitigate the effects of partial sample-volume shadowing by an object in the water. This is done using a model of the expected object return, allowing NLS to solve the likely backscatter and object return simultaneously. When the object power is less than the backscatter power, this approach achieved a MAPE of 1.504%. All results are compared to turbidity measurements from a Seabird AC-S operating over a turbidity range from 0.645 to 4.211 m⁻¹

Keywords: Low-cost, scalable oceanographic sensors and platforms, Open-source hardware, software, and community innovation, Citizen-science networks and participatory ocean observations
Abstract: Profiling floats are essential for global ocean observation. Models like the Argo float from Teledyne Webb Research use proprietary hardware and firmware developed by large institutions, collecting temperature and pressure profiles sent via satellite for climate and weather models. Autonomous profiling floats can also verify whether marine carbon dioxide removal (mCDR), using ocean processes to capture and store atmospheric CO2, works as intended. MRV (Monitoring, Reporting, and Verification) enables independent checks on mCDR data but remains in early field testing. Commercial float firmware, which controls movement, data collection, and transmission, remains inaccessible to students and researchers in the space. NanoFloat 2.0, from the University of Washington's Underwater Remotely Operated Vehicles (UWROV) team, tackles this issue. Its custom firmware, nanOS, delivers core profiling float functions, depth control, data collection, and wireless telemetry, on affordable, accessible hardware, specifically for dynamic, shallow, coastal areas. By open-sourcing firmware and hardware, NanoFloat 2.0 broadens access to oceanographic tools and provides a flexible, low-cost platform for MRV in emerging mCDR applications, especially in shallow coastal areas. nanOS runs on the Seeed Studio XIAO ESP32-C6 microcontroller, written in C++ for the Arduino/ESP-IDF framework. It executes a single closed main loop, keeping behavior predictable and easy to debug. Before a dive, the ESP32-C6 syncs to UTC with NTP (Network Time Protocol) and transmits one data packet. A Proportional-Integral (PI) depth controller adjusts the buoyancy-control piston, with sensor data as ground truth verification. Oceanographic data is saved to onboard flash using LittleFS, then transmitted over RH_RF95 radio upon surfacing, with fallback to a HTTP server if radio is unavailable. UWROV plans to expand NanoFloat 2.0’s payload to include CO2 and other biogeochemical (BGC) sensors, enabling direct measurement of carbon removal activity in the water column for verification.

Keywords: Low-cost, scalable oceanographic sensors and platforms, Rapid-deployment systems and modular instrumentation, Seafloor engineering
Abstract: This paper presents a modular, autonomous, underwater sensor network designed for long-duration, submerged water quality monitoring. Unlike existing low-cost marine sensor platforms, which typically rely on a surface buoy or shore-tethered gateway for power and data relay, the system presented here operates fully submerged and independent, requiring no surface presence and no external infrastructure beyond simple per-node anchors. The network consists of a master node, power node, and an arbitrary number of sensor nodes connected via an RS-485 daisy-chain bus, enabling single-diver deployment and retrieval with complete modularity. The system functions identically whether configured with one sensor node or several. Each sensor node integrates Atlas Scientific probes to measure temperature, electrical conductivity, pH, and dissolved oxygen, covering several GOOS Essential Ocean Variables. The system has been validated through benchtop endurance testing, a five-day field deployment in Lake Tahoe, and multiple shorter deployments to assess data quality, successfully completing dozens of autonomous collection events. This work is intended to support monitoring of the Palos Verdes artificial reef restoration site in partnership with the Vantuna Research Group at Occidental College, and to provide a data point on the long-term field accuracy of low-cost sensor probes as a monitoring solution.

Keywords: Systems and observatories, Pollution monitoring, Instrumentation, sensors and autonomous systems for sea-based aquaculture
Abstract: Coastal areas are dynamic, vulnerable, and valuable. Major population centers and 2.4 billion people exist within 100km of coasts worldwide. Recording high-frequency coastal environmental data, especially during transient meteorological events, has clear safety and conservation value. However, this is challenging to achieve non-invasively and cost-effectively. Bennett et al. propose a modular cable-mounted robot to collect data repeatably and autonomously. Enhancing this design with protection from a low-drag fairing extends the operating envelope of this system to include extreme events such as hurricanes, where monitoring data is especially valuable for developing coastal meteorological models. For fairing design, biomimetic inspiration comes from species endemic to the turbulent water of the coastal surf zone and swash zone: the California corbina fish, Menticirrhus undulatus, the Atlantic mole crab, Emerita talpoida, and the Atlantic surf clam, Spisula solidissima. Fairing models were designed and constructed from the fundamental geometries of the three species and scaled for dynamic similarity. A towing tank test was developed to simulate flow currents or waves. A drop test was developed to simulate wave impact. Towing tank results indicated that the mole crab-based design contributes significant drag reduction. The drop test served as a proof of concept to aid in future development. CFD analysis provided a qualitative visualization of pressure gradient surrounding the models. This effort develops technology that improves our understanding and stewardship of ocean natural resources.

Keywords: Machine-learning classification of acoustic signals, Passive and active acoustic monitoring, Distributed hydrophone arrays and networked acoustic systems
Abstract: Ship noise spectrograms are useful tools for visualizing the frequency content of passing ships. These spectrograms are often used to train machine learning models, but a lack of real-world labeled data often leaves a need for synthetic training datasets. This paper explores a method to estimate ship signatures from recorded spectrograms and model them in different shallow-ocean environments as a form of data augmentation. In shallow water, the ocean environment influences the measured spectrograms because different seabeds cause different amounts of loss in the signal as it propagates from the ship to a receiver. Varying the environment changes the transmission loss (TL) and the multipath interference patterns seen in the spectrogram. The TL can be approximated by modeling sound propagation in an assumed environment, and the basic sonar equation suggests that adding the TL back into the recorded spectrogram provides an estimate of the source levels of a ship. This method is successful when the seabed is modeled accurately and the interference patterns of the TL align with those of the recorded spectrogram. However, unwanted features are produced in the estimated source level spectrogram when there is modeling mismatch or ambient noise. Averaging estimated ship levels over time provides a way to minimize these unwanted anomalies while preserving tonal characteristics of the ship spectrum. Once the time-averaged ship spectrum is estimated, the TL of different environments can be subtracted to estimate the spectrogram of that ship in a different part of the world. This technique has the potential to create augmented datasets to train machine learning signal classification models, but further work is needed with more ship data to validate the process.

Keywords: Machine-learning classification of acoustic signals, Passive and active acoustic monitoring, Foundation models and large-scale training for marine applications
Abstract: Monterey Bay National Marine Sanctuary is an important habitat for three ecotypes of killer whales (Orcinus orca): Transient (Bigg's), endangered Southern Resident, and Offshore orcas. Routine automated detection of orcas is being developed for continuous passive acoustic monitoring via a broadband hydrophone deployed at ~900 m depth on the MARS cabled observatory. Manual annotation of the resulting thousands of hours of recordings is impractical, and acoustic training data from the MARS hydrophone for these ecotypes are scarce.

We evaluated two existing CNN models -- the Google multi-species whale model and OrcAI -- against Pacific Sound MARS recordings. While both yielded high-confidence orca detections, both produced significant confusion between orca calls, Pacific white-sided dolphin calls (Lagenorhynchus obliquidens), and other signals, limiting their operational utility.

Here we report the application of the Perch 2.0 bioacoustics foundation model with agile modeling (active learning) via the Perch-Hoplite vector database system. Using a small set of labeled audio files of orca calls, dolphin calls, and other signals, we perform transfer learning on fixed embeddings, enabling efficient one-shot retrieval through vector search. Acoustic detections are compared against visual sighting records from the California Killer Whale Project and others.

This work demonstrates how a bioacoustics foundation model, combined with a small labeled dataset and active learning, can produce a custom classifier efficiently -- requiring minimal researcher time and computational resources -- while supporting ecological assessment and protection of marine mammal populations.

Keywords: Low-cost, scalable oceanographic sensors and platforms, Foundation models and large-scale training for marine applications, Passive and active acoustic monitoring
Abstract: Floating offshore wind (FOSW) systems introduce complex interactions between mooring infrastructure, marine life, and environmental forces, creating risk of entanglement events that are difficult to detect and characterize in real time. Existing monitoring approaches are limited by sparse sensing, high data transmission costs, and an inability to distinguish entanglement signals from background ocean variability.

