Modern marine research requires high-precision three-dimensional (3D) underwater data. Underwater environments experience severe visible light attenuation, which leads to inferior imaging compared with air. In contrast, sound waves are less affected underwater; hence side-scan sonar is used for underwater 3D reconstruction. Typically, the shape-from-shading algorithm (SfS) is widely used to reconstruct surface normal or heights from side-scan sonar images. However, this approach has challenges because of global information loss and noise. To address these issues, this study introduces a surface-normal fusion method. Specifically, we propose a frequency separation SfS algorithm using a discrete cosine transform, which provides a surface-normal map with less noise. We then fuse the surface-normal map with a novel depth estimation network to achieve high-precision 3D reconstruction of underwater side-scan sonar images. We conducted experiments on synthetic, NYU-depth-v2, and real side-scan sonar datasets to demonstrate the effectiveness of the proposed method.
Ocean temperature prediction is significant in climate change research and marine ecosystem management. However, relevant statistical and physical methods focus on assuming relationships between variables and simulating complex physical processes of ocean temperature changes, facing challenges such as high data dependence and insufficient processing of long-term dependencies. This paper comprehensively reviews the development and latest progress of ocean temperature prediction models based on deep learning. We first provide a formulaic definition for ocean temperature prediction and a brief overview of deep learning models widely used in this field. Using data sources and model structures, we systematically divide ocean temperature prediction models into data-driven deep learning models and physically guided deep learning models; and comprehensively explore the relevant literature involved in each method. In addition, we summarize an ocean temperature dataset and sea areas, laying a solid foundation for ocean temperature prediction. Finally, we propose current challenges and future development directions in ocean temperature prediction research based on deep learning. This article aims to analyze existing research, identify research gaps and challenges, provide complete and reliable technical support for climate forecasting, marine disaster prevention, and fishery resource management, and promote the further development of ocean temperature research.
Given that clouds can absorb and scatter radiation signals in the visible and infrared bands, cloud detection is a key preprocessing step for ocean color and sea surface temperature retrievals. In this research, a Spectral-and-Textural-Information-Guided deep neural Network (STIGNet) is designed for cloud detection in global ocean data from the Haiyang-1C (HY-1C)/Chinese Ocean Color and Temperature Scanner (COCTS). Considering the spectral and textural properties of clouds, the model incorporates HY-1C/COCTS spectral data, differences in brightness temperature (BT), local statistical characteristics of BT, and geographical location information–all of which are closely related to cloud features. Notably, an edge learning module is implemented to emphasize edge features during the training process. We construct a HY-1C/COCTS cloud detection dataset to train and test the cloud detection model. In the dataset, labels are generated by combining the Bayesian cloud detection method with a manual mask. Analysis of the resulting cloud detection images indicates that STIGNet exhibits accurate performance across various types of clouds while showing minimal overestimated errors in areas such as ocean fronts or sun glints, where they tend to occur frequently. The ablation experiments performed on physical-based input features and edge learning modules show enhancements in cloud detection accuracy. Evaluation results demonstrate an overall accuracy of 96.64%, with a cloud overestimated error of 1.61% and a cloud missed error of 1.76%. These findings highlight the effectiveness of STIGNet in generating precise cloud masks for HY-1C/COCTS data.
A disposable miniature radiometer has been developed using optical filters for spectral separation. Limitations in accurately retrieving irradiance from the broad-band measurement results can be attributed to the broad-band filters. This paper proposes an algorithm for spectral irradiance using broad-band optical filter data (SIBOF algorithm) to achieve precise retrieved irradiance through four correction steps. First, the algorithm uses an energy ratio method to adjust the broad-band data to narrow-band data. The energy ratio is derived from the reference lamp spectrum and measured optical filter transmissivities. Second, the algorithm corrects for filter transmissivity differences by multiplying the normalized spectral transmissivities by calibration coefficients. The third step involves polarization correction, compensating for additional transmissivity caused by polarization effects from the film overlying on the cosine collector, thus eliminating errors due to film polarization. The fourth step involves radiative heating correction, where fitting curves and coefficients are used to analyze the relationship between irradiance deviation and actual irradiance to correct the data. Standardized tests indicate that, after applying the four corrections, the results are highly consistent with the irradiance from the reference radiometer, demonstrating that these correction steps constitute a reliable algorithm for spectral irradiance using broad-band optical filter data. In April 2024, a 20-day sea fog sounding observation was conducted at the Qianliyan Ocean Station. The irradiance data from the miniature radiometers before launch were corrected and compared with those measured by the reference radiometer on the ground. Results indicate that the irradiance retrieved through the algorithm was in good agreement with the measurements from the reference radiometer, validating its performance across various weather conditions.
