Deep learning-based data-driven methods have gained significant attention in underwater acoustic source localization. However, their performance is often constrained by environmental disturbances and the scarcity of real-world underwater acoustic data. To address these issues, this paper presents a novel network termed MTCL-Net, a multi-task learning network that incorporates contrastive learning as an auxiliary task for underwater acoustic source ranging. A standard dataset and a perturbed dataset to simulate real underwater interferences are constructed based on known environmental parameters in this method. A Siamese dual-branch architecture is employed, where a contrastive learning task enables the automatic extraction of position-related features. The network jointly optimizes three tasks: source localization in perturbed environments, localization on the standard dataset, and position similarity discrimination, which improves the robustness and generalization ability. The experimental results on simulated and sea trial data demonstrate that MTCL-Net outperforms traditional matched field processing (MFP), single-task learning (STL), and multi-task learning based on depth–range (MTL-DR) methods in terms of mean absolute error (MAE) and probability of credible localization (PCL-10%). Specifically, on SWellEx-96 sea trial data, MTCL-Net achieves an MAE of 0.17 km and a PCL-10% of 90.36%. Moreover, the proposed method only needs a few samples for fine-tuning and shows strong practicality in uncertain marine environments.

Seabed sediment detection is essential for obtaining seabed environmental information, supporting seabed resource exploitation, marine engineering monitoring, and underwater acoustic research. To overcome limitations of traditional station-based detection methods, such as bulky equipment and low efficiency, our team proposes an expendable in-situ technique for rapid sediment measurement. To investigate the motion characteristics of the measuring instrument in complex marine environment and its initial state of seabed penetration, this paper derives a dynamic model of underwater falling motion based on the momentum and angular momentum theorems of six-degree-of-freedom rigid body motion. Computational fluid dynamics (CFD) methods are then employed to determine the required hydrodynamic coefficients, which are incorporated into the dynamic model to predict the underwater motion state of the instrument. Based on this prediction method, the influence patterns of initial release conditions, currents, structural shape and mass distribution of the instrument on the falling motion characteristics are revealed. A field drop test validates the method, with predicted results in reasonable agreement with experimental data (maximum error 26.8% in vertical falling velocity). The findings provide critical initial-state inputs for predicting seabed penetration performance and offer theoretical support for structural optimization and hydrodynamic behavior analysis.

Mesoscale eddies significantly influence underwater acoustic propagation by modifying sound speed distributions. Existing studies have predominantly focused on the acoustical effects of surface-intensified eddies detectable via satellite altimetry. However, the acoustic impacts of subsurface-intensified eddies remain unclear due to the difficulty of observation. This study investigates a lens-shaped subsurface-intensified eddy (LS-SSE) observed by a spray glider off Mindanao in 2010, systematically examining its oceanographic and acoustic characteristics. The LS-SSE contains warm, salty water, exhibiting maximum temperature and salinity anomalies of 1 °C and 0.2 psu near the 26.75σθ isopycnal. These thermohaline anomalies substantially alter the local sound speed field, creating a subsurface acoustic duct (LS-SSE-Duct) between 250–350 m. Dynamic decomposition indicates that this duct is primarily attributed to the “spice sound speed” component, associated with water masses advected by the LS-SSE along isopycnals. Numerical simulations using decomposed “tilt” and “spice” sound fields reveal their distinct effects on acoustic propagation. In the tilt field, the upper sound energy turning points shoal by ∼50 m relative to the unperturbed field, and the first and second convergence zones shift forward by ∼4 km. In the spice field, the LS-SSE-Duct could trap sound energy with weak attenuation over ∼220 km. Acoustic modes within the LS-SSE-Duct shift to higher orders with increasing horizontal range away from the sound source, leading to the transfer of ducted energy to the shallower water layers. Beyond 80 km, average transmission loss (TL) increases with frequency. Frequency-dependent TL variations are mainly associated with mode coupling, and the contribution of diffractive effect is negligible.

The characteristics of seabed sediments have an important influence on the spatiotemporal structure of the underwater acoustic field in shallow sea. However, accurate geoacoustic parameters obtained is facing challenge in generally, raising difficulties for application of matched field processing (MFP) technology. In this paper, the Bartlett processor was used to carry out for vertical linear array (VLA) localization simulated in a shallow sea environment, and the MFP performance for localization was evaluated under eight types of sediments. Research indicates that great differences in the effects of different types of sediments on the underwater acoustic field, resulting in a poor localization performance for some sediments even if their accurate acoustic parameters can be obtained. The simulated localization results of mismatched sound velocity in different types of sediments show that the larger sediment sound speed, the better performance of MFP for localization, while the performance is also relied on the frequency in the same type of sediments. With the comparison of type and energy for normal modes excited in different types of sediments, the research further provided theoretical explanations for above result from the perspective of normal mode theory in shallow sea environments. This result is helpful for practical engineering application of sonar array in uncertain environmental parameters, especially uncertain geoacoustic parameters.