This work presents SEA-Scan, a modular sensing and analytics framework that integrates surface and subsurface sensor networks, low-power edge processing, machine learning (ML), and a physics-based digital twin to enable continuous, real-time entanglement detection. The system leverages Sofar Ocean’s Spotter platform and distributed subsea motion sensing modules to capture multi-modal environmental and structural data. A novel edge intelligence module performs onboard data compression and inference, reducing transmission requirements while preserving critical signals. ML modeling (led by Stanford University), including adversarial learning approaches, are used to detect and classify entanglement events within noisy datasets, while a subsea digital twin provides contextual understanding of structural behavior and potential damage mechanisms.

SEA-Scan advances ocean observing and offshore energy monitoring by enabling actionable, real-time insights for operators, regulators, and environmental stakeholders. This work – which is currently in early laboratory testing stages – contributes to the OCEANS community by demonstrating a pathway toward integrated, cost-effective monitoring systems that support both infrastructure integrity and marine ecosystem protection.

Keywords: Fiber-optic distributed sensing (DAS, DTS, DSS), Passive and active acoustic monitoring, Calibration of acoustic systems and metrology
Abstract: Distributed Acoustic Sensing (DAS) is a novel tool in underwater acoustics that measures apparent strain in an optical fiber from the phase change of backscattered coherent light to estimate acoustic pressure. This technology transforms fiber-optic cables into dense, synchronized arrays of thousands of acoustic sensors, offering unique opportunities for marine research. In contrast to conventional hydrophone arrays, DAS systems provide continuous spatial sampling over tens of kilometers with meter-scale resolution, enabling the observation of acoustic fields with unprecedented spatial coverage. However, the directivity and frequency response of DAS measurements remain poorly understood particularly in complex environments where propagation effects and cable-seafloor coupling may significantly influence the recorded signal. An experiment was conducted in October 2025 to characterize the DAS acoustic pressure sensitivity of the GeoLab cable off the southern coast of Madeira Island, Portugal. A controlled acoustic source experiment was designed to quantify both the amplitude response and the angular sensitivity of the DAS system under realistic oceanographic conditions. A sound source (VC2C, Lubell) was deployed from R/V Discovery at 14 m depth, with a reference hydrophone suspended 1 m below the source. Data acquisition covered three short towed transects and six stationary stations at 100–3000 m ocean depths. This configuration enabled sampling over a range of source-receiver geometries and grazing angles. At each station, the source repeatedly transmitted a signal comprising chirps in the 105–160 Hz band and 11 tonals between 55–195 Hz. The processing of the raw optical phase data involved converting it to strain rate using the standard conversion values, followed by linear detrending, and mean subtraction. Next, the DAS data were integrated in frequency domain to convert from strain rate to strain units. Results of the measurements of the frequency and angular dependence of DAS pressure sensitivity and signal-to-noise ratio will be presented. [Work supported by NSF and NERC.]

Keywords: Fiber-optic distributed sensing (DAS, DTS, DSS), Bioacoustics, Passive and active acoustic monitoring
Abstract: Distributed Acoustic Sensing (DAS) converts underwater fiber-optic cables into dense sensor arrays for marine monitoring, including ship and baleen whale detection. Detection performance is governed by the gauge length (GL), an acquisition setting set as a trade-off between optical signal-to-noise ratio (SNR) and bandwidth. Longer GLs improve optical SNR by integrating more backscattered light into strain measurements, at the cost of reduced frequency response.

Here, we investigate the effect of the GL on strain amplitude, SNR, and noise floor using data collected on a telecommunication cable crossing the St. Lawrence Estuary in Quebec, Canada with a FOSINA DxS interrogator during controlled-source playbacks. Raw optical data were processed with GLs of 5 to 20 m. Analysis of a playback tone at 100 Hz shows that increasing the GL reduced measured noise levels, with a GL of 10 m providing the best SNR overall, generally consistent with theoretical predictions. The estimated pressure-to-strain conversion factor increased systematically with GL by 0.76-0.77 dB/m, indicating a potential GL dependence not fully captured by the theoretical DAS response model considered here. The findings of this work provide an empirical basis for GL selection for monitoring whale calls near 100 Hz, while demonstrating that optimal configuration remains considerably influenced by the source-to-cable geometry.

Keywords: Fiber-optic distributed sensing (DAS, DTS, DSS), AI/ML for oceanographic and environmental data analysis, Classification and pattern recognition (parametric and non-parametric)
Abstract: Whale call detection is an important task of monitoring marine mammals. One approach to addressing whale call detection is passive acoustic monitoring using hydrophones. Whale call detection using hydrophones has been widely studied for many years.

More recently, distributed acoustic sensing (DAS), which is one of the fiber-optic sensing technologies, has been developed. Unlike hydrophones, few studies have investigated whale call detection using DAS. This is because the amount of open-source DAS data is extremely limited compared to that of hydrophones. This challenge hinders the development of machine learning models for DAS-based whale call detection, thereby limiting the use of DAS for large-scale whale monitoring in the ocean.

In this paper, we propose a training-less knowledge transfer from a hydrophone to DAS for whale call detection. In the proposed method, the domain gap between DAS and hydrophone is addressed without re-training of DAS data.

Keywords: Fiber-optic distributed sensing (DAS, DTS, DSS), Applications in geohazards, offshore energy, carbon monitoring, and environmental observation, Integration with autonomous platforms and observatories
Abstract: Monitoring dynamic seafloor processes remains challenging because most marine instruments record at discrete locations for limited periods, while permanent observatories are sparse and geographically restricted. Distributed Acoustic Sensing (DAS) provides dense spatial sampling of seismic and acoustic energy along optical fibres, but marine applications commonly rely on existing telecommunication cables, which rarely cross sites of greatest geomorphic activity. As a result, many important environments -including submarine canyons, active slopes, transform faults, hydrothermal systems, and recently disturbed seabeds - remain difficult to monitor with spatially distributed sensing.

Here we present Geo-Sense, a portable, autonomous DAS system designed for rapid, targeted seafloor deployment without pre-existing cable infrastructure. The system integrates a low-power interrogator, pressure-rated battery housings, and a 1 km sensing cable. In February 2025, Geo-Sense was deployed for eight days at 380 m water depth on the flank of Monterey Canyon, California. The system detected all catalogued earthquakes within 150 km, as well as additional uncatalogued seismic and hydroacoustic events. Broadband amplitudes in the 4-30 Hz band correlated with near-bottom tidal current velocities measured by a colocated Acoustic Doppler Current Profiler (ADCP), demonstrating sensitivity to oceanographic forcing.

Building on this fixed-site validation, we focus here on two recent trials designed to make Geo-Sense more flexible and easier to deploy using robotic ocean vehicles. In February 2026, a thinner, lighter 1 km fibre-optic cable was deployed by Remotely Operated Vehicle (ROV), improving handling and enabling targeted cable placement across selected seafloor features. In April 2026, a 100 m length of bare fibre was deployed from a Long-Range Autonomous Underwater Vehicle (LRAUV), providing proof of concept for mobile DAS sensing and short-aperture measurements in complex terrain.

These results show that portable DAS can move beyond opportunistic use of existing seafloor cables toward a deployable, repositionable observing system.

Keywords: Passive and active acoustic monitoring, Systems and observatories, Buoy Technology
Abstract: This study aims to develop an underwater target localizing buoy system and to evaluate its operational feasibility for sound source localization. The system includes the buoy platform, multi-channel data acquisition module, onboard signal processing, and satellite communication for real-time data transmission to the onshore station, where further processing is conducted for target localization.At the onshore station, a passive time-reversal mirror (PTRM) method combined with ray theory is applied to obtain the depth and range of an underwater target every 10 mins. The PTRM method utilizes the time-reversed the received signals and back-propagating them through an acoustic propagation algorithm, allowing the acoustic energy to refocus at the original source location and thereby enabling source localization.The field experiment was conducted in the southwestern offshore waters of Xiaoliuqiu, Taiwan,which has a water depth of approximately 600 m and represents a typical shallow-to-shelf-edge environment with frequent vessel activities and complex acoustic conditions,making it suitable for system validation.The results show that the estimated target depths are primarily concentrated between 90 m and 130 m, indicating a clear clustering of acoustic sources within this depth range. In terms of horizontal distribution, two dominant clusters are observed at ranges of 1000–1200 m and 1900–2000 m from the buoy,suggesting consistent spatial acoustic activities.The observed distribution patterns show good agreement with fishery resource survey data reported by the Fisheries Research Institute of the Council of Agriculture, Taiwan,supporting the physical validity of the localization results. Overall, the system demonstrates stable localization performance under real ocean conditions, confirming its potential for long-term deployment and practical applications in underwater acoustic monitoring and target localization.