This study employed OpenFOAM, the delayed detached-eddy simulation (DDES) turbulence model, and structured grids to develop numerical models for three centrifugal pumps with twisted blades. The internal pressure field, velocity field, forces, and fluctuation characteristics of the centrifugal pumps are comprehensively analyzed under various operating conditions. The findings indicate that the pressure is relatively higher in the flow passages near the volute tongue and the outlet within the impeller. Regions of high relative velocity (slip velocity) are mainly found on the suction side of the blades, indicating that the design of the blade suction side affects the fluid outward slip performance. As the flow rate increases, the forces and force fluctuation amplitudes of each pump component also rise. Conversely, as the rotational speed increases, the force on the blades or impeller gradually increases while the fluctuation amplitude decreases. In the stationary domain, the force on the volute gradually decreases while the fluctuation amplitude of this force increases. The shape of the volute tongue influences the rate at which pressure inside the volute is converted to outlet pressure. The power spectral density (PSD) of pressure fluctuations is smallest at the nominal flow rate, displaying a clear and distinct axial frequency pattern without complex low-frequency fluctuations. Under low flow and high-speed conditions, the PSD at the axial frequency is relatively small, whereas the pressure PSD at other low frequencies is relatively large. This indicates instability in the flow under these conditions.
El Niño-Southern Oscillation (ENSO) is a periodic climate phenomenon in the equatorial Pacific that significantly influences global climate patterns. Accurate prediction and monitoring of ENSO events are essential for meteorological agencies and governmental institutions. This study introduces a content-guided attention module within a convolutional neural network to improve prediction accuracy. This module models inter-channel relationships and enhances information interaction by integrating channel and spatial attention weights. These advancements substantially improve prediction accuracy and help overcome the spring prediction barrier in ENSO forecasting. The research emphasizes global feature modeling and proposes a novel content-guided ENSO prediction model. It also includes an ocean data generation model utilizing global attention. Furthermore, a layered rendering technique is employed to invert ocean data, facilitating detailed analysis and contributing to the development of an ocean synthetic dataset.
This work addresses the distributed predefined-time cooperative formation of heterogeneous multiagent systems comprising unmanned surface vehicles (USVs) and unmanned aerial vehicles (UAVs) with inherent dynamic uncertainties. By transforming the model coordinates, the underactuated heterogeneous USV-UAV systems can be converted into a fully actuated second-order multiagent framework. Subsequently, a predefined-time dynamic observer is designed to estimate the uncertain dynamics of each agent. Combining the backstepping method and the virtual leader model, a predefined-time distributed cooperative formation control based on uncertain dynamic estimation is designed for the heterogeneous USV-UAV systems. The convergence of the formation errors is rigorously demonstrated by constructing a suitable Lyapunov function under a predefined time framework. Ultimately, the two numerical cases in both the fixed and time-varying formation scenarios confirm the effectiveness of the constructed method.
Distributed acoustic sensing (DAS) is an emerging vibration signal acquisition technology that transforms existing fiber-optic communication infrastructure into an array of thousands of seismic sensors. Due to its advantages of low cost, easy deployment, continuous measurement, and long-distance measurement, DAS has rapidly developed applications in the field of marine geophysics. This paper systematically summarizes the status of DAS technology applications in marine seismic monitoring, tsunami and ocean-current monitoring, ocean thermometry, marine target monitoring, and ocean-bottom imaging; analyzes the problems faced during its development; and discusses prospects for further applications in marine geoscience and future research directions.
Large language models are widely used across various applications owing to their superior performance. However, their high computational cost makes deployment on edge devices challenging. Spiking neural networks (SNNs), with their power-efficient, event-driven binary operations, offer a promising alternative. Combining SNNs and transformers is expected to be an effective solution for edge computing. This study proposes an energy-efficient spike transformer accelerator, which is the base component of the large language models, for edge computing, combining the efficiency of SNNs with the performance of transformer models. The design achieves performance levels comparable to traditional transformers while maintaining the lower power consumption characteristic of SNNs. To enhance hardware efficiency, a specialized computation engine and novel datapath for the spike transformer are introduced. The proposed design is implemented on the Xilinx Zynq UltraScale+ ZCU102 device, demonstrating significant improvements in energy consumption over previous transformer accelerators. It even surpasses some recent binary transformer accelerators in efficiency. Implementation results confirm that the proposed spike transformer accelerator is a feasible solution for running transformer models on edge devices.
Refractive distortions in underwater images usually occur when these images are captured through a dynamic refractive water surface, such as unmanned aerial vehicles capturing shallow underwater scenes from the surface of water or autonomous underwater vehicles observing floating platforms in the air. We propose an end-to-end deep neural network for learning to restore real scene images for removing refractive distortions. This network adopts an encoder-decoder architecture with a specially designed attention module. The use of the attention image and the distortion field generated by the proposed deep neural network can restore the exact distorted areas in more detail. Qualitative and quantitative experimental results show that the proposed framework effectively eliminates refractive distortions and refines image details. We also test the proposed framework in practical applications by embedding it into an NVIDIA JETSON TX2 platform, and the results demonstrate the practical value of the proposed framework.
The pore structure of marine natural gas-hydrate-bearing sediments is a key factor related to the physical properties of reservoirs. However, the resolution of micro-computed tomography (micro-CT) images is unsuitable for the analysis of pore structures in fine-grained sediments. In this regard, super-resolution (SR) reconstruction technology is expected to improve the spatial resolution of micro-CT images. We present a self-supervised learning method that does not require high-resolution datasets as input images to complete the training and reconstruction processes. This method is an end-to-end network consisting of two subnetworks: an SR network and a downscaling network. We trained on a self-built dataset of hydrate samples from three different particle sizes. Compared with typical methods, the SR results indicate that our method provides high resolution while improving clarity. In addition, it has the highest consistency with the liquid saturation method with the subsequent calculation of porosity parameters. This study contributes to the investigation of seepage and energy transfer in sediments containing natural gas hydrates, which is particularly important for the exploration and development of marine natural gas hydrate resources.
Accurate and robust initialization is significant for visual-inertial simultaneous localization and mapping (VI-SLAM). Existing methods solve VI-SLAM initialization based on visual information. However inertial measurement unit (IMU) parameter estimation performed underwater is subject to two major limitations. First, IMU preintegration error accumulates over time, resulting in reduced accuracy. Second, it is difficult for robots to achieve sufficient movement underwater, which affects the reliability of initialization results. For a better balance between the efficiency and accuracy of VI-SLAM initialization, this study proposes a VI-SLAM initialization method using a designed marker calibration device. First, we utilize both marker points and ORB feature points for a fast and robust visual trajectory estimation with real motion scale, and we estimate the gravity direction using the marker calibration device. Second, the IMU trajectory is aligned with the trajectory, and the IMU parameters are solved using the initial gravity direction. Experiments verify the effectiveness of our developed method for improving the accuracy and efficiency of the VI-SLAM initialization. The code is available at https://gitee.com/litseaak/mmorb.