Keywords: Ocean noise, Passive and active acoustic monitoring, Air / sea interaction
Abstract: The interaction between the atmosphere and the ocean surface is a fundamental hydrodynamic process. The bubble plumes produced from breaking waves play a critical role in mediating air-sea gas exchange and the dissipation of wave energy. The nearshore hydrodynamic region is characterized by many factors, which alter the characteristics of breaking waves and the results of bubble size distribution compared to those observed in the open ocean. The experiments were conducted at the Hydraulics Laboratory of National Cheng Kung University. The facility is equipped with a wave flume measuring 20 m in length, 0.5 m in width, and 0.7 m in depth, operated by a piston-type wave maker. To simulate nearshore conditions, an artificial slope with a gradient of 1/20 was installed at the downstream end of the flume. The parameters of experiment are water depths of 20 and 30 cm, wave periods of 0.8 and 1.3 s, and wave heights of 4, 6, and 8 cm. These factors lead to the relationship between characteristics of bubbles and breaking intensity. A dual hydrophone system was facilitated by hydrophones mounted on the flume bottom capturing the acoustic signal of entrained bubbles. The physical basis relies on bubble oscillation upon formation, emitting acoustic signals with frequencies inversely proportional to their radii, as established by the Minnaert (1933) resonance relationship. Raw acoustic signals are processed using Fast Fourier Transform and Short-Time Fourier Transform to result in spectrogram. Current results have isolated bubble acoustic signatures with a characteristic duration of approximately 0.025 s. Concurrently, a high-speed camera was utilized to record the evolution of the breaking process. These images were used to estimate the void fraction per unit volume, further facilitating the estimation of wave energy dissipation. Future work will involve a comparative analysis between acoustic and optical results to evaluate the feasibility and applicability of passive acoustic techniques for studying bubble size distributions in nearshore breaking waves.

Keywords: Machine-learning classification of acoustic signals, Soundscape ecology and marine organism acoustic behavior, Passive and active acoustic monitoring
Abstract: Humpback whales (Megaptera novaeangliae) produce long, complex songs, hierarchically structured with sound units arranged in phrases that repeat as themes, typically with 5-6 themes per song. Song composition changes over time, yet all singers in a population sing the same version at any one time. Components may be shared between populations across the North Pacific. Characterizing the acoustic unit vocabulary that composes these songs is essential for tracking song evolution, comparing populations, and scaling analyses across ocean-wide studies. Yet manual annotation does not scale, and reliable automated labeling schemes remain limited. In collaboration with the North Pacific Song Comparison Study, we present a comparative study of unsupervised methods for discovering and characterizing humpback song units and phrases, applied to multi-year hydrophone recordings across the Pacific. The study brings together complementary unsupervised approaches: topic modeling over discretized acoustic tokens from small unit- and phrase-level datasets, building on prior results, alongside self-supervised audio embeddings combined with dimensionality reduction and density-based clustering. In preliminary experiments, AVES self-supervised bioacoustic embeddings were projected with UMAP and clustered with HDBSCAN, with evaluation against curated labeled songs using agreement metrics (detection similarity, normalized mutual information, homogeneity) and, for topic modeling, perplexity and the Adjusted Rand Index. Refining and extending this evaluation across methods and datasets is a core focus of this work. Initial results focus on distinct phrase types transmitted across Pacific populations, showing that method choices (temporal granularity, embedding, clustering) strongly influence the resulting unit vocabulary. We will report a comparative evaluation across techniques and datasets to recommend a configuration for song-structure analysis. This work is directly relevant to the OCEANS community as it illustrates how unsupervised and self-supervised learning can be operationalized on ocean acoustic datasets, lowering the labeling burden for marine mammal monitoring and supporting soundscape ecology and population assessment.

Keywords: Passive and active acoustic monitoring, Anthropogenic noise characterization and mitigation, Benchmark datasets, reproducibility, and open-source frameworks
Abstract: To facilitate comparisons of underwater soundscapes across vast regions, we present PBP (PyPAM-Based Processor), an open-source software designed for large-scale passive acoustic analysis. Extending the widely used PyPAM framework, PBP supports both local and cloud archives. The tool includes two core applications: meta-gen, which efficiently extracts essential metadata and generates quality assurance plots; and hmb-gen, which computes calibrated hybrid millidecade spectra.

Meta-gen enables quick data extraction (timestamps, duration) from remote or local archives, while comprehensive summary plots maintain strict quality control before batch processing through hmb-gen. This process yields calibrated time series products of sound pressure levels within a user-defined temporal window. Crucially, the results are stored in self-describing, compressed NetCDF format and include a quality control flag, ensuring both rigorous community archiving standards and easy data sharing.

Here, we show examples on its use, including how to characterize and quantify sources of ship noise in ecologically sensitive habitats with the calibrated hybrid millidecade spectra, and how it was used to efficiently compute sound metrics from a near-contiguous 11-year recording series created by the MBARI deep-water hydrophone deployed on MARS. Integration with the NCEI SoundCoop web data viewer provided by NOAA showcases its utility across diverse regions with multiple hydrophone recording types, and its broader adoption within the sound community.

In summary, PBP is mature open-source software that provides a standardized, accurate, scalable, and compact open-source pipeline that greatly simplifies the analysis of complex passive acoustic datasets for global scientific use.

Keywords: Distributed hydrophone arrays and networked acoustic systems, Benchmark datasets, reproducibility, and open-source frameworks, Passive and active acoustic monitoring
Abstract: Recent developments have increased interest in using Distributed Acoustic Sensing (DAS) for detecting baleen whale calls, as its spatial component offers positional insights not available in conventional passive acoustic monitoring methods. Prior recognition methods have been limited in train/test data to one or two experiments and typically focus on low-frequency blue and fin whale calls, due to scarce publicly available labeled datasets. Robust mutli-species labels across multiple datasets would enable researchers to test generalizability, compare algorithms, and expand DAS species monitoring abilities. However, DAS generates extremely large volumes of data, up to hundreds of gigabytes per hour, making efficient processing and storage essential. Deep learning methods are well-suited for analyzing these datasets and producing actionable detection products, but require strong labeled datasets for training and evaluation. We present a workflow for minimally preprocessing, efficiently storing, and labeling DAS datasets. This workflow is intended for developing: 1) a research dataset, with minimal preprocessing, where original dataset parameters, like sampling rates and intervals, are retained whenever possible; and 2) benchmark datasets (for challenges such as BioDCASE and Kaggle), where dataset parameters are standardized to reduce preprocessing demands on developers. This structure ensures that datasets are easily accessible, reproducible, and suitable for training machine learning models. The workflow also includes a user interface for fast and intuitive labeling of whale calls in multiple linked representations between time, space, and frequency. We use this system to label multi-species baleen whale calls in DAS data at several sites. By combining large-scale compression, standardized parameter settings, intuitive labeling tools, and public hosting, this approach provides a robust workflow for developing generalized detection algorithms, promoting reproducibility, and supporting expanded DAS applications for monitoring baleen whales across species and geographic regions.

Keywords: Environmental DNA/eRNA for biodiversity assessment, Low-cost, scalable oceanographic sensors and platforms
Abstract: Environmental DNA (eDNA) sampling has transformed biodiversity monitoring, ecosystem health assessment, and detection of invasive or rare species. However, researchers monitoring large interconnected biological systems often face funding, site access, weather, and personnel challenges. Robotic eDNA collection technology offers a non-invasive, cost-effective, and scalable solution for consistent, long-term monitoring. For two decades, the Monterey Bay Aquarium Research Institute (MBARI) has developed robotic sampling systems to address these challenges, most notably the Environmental Sample Processor (ESP), which has demonstrated automated eDNA collection in marine and freshwater environments over the past five years. While the ESP has reduced cost, personnel, and access burdens, its size, complexity, and cost limit broader adoption.

To overcome these scalability barriers, MBARI has collaborated with the United States Geological Survey (USGS) over the past four years to design FIDO (Filtering Instrument for DNA Observation), a new robotic eDNA sampler engineered for affordability, ease of use, and scalability. Here we present the challenges we have overcome in building and operationalizing ten of these devices relating to hardware reliability, instrument calibration, and sample protocol development. Our lessons learned center on the importance of exhaustive testing and on designing interfaces and workflows that meet users where they are, minimizing the technical expertise required for successful deployment. We will also present data from our first shipboard deployment aboard an MBARI research vessel. Here FIDO samplers collected paired inshore and offshore samples aboard the R/V David Packard. We plan continued field deployments across diverse environments and are working toward licensing FIDO to a commercial partner to facilitate wider scale use of this technology.

Keywords: Environmental DNA/eRNA for biodiversity assessment, Oceanographic instrumentation and sensors, Deepwater development technology
Abstract: Deep-sea biodiversity remains difficult to assess because of the constraints associated with sampling at high depths. Environmental DNA (eDNA) provides an alternative to traditional observation and capture methods by enabling indirect detection of organisms from genetic material present in seawater. While widely used in coastal environments, its application in the deep sea is still limited by low DNA concentrations and strong spatial heterogeneity. These conditions require the filtration of large water volumes and the use of replication to obtain reliable results. To address these issues, we developed SANDe, an automated in situ eDNA sampler designed for deployment down to 6,000 m. The system is operated from a remotely operated vehicle (ROV), such as VICTOR 6000, from which it receives power and control, as it is not energy-autonomous. Sampling is based on microfiltration using commercial capsule filters (e.g. Spygen, Waterra), typically with a pore size of 0.22 μm. The system can filter large seawater volumes, up to 60 L per filter, with adjustable flow rates adapted to avoid clogging. The filtered volume is monitored by an integrated flowmeter. Several independent samples can be collected during a single deployment, each in triplicate, allowing improved reproducibility compared to sequential approaches. Between samplings, a cleaning sequence using a 5% HCl solution followed by rinsing reduces contamination. Samples are preserved in situ using stabilizing buffers to prevent degradation. The system is controlled via an embedded web application developed within the DeepSea’nnovation project, with user and expert modes. The prototype was tested in hyperbaric conditions and during laboratory and coastal experiments, confirming its operational performance. This work is supported by the French National Research Agency (ANR) under the Programme d’Investissements d’Avenir (ANR-21-ESRE-0042).

Keywords: Environmental DNA/eRNA for biodiversity assessment, Midwater biodiversity and ecosystem structure, Linking omics data to physical, chemical, and ecological processes
Abstract: Larvaceans, or appendicularians, are common throughout the world ocean. Although relatively little is known about most mesopelagic species, all larvaceans feed on particles using mucus filters they inhabit, and through which they pump water to concentrate food. Various larvacean species are found throughout the mesopelagic, with different species often occupying distinct depth zones. Consequently, their houses, as accumulation sites, may provide a molecular snapshot of the surrounding midwater community. They may tell other stories as well. We’ve sampled various species and their houses across a range of mesopelagic depths. Here we compare the eDNA collected on their houses to eDNA sampled from the water column and report what we know so far.

Keywords: Biological and ecological processes in midwater environments, Pollution monitoring, Biogeochemical roles of midwater communities
Abstract: The Rio Grande River originates in the Southern Colorado, USA, flows through the heart of New Mexico, USA, and finally meets the Gulf of Mexico. This River plays a vital role in replenishing the high and low deserts of New Mexico; western Texas, USA and the North-Eastern part of the country of Mexico. However, the Global Warming and climate change is effecting various fresh water systems including Rio Grande in a negative way. Currently, there is no solid data on the changes of the state of Rio Grande. This study looks into various parameters of the Rio Grande River at different locations throughout its journey from its origin to the Gulf of Mexico. These parameters include, but not limited to, change in water level, change in water temperature, change in water quality, change in microbes in the water, and change in vegetation. Data is collected at different locations of Rio Grande throughout the year to include seasonal changes of the river. A database is created in the clouds to store the data. The data is then made available to users through a newly developed mobile application. Although this research is conducted on the Rio Grande River, its impact is closely tied to the Oceans. Rio Grande carries the water and everything in it to the ocean, or in this case the Gulf of Mexico. The changes in the rivers affect the oceans. Therefore, this study is important to the ocean community. This is a continuing project to collect and observe data over the length of 5 years.

Keywords: Sea-ice, glacier, and ice-shelf dynamics, AI/ML for oceanographic and environmental data analysis, Low-cost, scalable oceanographic sensors and platforms
Abstract: Accurate 3D reconstruction of iceberg sails is critical for maritime hazard assessment and climate change monitoring, yet operational practice commonly reduces iceberg geometry to simple dimensional classifications. We apply Neural Radiance Fields (NeRFs) to reconstruct iceberg sail geometry from consumer drone imagery and evaluate results against traditional structure-from-motion and multi-view stereo photogrammetry, both quantitatively and qualitatively. Using an end-to-end opensource pipeline on commercial off-the-shelf hardware, we demonstrate geometrically precise sail reconstructions from which surface area and volume are derived. Photometric evaluation via mesh reprojection across a subset of registered camera poses shows NeRF achieving consistently higher structural similarity and color fidelity than Multi-View Stereo (MVS) across both datasets. NeRF additionally captures geometric features absent in MVS reconstruction, supporting NeRF as a scalable, low-cost approach to remote iceberg monitoring.

Keywords: Oceanography: physical, geological, chemical, biological
Abstract: High biological productivity in eastern boundary upwelling systems (EBUS) is largely driven by regional upwelling-favorable winds, which bring cold, nutrient-rich waters from depth to the surface. At smaller scales, coastal embayments can modify local wind forcing, vertical stratification, thermal fronts, submesoscale features, and circulation patterns in complex ways. These physical changes impact small-scale mixing dynamics, which regulate nutrient delivery, biological productivity, and oxygen dynamics. Despite their importance, the mechanisms by which winds modulate small-scale turbulence within upwelling embayments remain poorly understood due to the challenges of observing submesoscale processes across varying forcing conditions. To address these gaps, this study quantifies the influence of regional and local wind stress on turbulence, stratification, and frontal evolution. San Luis Obispo Bay (SLO), a small upwelling embayment along the central California coast, provides an ideal setting, as a persistent submesoscale temperature front and variable wind regimes interact to modulate turbulence. As part of the Submesoscale Upwelling Bay Frontal Exchange (SUBFEx) project, data from three ~10-day research cruises (June 2024, June 2025, and August 2025) captured contrasting stratification and forcing conditions. The primary dataset comes from a Rockland Scientific MicroCTD microstructure profiler, which includes shear probes to estimate turbulent kinetic energy dissipation, along with temperature, salinity, and pressure sensors. Profiles were collected at nine stations along repeated transects spanning the front. Results from June 2024 show that turbulence in a shallow (<10 m) stratified surface layer scales with wind speed, indicating wind-driven shear as the primary control. In contrast, June and August 2025 observations show a weaker relationship between wind and turbulence with evidence of energetic turbulence below the surface layer, possibly linked to internal wave activity. By investigating the links between turbulence and nearshore submesoscale dynamics in upwelling bays, this study aims to improve our understanding of how these small-scale processes shape coastal environments.

Keywords: Ocean energy, Ocean resources, Protection of offshore energy and critical maritime infrastructure
Abstract: Coastal communities in the Mediterranean often face, at the same time, a fragile energy supply and a structural shortage of freshwater. This paper assesses the integration of a wave energy converter (WEC) for the direct pumping of pressurized seawater into the harbor infrastructure of Castellammare del Golfo (Sicily, Italy), where the pressurized flow feeds a reverse osmosis unit and a hydraulic turbine, so that freshwater and electricity are produced without any intermediate electrical conversion stage. The hourly wave climate measured by the RON buoy of Capogallo was transferred to the nearshore deployment site through a backward ray-tracing procedure based on refraction-shoaling theory, applied to a composite bathymetric model of the study domain. The resulting nearshore climate was then combined with the production characteristic of a single ExoMare-04 unit. Two operating strategies were compared: combined production of freshwater and electricity, yielding 198 m3/y of freshwater and 1236 kWh/y, and electricity-only operation, yielding 2229 kWh/y. The results encourage the potential installation on the test site and scale-up of the system.