The rise in oil extraction and transportation in marine environments has led to frequent oil spill incidents, posing a severe threat to marine ecosystems and becoming an urgent environmental issue. This paper presents a laser-induced fluorescence light detection and ranging (LiDAR) system specifically designed for monitoring marine oil spills. The system comprises a laser emission module, a receiving module, a data processing module, and a wireless transmission module. Through outdoor experiments, the system has demonstrated its effectiveness and reliability in detecting and identifying various oil types, including crude oil, diesel, heavy oil, gasoline, and lubricating oil. Additionally, a BP neural network model was employed to process the fluorescence spectral data collected by the LiDAR system. This model successfully predicted oil types with an accuracy of 96.58%. This research presents a new technological solution for marine oil spill monitoring, offering significant potential for practical applications and further research.
Exploring the underwater world presents significant physical and financial barriers. Telepresence technologies offer a potential solution by a providing a more accessible version of an underwater experience. Our research involved a scoping review to consolidate previous findings on vision-based technologies that aim to recreate underwater experiences and their user evaluations. We searched 5 academic databases for papers describing or evaluating technologies providing visual underwater experiences without actual submersion. We systematically searched YouTube to include immersive experiences not documented in academic publications. Our review included 45 academic papers and 23 YouTube videos classified by their level of ‘reality’, ‘degrees of freedom’, and presence of interactive elements. The technologies reviewed included virtual reality, 360 video and imaging, augmented and mixed reality, head-mounted displays, mobile devices, cameras, sensors, remotely operated vehicles (ROVs), and games. Half of the selected papers featured user evaluations (with sample sizes ranging from 5 to 1006 participant); these methods included interviews, performance tracking, and questionnaires. We identified six main application areas for these technologies: (i) general environmental awareness, (ii) formal education about marine life, (iii) therapeutic interventions, (iv) access to underwater heritage sites, (v) ROV teleoperation and simulations, and (vi) entertainment. Immersive technologies, such as head-mounted displays and augmented reality, were prevalent across all application categories, though their usability varied. Cost considerations were also diverse, with costs ranging from expensive ROVs and simulated environments to cheaper 360 videos. Our findings indicate a need for more robust user studies, including long-term research and comparisons among real-time, pre-recorded, and simulated experiences. A better understanding of entertainment-driven applications could benefit education, environmental conservation, and healthcare. The findings of the scoping review are discussed with respect to the technologies identified and the corresponding user studies.
Coral reefs are essential ecosystems in the vast expanses of oceans, nurturing various forms of marine life within their vibrant and expansive structures. However, these underwater paradises suffer considerable threat from the population explosions of crown-of-thorns starfish (COTS), which detrimentally affect scleractinian corals across the Indo-Pacific region. This study addresses the early drawback of solely relying on texture analysis for COTS detection, recognizing the associated insufficiency due to variability in reef substrates. By integrating multiresolution analysis employing wavelet transform, edge information, and texture analysis using gray-level co-occurrence probability, this approach employs crucial Haralick features refined for pattern recognition. This enables a more detailed understanding of COTS traits, including the detection of the numerous sharp spines that cover their upper bodies. This approach considerably enhances classification reliability, making notable progress with an impressive accuracy of 95.00% using the eXtreme Gradient Boosting (XGBoost) Classifier. Moreover, this model streamlines processing requirements by increasing computational and memory efficiencies, making it more resource-efficient than the current models. This advancement enhances detection and opens avenues for early intervention and future research. Furthermore, integrating the model with underwater imagery could enable citizen science initiatives and autonomous underwater vehicle (AUV) surveys. Empowering trained volunteers and equipping AUVs with this technology could considerably expand coral reef monitoring efforts. Early COTS outbreak detection allows for shorter response times, potentially mitigating the damage and facilitating targeted conservation strategies.