Keywords: Ocean energy, Ocean economic potential, Ocean resources
Abstract: The decarbonization of non-interconnected island power systems is constrained by dependence on diesel generation, limited operational flexibility and pronounced tourism-driven demand seasonality. This paper presents an integrated techno-economic optimization framework for a hybrid microgrid combining photovoltaic, wind and wave energy conversion with battery storage and existing diesel backup, applied to Lampedusa Island (Italy) as a representative Mediterranean case study. A deterministic hourly dispatch model, coordinated by a hierarchical rule-based energy management system (EMS), is coupled with a genetic algorithm (GA) minimizing the levelized cost of energy (LCOE) over a discrete design space. Building on this framework, the paper introduces an optimal load-scheduling strategy for the island seawater desalination plant, formulated as a mixed-integer linear program (MILP) that reschedules the weekly operating pattern of the desalination units – within water-tank storage limits – to minimize renewable energy curtailment at fixed weekly water production. Under an annual electricity demand of 33.8 GWh, the optimal configuration achieving 68.66% renewable penetration allows an LCOE of 0.1750 €/kWh, with the installation of 6.0 MW photovoltaic, 3.0 MW wind, 2.5 MW sea wave and 10.0 MWh battery capacity. Applying the proposed MILP rescheduling to the desalination plant recovers 0.84% of the otherwise curtailed renewable energy and cuts residual diesel generation by 1.10% with respect to the current desalination profile at the same renewable share, demonstrating the value of coupling flexible desalination loads with renewable-dominated island microgrids.

Keywords: Energy-harvesting and long-duration monitoring solutions, Applications in geohazards, offshore energy, carbon monitoring, and environmental observation, Ocean energy
Abstract: The U.S. Department of Energy’s (DOE) Hydropower and Hydrokinetic Office (H2O) initiative is accelerating the development and deployment of marine energy technologies to unlock their potential as reliable sources of power for nearshore and offshore applications. This poster will highlight recent progress across a portfolio of DOE-supported projects that demonstrate the pathway from technology innovation and in-water validation to industry-led commercialization.

We will spotlight three National Laboratory H2O projects—HEROWEC, the Pioneer WEC Buoy, and SurfWEC—that are advancing wave energy technologies through technology development, ocean testing, and performance validation. Recent milestones from these projects demonstrate the value of real-world operational experience in generating critical performance data, evaluating system reliability and survivability, and identifying lessons that can inform future designs and deployments.

The presentation will also highlight industry-led efforts, including CalWave, and Ocean Motion Technologies, that are translating technology advances into systems and applications with commercial potential. Together, these efforts illustrate how DOE investments can bridge the gap between research and development, in-water demonstration, and market adoption.

Across this portfolio, H2O is helping generate the data and operational experience needed to reduce technical risk, improve technology readiness, strengthen industry partnerships, and enable future private investment. These projects demonstrate tangible progress toward a robust marine energy industry and the deployment of reliable marine energy systems capable of supporting offshore, remote, and other emerging energy applications.

Keywords: Pollution clean-up and pollution remediation, Pollution monitoring, Maritime domain awareness and autonomous monitoring
Abstract: This study developed a standard accident scenario framework to ensure safety during the port operation phase in response to the introduction of vessels using eco-friendly alternative fuels (LNG, methanol, and ammonia) in accordance with the IMO 2050 Net-Zero goal. By integrating domestic and international technical guidelines, the accident causal relationships were structured into a six-stage standard flow: ① preconditions, ② accident occurrence, ③ primary impact, ④ secondary diffusion, ⑤ tertiary impact assessment, and ⑥ response and recovery. Standard data fields were also defined for the quantitative reproduction of each stage. Analysis of the worst-case accident scenario (total spill within 10 minutes) for the Ulsan Port model confirmed massive ripple effects, including port paralysis and approximately 125 casualties within a 250m radius, along with the unique risk characteristics of each fuel type. The dataset from this study will be utilized as a core input for CFD analysis and digital twin simulators, contributing to the revision of domestic bunkering safety guidelines and the advancement of management systems.

Keywords: Maritime domain awareness and autonomous monitoring, Surveillance, detection, and threat-assessment techniques, Data fusion, uncertainty quantification, and predictive modeling
Abstract: Maritime target tracking over large distances often requires multi-agent teams without centralized coordination, and intermittent communication. Each agent must maintain an independent estimate that can take advantage of opportunistic communications availability when possible. This can lead to overly confident local estimates in the absence of external data. In this work, we propose an augmentation to a classical particle filter implementation that accounts for this potential source of error by forcing particles to spread strategically in the absence of informative updates from other sensor nodes. We demonstrate our method using Unmanned Surface Vessels (USVs) on a lake, and show that our augmentations do not deteriorate nominal performance, and provide an advantage in some specific edge cases.

Keywords: Maritime domain awareness and autonomous monitoring, Citizen-science networks and participatory ocean observations, Oceanographic instrumentation and sensors
Abstract: The Alaska Ocean Observing System (aoos.org; AOOS) enhances the nation’s ability to collect, deliver, and use ocean information as part of the the national Integrated Ocean Observing System (IOOS). Alaska’s immense geographic scope includes five Large Marine Ecosystems and more coastline than all other US states combined. The size of the region, along with environmental factors such as extreme cold, seasonal daylight, and remote/inaccessible coastlines makes thorough observing coverage uniquely challenging. Alaska’s coasts and oceans support high-value fisheries and mariculture, oil and gas development, national security, shipping and commerce, community subsistence and food security, and many other economic uses. In this context, agency-supported observations must leverage external resources and partnerships to meet Alaska’s observing needs. AOOS facilitates five topical networks that provide focused collaboration around observing needs, technology and tool development, resource sharing, and data and product delivery to stakeholders. AOOS-facilitated networks include the Alaska Harmful Algal Bloom Network, Alaska Ocean Acidification Network, Alaska Water Level Watch, Sea Ice for Walrus Outlook, and Northeast Pacific Acoustic Telemetry. These network-based partnerships accelerate technology development from ideas to sustained operations, and improve decision-support for communities, industry, and resource managers facing rapid environmental change. This poster will present case studies showing how these networks co-developed ocean observing tools and approaches serving the unique needs of Alaska and the Arctic, and will describe ways to get involved with these regional networks and broader efforts.

Keywords: Marine education and outreach, Ocean resources, Ocean economic potential
Abstract: Science-based educational programming to educate stakeholders on local issues can help to encourage responsible stewardship of marine resources. In the northeast United States, the marine ecosystems and resources in New Jersey support very valuable tourism, aquaculture, and commercial and recreational fishing industries. In addition to commercial businesses, a large and active community of recreational surfers, divers, fishers, boaters, and water sport enthusiasts depends on the state's coastal waters for recreation and employment. The continued health of these marine ecosystems is therefore crucial. The sustainability of New Jersey’s marine resources and the associated industries rely on the implementation of responsible stewardship practices by a diversity of stakeholders. Therefore, the objective of our science-based programs through Rutgers Cooperative Extension is to educate community members on the science and management of New Jersey’s marine resources to encourage responsible stewardship practices. Our educational programs include two 10-week courses offered annually for local stakeholders, including the Coastal Stewardship and the Introductory Fisheries Science for Stakeholders (IFISSH) courses. Short and long-term program evaluation data have documented significant increases in knowledge gained. Another resource highlighting stewardship is our Marine Debris webpage, which provides a list of opportunities in the Mid-Atlantic to properly dispose of marine debris. The website went live in November 2024, with maintenance conducted regularly, and was commended by the NOAA Mid-Atlantic Marine Debris Action Plan Team. The website has received 595 views from 314 active users since launching. In a related effort, we have an ongoing neoprene recycling drive that is collecting waders, wetsuits, and survival suits. New Jersey’s reliance on neoprene products for its large surfing, diving, fishing, and boating community has driven an effort of recycling 1,195 pounds of neoprene since 2020. These spotlighted programs and additional resources continue to successfully educate stakeholders in best practices, with a means to incorporate the responsible stewardship of New Jersey’s marine resources into everyday life.