Underwater sound classification presents a unique challenge due to the complex propagation characteristics of sound in water, including absorption, scattering, and refraction. These complexities can distort and alter spectral features, hindering the effectiveness of traditional feature extraction methods for vessel classification. To address this challenge, this study proposes a novel feature extraction method that combines Mel-frequency cepstral coefficients (MFCCs) with a spectral dynamic feature (SDF) vector. MFCCs capture the spectral content of the audio signal, whereas SDF provides information on the temporal dynamics of spectral features. This combined approach aims to achieve a more comprehensive representation of underwater vessel sounds, potentially leading to improved classification accuracy. Validation with real-world underwater audio recordings demonstrated the effectiveness of the proposed method. Results indicated an improvement of up to 94.68% in classification accuracy when combining SDF with several classical extractors evaluated. This finding highlights the potential of SDF in overcoming the challenges associated with underwater sound classification.
Current research on energy management strategies (EMSs) often neglects the impact of system topology and local control. This study tackles this issue by optimizing the topology of the hybrid power system on the ’FCS Alsterwasser’ cruise ship and enhancing EMS performance using various local controllers. First, the paper outlines the objectives of the research and provides an analysis of the current domestic and international research status. Second, the methods used in this study are introduced, including the topology optimization method and EMS. Subsequently, a model of the hybrid power system is constructed and verified through simulations. Finally, the effectiveness of different strategies is evaluated according to simulation results. Compared with an EMS based solely on a proportional-integral controller, the combination of a state machine and droop controller achieves better results, reducing battery power fluctuations by 86.5% and fuel cell power fluctuations by 16.2%.
In this paper, an adaptive fuzzy disturbance observer (FDO)-based control is proposed for the swing suppression of the mooring-heavy lift crane (HLC)-cargo system in the presence of a wave disturbance. First, the dynamic model of the HLC system is determined by employing the Lagrangian method, and the wave force is modeled as an exosystem with unknown terms based on the Jonswap spectrum. Then, based on the HLC model and the wave force exosystem, an FDO is established to determine the wave disturbances, and a fuzzy approximator is developed to estimate the unknown terms. A novel disturbance estimation error observer is first developed to facilitate the parameter adaptive updating law. Subsequently, by augmenting the HLC system and disturbance estimation error system, an FDO-based fuzzy antiswing control method is proposed in terms of the linear matrix inequality technique to suppress the swing. The closed-loop system stability is examined by using the Lyapunov method. Finally, the effectiveness of the proposed method is validated by numerical simulation.
This paper proposes an ocean front database and a method for its construction tailored for studying the dynamic evolution of ocean fronts. Ocean fronts play a crucial role in the interactions between the ocean and atmosphere, affecting the transfer of heat and matter in the ocean. In recent years, research on ocean fronts has emerged as a significant and rapidly evolving area within oceanography. With the development of ocean remote sensing technology, the amount of available ocean remote sensing data has been increasing. However, the potential of this expanding volume of ocean front data remains largely untapped. The lag in data processing technology has hindered research progress in understanding ocean fronts despite the growing amount of data available. To bridge this gap, this paper proposes an ocean front dynamic evolution database along with a method for its construction to further promote research into the variations and interactions of ocean fronts. This is especially relevant for studies utilizing deep learning to explore the dynamic evolution of ocean fronts. Specifically, the proposed database is designed to capture the variation processes of ocean front enhancement and attenuation, as well as the interactions during ocean front splitting and merging. The proposed database construction method allows for the segmentation and extraction of specific ocean fronts of interest from ocean front images. The proposed method is beneficial for analyzing the dynamic evolution between multiple ocean fronts on the same timeline.
In the field of underwater acoustics, forward-looking sonar represents a pivotal tool for acquiring subaqueous imagery. However, this technique is susceptible to the inherent ambient noise prevalent in underwater environments, resulting in degraded image quality. A notable challenge in this domain is the scarcity of pristine image exemplars, making it difficult to apply many advanced deep denoising networks for the purification of sonar images. To address this issue, the study introduces a novel self-supervised methodology specifically designed for denoising forward-looking sonar images. The proposed model employs a blind-spot network architecture to reconstruct unblemished images. Additionally, it integrates wavelet transform technology within a convolutional neural network (CNN) framework, combining frequency and structural information. Furthermore, the model incorporates contrastive regularization to augment denoising efficiency. This innovative denoising network, which leverages wavelet transform and contrastive regularization (CR), is henceforth referred to as WTCRNet. To evaluate the performance of WTCRNet, this study constructs a dual dataset comprising both simulated and authentic forward-looking sonar images, thereby furnishing a comprehensive dataset for network training and evaluation. Empirical assessments conducted on these datasets demonstrate that WTCRNet substantially outperforms existing denoising methodologies by effectively mitigating noise. The code is available at https://gitee.com/sichengling/wtcrnet.git.
SeaOrbiter’s vision is to become the first International Space Station of the ocean, a symbol of humanity's capacity to explore and understand the most uncharted and critical part of our planet. SeaOrbiter will bring together scientists, explorers, and innovators from around the world to conduct ground-breaking research, notably in the context of global change, explore the unknown, and develop new technologies to better understand our ocean, its fine relationship with the atmosphere and its complex ecosystems. With its unique design and innovative operational mode–a quiescent slow drifting pace–SeaOrbiter aims to become an iconic symbol of human’s commitment to protecting our planet's most valuable resource: the ocean.
The influence of the viscosity on a wave attractor flow has been previously studied, particularly in relation to the widening of the hydrodynamical structures. In this work, we simulate an attractor flow with a peculiar bottom shape that includes an underwater hill. During the simulation, we discovered a side structure appearing beyond the wave attractor. We determined that the appearance of this structure is connected to viscosity. In this article, we consider the behavior of this newly found structure. Additionally, we discuss the challenges of energy accumulation and the estimation of the Reynolds number, which is a non-trivial problem in the context of wave attractor flows.
Earthquake data are one of the key means by which to explore our planet. At a large scale, the layered structure of the Earth is revealed by the seismic waves of natural earthquakes that go deep into its inner core. At a local scale, seismology for exploration has successfully been employed to discover massive fossil energies. As the volume of recorded seismic data becomes greater, intelligent methods for processing such a volume of data are eagerly anticipated. In particular, earthquake focal mechanisms are important for assessing the severity of tsunamis, characterizing seismogenic faults, and investigating the stress perturbations that follow a major earthquake. Here, we report a novel deep reinforcement learning method for inverting the earthquake focal mechanism. Unlike more typical deep learning applications, which require a large training dataset, a deep reinforcement learning system learns by itself. We demonstrate the validity and efficacy of the proposed deep reinforcement learning method by applying it to the Mw 7.1 mainshock of the Ridgecrest earthquakes in southern California. In the foreseeable future, deep learning technologies may greatly contribute to our understanding of the oceanographic process. The proposed method may help us understand the mechanism of marine earthquakes.
The rapidly growing exploitation and utilization of marine resources by humans has sparked considerable interest in underwater object detection tasks. Targets captured in underwater environments differ significantly from those captured in general images owing to various factors, such as water turbidity, complex background conditions, and lighting variations. These adverse factors pose a host of challenges, such as high intensity noise, texture distortion, uneven illumination, low contrast, and limited visibility in underwater images. To address the specific difficulties encountered in underwater environments, numerous underwater object detection methods have been developed in recent years in response to these challenges. Furthermore, there has been a significant effort in constructing diverse and comprehensive underwater datasets to facilitate the development and evaluation of these methods. This paper outlines 14 traditional methods used in underwater object detection based on three aspects that rely on handmade features. Thirty-four more advanced technologies based on deep learning were presented from eight aspects. Moreover, this paper conducts a comprehensive study of seven representative datasets used in underwater object detection missions. Subsequently, the challenges encountered in current underwater object detection tasks were analyzed from five directions. Based on the findings, potential research directions are expected to promote further progress in this field and beyond.