Keywords: Public engagement through art, media, and storytelling, Ocean literacy, outreach, and education initiatives, Strategies to broaden participation and promote inclusivity
Abstract: The Western Flyer Foundation is a marine science research and education nonprofit rooted in one of science's most celebrated interdisciplinary collaborations— the friendship of ecologist Ed Ricketts and author John Steinbeck. In 1940, the pair chartered the fishing vessel Western Flyer to survey the Gulf of California, producing Sea of Cortez, a work that transformed ecological observation into public wonder. This expedition remains a model for how science, art, and narrative can foster deep, lasting appreciation for the ocean.

Today, the restored Western Flyer is a hybrid diesel-electric research vessel and a living cultural artifact, connecting this legacy to the urgent ocean challenges of the present. Through onboard experiential learning programs, we engage students, scientists, artists, and community members from the Pacific Northwest to the Baja Peninsula, fostering place-based ocean appreciation rooted in their local ecosystems.

We emphasize approaches that make science tangible and accessible. Through a partnership with the California State University Monterey Bay Science Illustration Program, we host an annual science illustration intern, supporting student artists working at the intersection of art, technology, science communication, and ocean literacy. In 2025, our intern developed three-dimensional models of Monterey Bay plankton, making microscopic marine life visible, and deepening understanding of its role in regional ecosystems.

This work extends across regions, from the Pacific Northwest to the Gulf of California. The Western Flyer's landmark 2025 return to the Gulf of California—its first visit in 85 years— established partnerships with scientists and communities in Baja California Sur. With a 2027 return expedition underway, we offer a replicable model for fostering ocean appreciation across cultures through the integration of science, art, storytelling, and community engagement.

Keywords: Ocean energy, Offshore structures, Hydrodynamics
Abstract: Offset-column mass is one of the largest single cost drivers in semisubmersible floating-wind support structures, but conventional lightweighting strategies forfeit waterplane stiffness and degrade motion response. This paper presents a feasibility analysis for an alternative: hollowing the offset columns of an OC4 semisubmersible to host oscillating water column (OWC) chambers, so that the geometric step that removes mass from the offset columns simultaneously creates an air-capped internal water column that supplies pneumatic restoring and damping. To select a viable hollow geometry, a two-fidelity approach is used, followed by a RANS-VOF verification under a moderate sea state to quantify the resulting motion behavior. The most consequential finding is a regular-to-irregular response reversal: under monochromatic excitation, the lightweighted platform exhibits a modest heave penalty relative to the solid baseline, but under broadband irregular excitation at the same nominal sea state, it achieves a roughly 20% heave reduction. The viscous damping recovery in the CFD model relative to linear theory further amplifies this trend. A passive-orifice study at the same sea state maps a clear hydro-pneumatic operating envelope: a wide restrictive band sustains hydro-pneumatic coupling and useful pneumatic power, while an over-open limit collapses chamber pressure and breaks the compensation mechanism. The combined evidence supports OWC-integrated lightweighting as a credible direction for cost-effective stabilized FOWT platforms and motivates future chamber-specific active airflow control within the identified envelope.

Keywords: Buoy Technology, Hydrodynamics, Ocean energy
Abstract: Access to safe drinking water remains a critical global challenge, particularly in coastal regions affected by natural disasters, where infrastructure and electricity are often disrupted. Although seawater is abundant in these environments, converting it into potable water typically requires energy-intensive desalination systems that depend on stable power sources. This project develops and evaluates a floating wave-powered system that uses a two-body wave energy converter (WEC) with a mechanical power take-off (PTO) to amplify low-speed wave motion into mechanical power and drive an axial piston pump for pressurized, unidirectional flow. A Berkeley wedge is incorporated to reduce vortex formation and improve hydrodynamic performance. Time-domain simulations using WEC-Sim were conducted to study how wave period, mooring stiffness, and PTO damping affect power output, guiding design choices. A 1:8 prototype was designed, fabricated via 3D printing, assembled, and tested in a controlled wave tank to quantify pumping behavior and identify failure modes that informed subsequent design iterations. Using modeled power and conservative reverse osmosis (RO) specific-energy assumptions, a full-scale unit (~2.0 m × 1.25 m) is projected to produce >400 L/day of freshwater, enough to supply ~200 people during the “golden 72 hours” after disasters.

Keywords: Low-cost, scalable oceanographic sensors and platforms, Vehicle design, Rapid-deployment systems and modular instrumentation
Abstract: Abstract — Oceanographic research is frequently hindered by a market gap between cost-prohibitive enterprise surface vessels and payload-limited micro-USVs. To address this barrier, this work presents the design, methodology, and validation of the All Purpose Low-cost Uncrewed Surface Vessel (APLUSV). Engineered to maximize scientific versatility, the APLUSV is a highly modular mid-sized catamaran constructed with commercial off-the-shelf components modified with minimally specialized manufacturing. The vessel features a structural semi-monocoque deck and a frame constructed of aluminum tubing with universal mounting features to offer a total of 7.87 square meters of mounting space, that can support up to 82 kilograms of diverse payloads. The collapsible mast and thruster mounts enable convenient transport in a pickup truck. The vessel utilizes an open-source autonomy architecture equipped with differential thrust and redundant Beyond Visual Line of Sight (BVLoS) compatible telemetry. Field trials validated APLUSV’s high-precision autonomous navigation, reliable propulsion, and an extended mission duration. Ultimately, the APLUSV democratizes accessibility for marine data collection.

Keywords: Aquaculture engineering, Remotely operated vehicles, Vehicle design
Abstract: Aquaculture serves as a sustainable alternative to land-based agricultural systems. However, they struggle to match the output of their terrestrial counterparts due to several factors. In oyster aquaculture, one such factor is the heavy reliance on manual labor. Floating systems typically require the flipping of bags or cages to clear biofouling, which restricts water and nutrient flow into the containers. There can be thousands of these bags and each can weigh up to 30 kgs (60+ lbs). MIT Sea Grant has developed a prototype unmanned surface vessel (USV) for use in flip-bag oyster aquaculture systems to remove this reliance on manual labor. Several versions have been made, each building off of the results of the previous iterations' results. Using findings from the previous vessels, the third generation of the Oystermaran makes improvements on oyster field maneuverability and flipping ability. In this paper, we describe the overall design and fabrication of the system and the results of field testing. The system has been tested in MIT Sea Grant and the Charles River, demonstrating considerable progress over previous generations in terms of hydrodynamic performance and structural integrity while flipping baskets.

Keywords: Vehicle navigation, Vehicle performance, Autonomous underwater vehicles
Abstract: This paper proposes a disturbance observer-based model predictive control (DOB-MPC) framework for precise trajectory tracking of unmanned surface vehicles (USVs) under environmental disturbances and model uncertainties. While MPC can explicitly handle system constraints and underactuation, its performance degrades in the presence of disturbances and modeling errors. To address this limitation, the proposed approach incorporates disturbance estimates into the prediction model, ensuring disturbance-aware and feasible control actions while avoiding instability caused by actuator dynamics. The proposed method is validated through open-sea experiments using an 8 m class USV, demonstrating significant improvements in trajectory tracking performance compared to nominal MPC.

Keywords: Real-time decision systems for ocean robotics and autonomous operations, Vehicle performance, Automatic control
Abstract: Hydrofoiling sailing vessels present a strongly nonlinear control problem in which aerodynamic, hydrodynamic, and inertial forces are tightly coupled and environmental conditions vary continuously. This work evaluates classical and learning-based control strategies for a hydrofoiling catamaran under variable model and environmental conditions, using a validated six-degree-of-freedom nonlinear simulation. Two controllers are compared: a Linear Quadratic Regulator (LQR) and a Proximal Policy Optimisation (PPO) reinforcement learning agent. Under nominal conditions, LQR demonstrates superior tracking performance with more stable actuation effort. However, Monte Carlo analysis across 5000 trials shows that PPO maintains stable foiling flight in 67% of cases, compared to 12.6% for LQR. A delay-aware PPO variant trained under 250 ms actuation delay retains stability in 63% of trials in the delay environment, compared to 3% for the standard PPO, with no architectural modification required. These results demonstrate that while LQR provides efficient control near its design point, the substantially greater robustness of PPO to environmental variability and hardware constraints makes it the more practically deployable controller for vessels operating across a broad and uncertain envelope, where robustness is a more appropriate design objective than local optimality.