Solving differential equations is one of the most promising applications of quantum computing. The Poisson equation has applications in various domains of physics and engineering, including the simulation of ocean current dynamics. Here, we propose an efficient quantum algorithm for solving the one-dimensional Poisson equation based on the controlled R y rotations. Our quantum Poisson solver (QPS) removes the need for expensive routines such as phase estimation, quantum arithmetic or Hamiltonian simulation. The computational cost of our QPS is 3n in qubits and 5/3n 3 in one- and two-qubit gates, where n is the logarithmic of the number of discrete points. An overwhelming reduction of the constant factors of the big-O complexity is achieved, which is critical to evaluate the practicality of implementing the algorithm on a quantum computer. In terms of the error ε, the complexity is log(1/ε) in qubits and poly(log(1/ε)) in operations. The algorithms are demonstrated using a quantum virtual computing system, and the circuits are executed successfully on the IBM real quantum computers. The present QPS could exhibit a potential real-world application for solving differential equations on noisy intermediate-scale quantum (NISQ) devices.
Due to its importance in marine engineering and aquatic robotics, underwater image enhancement works as a preprocessing step to improve the performance of high-level vision tasks such as underwater object detection and recognition. Although several studies exhibit that underwater image enhancement algorithms can boost the detection accuracy of detectors, no work has focused on studying the relationship between these two tasks. This is mainly because current underwater datasets lack either bounding box annotations or high-quality reference images, based on which detection accuracy or image quality assessment metrics are calculated. To examine how underwater image enhancement methods affect underwater object detection tasks, we provide a large-scale underwater object detection dataset with both bounding box annotations and high-quality reference images, namely, the WaterPairs dataset. The WaterPairs dataset offers a platform for researchers to comprehensively study the influence of underwater image enhancement algorithms on underwater object detection tasks. We will release our dataset at https://github.com/IanDragon/WaterPairs once this paper is accepted.
To compute the vibroacoustic radiation of the shaft-hull system in an ocean acoustic channel, we apply the finite element method/boundary element method. For the convergence zone effect, we use the beam displacement ray normal mode to examine the profile of the speed of sound in seawater. To evaluate the radiation property of the system, we propose the concept of effective radiated acoustic power, where only reversing acoustic rays and seabed totally reflected rays are included. We reveal that the far-field radiation can be approximately inferred from the effective radiated acoustic power. Effective radiated acoustic power can be utilized to extract the main radiating modes, which dominate the far-field radiation. The axial force induces vibration, including breathing and bending modes, resulting in the highest peak of the radiated sound pressure level (SPL). The lateral and vertical excitation results in horizontal and vertical bending modes, respectively, and the SPL in convergence zones induced laterally are higher than those induced vertically.
Fine-grained image classification of marine organisms involves dividing subcategories within a larger category. For instance, this could mean distinguishing specific species of fish or types of algae. This type of classification is more intricate than regular image classification, as the minor feature differences between subcategories are often concentrated in one or a few specific areas. Therefore, accurately identifying these critical regions and effectively using local features are crucial in improving the accuracy of fine-grained image classification. Existing methods for fine-grained image classification primarily rely on single-branch models based on convolutional neural networks (CNNs) or vision transformers (ViTs). Consequently, merging them allows for a more comprehensive understanding of marine organism images. In addition, marine organism images are affected by the distance and angle of the shot, making it challenging to capture detailed local nuances at a single scale. To address these challenges, we propose a multi-scale dual-branch network (MSDBN) that combines the strengths of ViT and CNN for fine-grained image classification of marine organisms. Our model uses a novel two-stage selection module to select discriminative regions from the ViT branch. Following this, the CNN branch executes a more detailed feature extraction on the local regions. To effectively utilise the multi-scale information of marine organisms, we introduce our designed multi-scale shift-window self-attention, specifically for the ViT branch. MSDBN demonstrates improved performance compared to existing classical methods and the best-performing dual-branch methods on three marine datasets. Our code is released publicly at https://github.com/Xiaosigz/MSDBN.
The exponential progression in oceanic observational technology has fostered the accumulation of substantial time series data pivotal for predictions in ocean meteorology. Foremost among the phenomena observed is El Niño-Southern Oscillation (ENSO), a critical determinant in the interplay of global ocean atmosphere interactions, with its severe manifestations inducing extreme meteorological conditions. Therefore, precisely predicting ENSO events carries immense gravitas. Historically, predictions hinged primarily on dynamic models and statistical approaches; however, the intricate and multifaceted spatiotemporal dynamics of ENSO events have often impeded the accuracy of these traditional methodologies. A notable lacuna in contemporary research is the insufficient exploration of long-term dependencies within oceanic data and the suboptimal integration of spatial information derived from spatiotemporal data. To address these limitations, this study introduces a forward-thinking ENSO prediction framework synergizing multiscale spatial features with temporal attention mechanisms. This innovation facilitates a more profound exploration of temporal and spatial domains, enhancing the retention of extensive-period data while optimizing the use of spatial information. Preliminary analyses executed on the global ocean data assimilation system dataset attest to the superior efficacy of the proposed method, underscoring a substantial improvement over established methods including SA-convolutional long short-term memory, particularly in facilitating long-term predictions.The source code and datasets are provided. The code is available at https://github.com/tse1998/ENSO-prediction.
To achieve the requirements of lightweight, low energy consumption, and low inertia of an underwater vehicle manipulator system, a cable-driven manipulator is installed on the underwater vehicle to form a cable-driven flexible-joint-based underwater vehicle manipulator system (CDFJ–UVMS). The CDFJ–UVMS is a complex nonlinear system subject to model uncertainties, complex marine environment disturbances, and actuator dead-zone nonlinearity. To design track controllers, the CDFJ–UVMS dynamics is divided into two parts: known and unknown. Subsequently, a radial basis function neural network is adopted to approximate the unknown nonlinearity. A neural network performance observer is constructed, whose estimation error is then used to design a novel neural disturbance observer (NDO) to estimate the total disturbance. Finally, an adaptive neural network control method is proposed for the CDFJ–UVMS based on the NDO, neural network compensator, and neural performance observer. The stability of the closed-loop system is analyzed using the Lyapunov method. The proposed control algorithm is applied to a CDFJ–UVMS with two cable-driven joints and compared with other control methods to show the effectiveness of the proposed control algorithm.
Sea surface temperature (SST) prediction is a subject of great significance to the marine environment and human society. Changes in SST not only impact marine ecosystems and fishery resources but also trigger extreme weather events and disastrous consequences. Therefore, the precise prediction of SST is essential to avoiding these problems. Although numerous data-driven SST prediction models have emerged in recent years, these models are characterized by a lack of physical mechanisms related to sea temperature changes as well as insufficient generalization capabilities and interpretability. In our work, attempts were made to integrate physics-related convection phenomena into deep learning models, and traditional deep learning models were improved by incorporating time and space attention modules. The results of a series of experiments showed that the incorporation of physical mechanisms enhanced the performance of data-driven models. Furthermore, attention mechanisms were similarly helpful, of which temporal attention proved to be more important. The modules proposed in this work also improved the baseline model’s accuracy by 22%. In addition, seven-day SST predictions were carried out for the world’s five major fishing grounds. The results demonstrated that the application of transfer learning strategies yielded superior performance, further improving prediction accuracy by 1%–5%.
Underwater source localization, such as matched-field processing (MFP), triangulation, and waveguide invariant, have been extensively investigated in temperate oceans. Seasonal or yearlong ice floes or ice cover exist in high-latitude sea regions and the polar ocean. In under-ice shallow water, sound interacts frequently with ice and sea bottom, which results in dramatic reflection, attenuation, and modal dispersion. The boundary effects generate more uncertainties in model-based source localization methods, for example, the MFP method. In this work, we develop the preliminary scheme of the under-ice MFP. The performances of the incoherent Bartlett and minimum variance algorithms are verified by real data collected by a 12-element Vertical line array with a space of 1 m for a source (650–750 Hz) at 2.7-km range and 5-m depth in the northern Yellow Sea in winter. The experimental findings demonstrate that the range error is within 2% and the depth error is within 10%. The error primarily originates from the uncertainty of sea bottom parameters.