Keywords: Low-cost, scalable oceanographic sensors and platforms, Open-source hardware, software, and community innovation, Dynamic positioning
Abstract: This work presents the development of an active heave compensation (AHC) system for an existing fiber optic tether system and targets the sea conditions faced during the tether breakage of a past ROV deployment. The system adapts the active heave compensation approach intended for offshore cranes and implements it onto an affordable off-the-shelf microcontroller with estimation, prediction, and control methods. The estimation system produces attitude estimation from sensor fusion of measured IMU data and uses adaptive filtering to estimate the heave. The prediction algorithm using auto-regressive modeling and a Levinson recursive least squares algorithm is used to produce a forecasted heave and velocity. This is then successfully used as a pure feedforward command to drive compensator action from the motor to correct for heave. Simulation and bench tests are used to demonstrate the performance of the system.

Keywords: Foundation models and large-scale training for marine applications, Benchmark datasets, reproducibility, and open-source frameworks, Sonar imaging
Abstract: Acquiring large-scale annotated sonar datasets for underwater perception is prohibitively expensive due to the high cost of marine operations and the limited availability of controlled test environments. We developed FURo-Sim, a GPU-accelerated, physics-based sonar simulation framework built on Unreal Engine 5 (UE5) that generates physically realistic synthetic sonar imagery with pixel-accurate ground-truth annotations at negligible marginal cost.

FURo-Sim couples UE5's GPU rasterization backend with an acoustic model based on the sonar equation, supporting forward-looking sonar (FLS) and side-scan sonar (SSS) modalities. Target strength is computed via a roughness-dependent BRDF combining diffuse Lambertian and specular Phong lobes; transmission loss models spherical spreading and frequency-dependent absorption; multipath propagation is handled through recursive GPU ray tracing. Three independent noise sources (speckle, crosstalk, ambient) are composited to reproduce the stochastic appearance of real sonar returns.

A key contribution is automatic multi-modal ground-truth generation: per-pixel 2D segmentation masks, 2D bounding boxes (highlight-only and highlight-with-shadow variants), semantically labeled 3D point clouds, and 3D oriented bounding boxes are produced alongside every rendered frame without manual labeling. The simulator also includes 6-DoF AUV and ASV vehicle models, an auxiliary sensor suite (RGB camera, IMU, GPS, pressure sensor, magnetometer, LiDAR, DVL), and ROS 2 / Python interfaces.

Quantitative validation against real DIDSON 300m and Oculus M750d imagery — using MSE, PSNR, MS-SSIM, and LPIPS metrics — confirms that each acoustic component contributes meaningfully to realism and that higher-order reflections improve structural fidelity in complex scenes. FURo-Sim provides a complete, end-to-end simulation and dataset generation platform for field and underwater robotics research where real annotated data remains scarce.

Keywords: AI/ML for oceanographic and environmental data analysis, Sonar imaging, Ocean exploration
Abstract: Bathymetry measurement and estimation is critical for understanding the structure of underwater environments. While multibeam echosounders (MBES) provide accurate depth measurements, their deployment on surface vessels limits spatial resolution in deep water. Sidescan sonar (SSS), in contrast, can be deployed on autonomous underwater vehicles (AUVs) to operate closer to the seafloor, enabling high-resolution acoustic imaging. However, the inherent ambiguity in measurements and the limited observation coverage of sonar datasets degrade the performance of state-of-the-art methods for underwater bathymetry estimation from SSS. In this paper, we propose a differentiable neural rendering framework for SSS that enables effective estimation of bathymetry from SSS measurements. Our approach incorporates a measurement model based on SSS geometry, explicitly accounting for its wide elevation aperture and associated ambiguities. We evaluate our method on both synthetic and real-world datasets, demonstrating improved robustness and accuracy compared to alternative approaches.

Keywords: Passive and active acoustic monitoring, Anthropogenic noise characterization and mitigation, Sonar signal processing
Abstract: Single-hydrophone Doppler analysis of ship tonals often supports several speed modes. In 18 waterjet craft pass-bys, a permissive Doppler screen nominated the low-speed region for 10 passes, although only 6 were commanded low. We test tonal–pedestal contrast (TPC) as a gate criterion for applying a low-speed restriction to the Doppler set, rather than as a speed estimator. TPC compares a tracked ridge with its local broadband pedestal and is aggregated as a ratio of temporal medians. The six commanded 10-kn/2000-rpm passes had higher mean TPC than the rest (Welch p = .0014). Among the other 12 passes, Doppler nominated low for four and the TPC criterion was met for two different passes. With each period held out, a midpoint AND gate allowed contraction for all 6 targets and none of the 12 others. For the six contracted passes, the median candidate-set measure changed from 46.0 to 22.5 km/h. Across all 18 passes, reference inclusion remained 16 of 18. The smallest TPC decision margin was +0.13 dB.

Keywords: Acoustical oceanography, Signal coherence and fluctuation, Sound propagation and scattering
Abstract: This paper examines tide-influenced semi-diurnal travel-time perturbations along the 166 kilometer path from the Kauai Beacon (KB) to the ALOHA Cabled Observatory (ACO). The path lies close to the Ka'ena Ridge, one of the world’s most energetic baroclinic tide-generation sites. For 72 hours, starting December 4th, 2025, at 00:00 UTC, the KB duty cycle was raised from 2% (normal operation) to 33% (20-min transmissions every hour). This time frame coincides with a full moon (December 4, 2025, 23:14 UTC) and brackets the maximum tidal forcing (spring tide). The travel time perturbations for each transmission were calculated using the integrated autocorrelation phase at one period lag (IAP1) method, which uses the phase of the full received field of each transmission (rather than individual peaks in the arrival pattern) to resolve perturbations. IAP1 was used because strong bottom interactions along the KB-ACO path (all ray paths are bottom-interacting) produce variable multipath structure that hinders tracking individual arrivals over time. IAP1 results yield a clear semidiurnal signal with an M2-fit amplitude of 7.9 ms with a phase of 55.66° (phase value relative to beginning of time series), confirmed by an independent envelope cross-correlation estimate (8.1 ms, 54.75°). A barotropic prediction derived from a Princeton Ocean Model simulation of the M2 tide accounts for 1.7 ms of this amplitude (phase 17.55°), and a BELLHOP ray-trace through a ROMS sound-speed field, which captures a portion of the sound speed variability (including baroclinic and mesoscale temperature and salinity variability), contributes 1.4 ms (phase 69.46°). The combined model contributions have an amplitude of 2.8 ms and a phase of 46.43°. The difference between the combined model and the observed data is roughly 5.1 ms in the semidiurnal band. The unresolved signal leaves room for further investigation of tidal energy at the location. The results show that the KB–ACO path, combined with IAP1 processing, is a practical tool for near-field observation of the Hawaiian Ridge tide, complementing basin-scale tide studies and results from the Hawaii Ocean Mixing Experiment (HOME).[ONR Ocean Acoustics program, grant no. N00014-24-1-2546.]

Keywords: AI/ML for oceanographic and environmental data analysis, Foundation models and large-scale training for marine applications, Benchmark datasets, reproducibility, and open-source frameworks
Abstract: Accurate benthic characterization is essential for coral reef monitoring, yet existing image-based approaches have largely focused on coral cover, which provides only limited insight into reef condition, ecological resilience, and functional composition. In particular, coral growth-form composition constitutes an important ecological dimension because different morphologies contribute differently to recovery dynamics, structural persistence, and habitat provision. To address this gap, we developed a morphology-aware semantic segmentation framework for pixel-level classification of seven canonical coral growth forms in underwater benthic imagery, including branching, corymbose, digitate, massive, encrusting, foliose, and submassive corals, together with background. The model was trained and evaluated using 3,024 manually annotated orthomosaic patches generated from high-resolution underwater surveys conducted in Kenting National Park, Taiwan. UNet++ was adopted as the backbone network because its dense skip connections and multi-scale feature fusion are well suited to the delineation of fine colony boundaries and spatially adjacent coral morphologies. The proposed framework achieved an overall mean Intersection over Union of 0.833, an overall accuracy of 0.988, and an F1-score of 0.879 on the test set. Performance was highest for branching and massive corals, whereas lower accuracy for digitate, encrusting, and submassive classes reflected greater morphological ambiguity and stronger visual similarity to surrounding substrate. These results demonstrate that semantic segmentation of coral growth forms can move reef-image analysis beyond coral-cover estimation toward ecologically richer benthic characterization. The proposed framework provides a practical basis for scalable, morphology-resolved reef monitoring and offers a foundation for future integration of structural complexity metrics into unified ecological assessment workflows.

Keywords: AI/ML for oceanographic and environmental data analysis, Model-based signal processing techniques, Hydrography / Seafloor mapping / geodesy
Abstract: Forward-looking sonar (FLS) is widely used on autonomous underwater vehicles (AUVs) for seafloor mapping and collision avoidance. However, it usually has a wide vertical aperture. As a result, elevation information is unavailable because echoes returned from different elevation angles at the same range and azimuth will merge into a single image pixel. This will make terrain reconstruction challenging. To this end, this paper presents a three-stage convolutional neural network to predict per-pixel elevation angles from a FLS image for 3D cloud generation. The network is trained using co-registered multibeam echosounder (MBES) measurements. The staged decomposition is designed to perform no-return rejection, partial-return clas- sification and elevation regression. It is motivated by the fact that 86.5 % of pixels in an FLS image have no valid returns. The elevation regression stage is trained with a locally weighted Huber loss function normalized by the total-variation and Laplacian smoothness terms along each beam. The method is validated using the simulated data generated in the Stonefish simulator. The pipeline attains an F1 score of 86.3 % on the classification (stage 1&2), an RMSE of 1.03 deg (0.018 rad) in simulation and RMSE of 6.51 deg (0.11 rad) deg on the field dataset. To make the trained model applicable to field data, speckle noise was applied on the simulated sonar images and bilateral filtering was applied with intensity thresholding to enhance the simulated images that are used in model training. Then the trained model is applied to the field data to create a point cloud of the terrain and compared with the results collected using a Norbit MBES mounted on an ASV.

Keywords: AI/ML for oceanographic and environmental data analysis, Oceanographic instrumentation and sensors, Pollution monitoring
Abstract: The rise in plastic pollution in our oceans and rivers necessitates automated systems for microplastic detection. Scout is a recently developed impedance-based microplastic sensor that can detect and distinguish microplastics from other particles commonly found in waterways. This work evaluates signal processing and machine learning techniques for analyzing Scout’s data output to detect particles passing through the sensor and classify those particle detections as plastic or non-plastic. Peak detection, Short-Time Fourier Transform (STFT), and wavelet transforms were tested for particle detection. Six machine learning classification models were evaluated to differentiate plastics from biological materials and bubbles. Detection results showed wavelet-based detection as the best performing (93.7% recall) with peak-based detection having acceptable performance (87.9% recall) with a much smaller compute time. Results indicated that convolutional neural networks (CNN) achieved the highest classification accuracy (98%), but random forest models offer a highly efficient alternative (93% accuracy) with significantly lower inference time, making real-time microcontroller processing for an impedance-based microplastic sensor theoretically feasible.

Keywords: Autonomous underwater vehicles, Open-source hardware, software, and community innovation, Oceanographic instrumentation and sensors
Abstract: Cylindrical watertight enclosures are widely used across various underwater systems, from unmanned underwater vehicles (UUVs), to remotely operated vehicles (ROVs), to various sensor platforms. However, electronics are typically built on rectangular PCBs arranged in horizontal stacks, which inefficiently occupy the circular cross-section volume that is critical for both payload capacity and buoyancy management. This paper presents CUB (Central Utility Board) and CougarTail, a general-purpose system designed to address this gap. CUB is a circular PCB sized for 4-inch-diameter enclosures that consolidates a Raspberry Pi Compute Module 5 (CM5) and an STM32 microcontroller, while also providing power management features and auxiliary connections. Mounted coaxially, CUB reduces the electronics stack of our CougUV from 200 mm of tube length and 709 g to 25 mm and 156 g, returning that length and mass budget to payload and buoyancy trim. CougarTail is an open-source companion sensor mast that houses a GPS antenna and two dual-band (2.4 and 5 GHz) omnidirectional PCB antennas. Both components are validated through bench testing and integration on a CougUV platform, our small open-sourced torpedo UUVs.

Keywords: Low-cost, scalable oceanographic sensors and platforms, Energy-harvesting and long-duration monitoring solutions, Oceanographic instrumentation and sensors
Abstract: Maritime sensing faces challenges from lack of energy and communications in remote regions. Especially when constrained by cost, sensors struggle to provide effective coverage of the maritime environment in ports and waterways. This work introduces a standardized, small, low-cost Shore Pod sensor platform for deployment in large numbers to autonomously monitor the maritime domain. Using this platform, we evaluate solar energy harvesting for moored Shore Pods. Given historical irradiance data of a target deployment environment and onboard sensor characteristics, we propose and evaluate an energy buffer sizing process to maximize uptime while minimizing volume, mass, and cost. We identify three energy-scheduling pitfalls and evaluate operating strategies to avoid these pitfalls and maximize uptime across operating environments. Top-performing Shore Pod operating strategies achieve 98.9% of maximum uptime.

Keywords: Machine-learning classification of acoustic signals, Passive and active acoustic monitoring, AI/ML for oceanographic and environmental data analysis
Abstract: Underwater acoustical signal classification became a matter of significance over a century ago when submarines became threats to national security amidst World War I. Since then, our understanding of the depths of the ocean has expanded, yet we are barely scratching the surface. If classification could be done without the sending of signals, but rather just with a passive listening of the underwater soundscape, the rate at which our knowledge of the ocean expands would accelerate dramatically.

The Monterey Accelerated Research System (MARS) lies 52 km off the coast of California and uses a hydrophone and three accelerometers to record raw acoustic data year round. For the purpose of this research, the data was divided into minute-long files. This data was initially processed into time-frequency representations of power spectral density levels and bearing. This research focuses on how these time-frequency representations can best be used for signal classification.

Recent research by Artorix de la Crus and Mae L. Seto indicated that Rényi entropy may be key in appropriate signal classification. Rényi entropy is a specialized form of Shannon Entropy which describes a measure of the unpredictability of an event.

In addition to entropy, other acoustic features were evaluated for their potential in signal classification: kurtosis, spectral flux, and spectral centroid. For each combination of acoustic features, unsupervised learning methods were applied and the classification results were compared. Initial efforts were made with KMeans Clustering and future efforts will implement additional methods to improve preliminary labeling. After removing features one by one, it was evident that when feature vectors included a measure of entropy, the clustering performed considerably better.

These results highlight the potential for Renyi entropy to advance the validity of unsupervised signal classification for providing preliminary labels to passively recorded data.

Keywords: Passive and active acoustic monitoring, Distributed hydrophone arrays and networked acoustic systems, Low-cost, scalable oceanographic sensors and platforms
Abstract: Accurate underwater localization and reliable ocean-to-land data transmission are essential for ocean monitoring and autonomous underwater vehicle operations. Existing solutions often suffer from either high complexity or limited localization robustness. This work presents a low-cost and low-complexity framework that integrates acoustic source localization with long-range wireless data transmission. For underwater localization, three passive hydrophones are employed with Generalized Cross-Correlation with Phase Transform (GCC-PHAT) to estimate the time difference of arrival (TDOA). To mitigate multipath-induced false peaks in reverberant underwater environments, a grid-based spatial evaluation method combines the correlation information from multiple hydrophone pairs to identify the most spatially consistent source location. Experimental measurements in an enclosed water tank demonstrate successful localization in six of seven tested positions. In parallel, an ocean-to-land wireless communication system integrating a buoy-mounted RF modem and an onshore software-defined radio (SDR) base station is developed. The proposed framework provides a practical and scalable solution for low-cost underwater sensing, localization, and long-range maritime data transmission.

Keywords: Integration with autonomous platforms and observatories, Sensor synergy, Autonomous underwater vehicles
Abstract: This work presents the development, implementation, and testing of a sensor-enabled towfish with the use of autonomous systems. The towfish is equipped with magnetic field sensors, an inertial measurement unit (IMU), and pressure sensors to allow for subsea magnetic field mapping. This research focuses on four primary goals: the characterization of the performance of the towfish and electromagnetic sensors, an assessment of the quality of the measured data, determination of the minimum length of the tether needed for sensor package isolation, and the assessment of the impact of towing on vehicle drag and its maneuverability. Tests are currently being conducted to assess the towing behavior of the towfish from an autonomous surface vehicle (ASV). Following successful tests on board the ASV, the towfish is being optimized for use on board the REMUS 100 autonomous underwater vehicle (AUV).