The surface current system in the tropical western Pacific is pivotal for regulating local-to-global air-sea interactions by modulating ocean-atmosphere heat and mass exchange. However, two key knowledge gaps persist: the evolution of the region's hydro-ecological conditions (e.g., nutrient dynamics, phytoplankton biomass) and their response mechanisms to the El Nino-Southern Oscillation (ENSO) cycle. To address these, this study analyzed eco-hydro-climatic variations in the tropical western Pacific during ENSO events (2003-2022). Results show ENSO modulates surface chlorophyll-a (Chl-a) via a "three-level linkage mechanism": ENSO signals first alter physical processes (sea level anomalies [SLA], boundary current velocities), which then regulate nutrient supply (vertical mixing, horizontal transport), ultimately driving differentiated Chl-a responses across subregions. Specifically, during El Nino: decreased SLA weakens upper-ocean stratification to enhance vertical nutrient supply; boundary currents (NGCC, NECC) show intensity-dependent responses (decreasing in weak-to-strong El Nino, increasing in the 2015/2016 super El Nino), leading to significant Chl-a increases in the Western Pacific Warm Pool (WP) and Halmahera Eddy (HE) regions. During La Nina: stable SLA limits vertical nutrient supply, but enhanced westward currents compensate via horizontal transport to sustain WP region Chl-a growth; the HE region shows no Chl-a change due to reduced boundary current-driven nearshore nutrient input. Nutrient sensitivity further shapes responses: nitrate responds to super El Nino and all La Nina events, phosphate to moderate-to-strong La Nina, and silicate (HE region's key limiting nutrient) exhibits ENSO-linked extreme values. Open-ocean regions (A, B) and the Mindanao Eddy (ME) region have weak/no Chl-a-ENSO correlation, due to persistent stratification (A, B region) or weak eddy-driven supply (ME region). This study clarifies the ENSO-Chl-a cascade in the tropical western Pacific, filling gaps in hydro-ecological response understanding. By highlighting boundary current dynamics, sub-regional heterogeneity, and nutrient sensitivity, it provides a theoretical framework for predicting marine ecosystem changes under climate variability.

A large number of seabed depressions, covering an area of 2500 km2 in the Xisha Massif of the South China Sea, are investigated using newly collected high-resolution acoustic data. By analyzing the morphological features and seismic attributes of the focused fluid flow system, five geological structures are recognized and described in detail, including pockmarks, volcanic mounds, pipes, faults, and forced folds. Pockmarks and volcanic mounds occur as clustered groups and their distributions are related to two large-scale volcanic zones with chaotic seismic reflections. Pipes, characterized by disordered seismic reflections, mainly occur within the focused fluid flow zone (FFFZ) and directly link with the large-scale deep volcano and its surrounding areas. Faults and fractures mainly occur along pipes and extend to the seafloor, commonly presenting lateral walls of mega-pockmarks. Forced folds are primarily clustered above volcanic zones and commonly restricted between faults or pipes, characterized by sediment deformations as indicated in seismic profiles. By comprehensive analysis of the above observations and a simplified simulation model, the volcanism-induced hydrothermal fluid activities are argued herein to contribute to these focused fluid flow structures. In addition, traces of suspected submarine instability disasters such as landslides have been found in this sea area, and more observational data will be needed to determine whether seafloor fluid flow zones can be used as a predictor of seafloor instability in the future.

To address the limitations in identifying complex anomaly patterns and the heavy reliance on manual labeling in traditional oceanographic data quality control (QC) processes, this study proposes an intelligent QC method that integrates Gated Recurrent Units (GRU) with a Mean Teacher–based semi-supervised learning framework. Unlike conventional deep learning approaches that require large amounts of high-quality labeled data, our model adopts an innovative training strategy that combines a small set of labeled samples with a large volume of unlabeled data. Leveraging consistency regularization and a teacher–student network architecture, the model effectively enhances its ability to learn anomalous features from unlabeled observations. The input incorporates multiple sources of information, including temperature, salinity, vertical gradients, depth one-hot encodings, and seasonal encodings. A bidirectional GRU combined with an attention mechanism enables precise extraction of profile structure features and accurate identification of anomalous observations. Validation on real-world profile datasets from the Bailong (BL01) moored buoy and Argo floats demonstrates that the proposed model achieves outstanding performance in detecting temperature and salinity anomalies, with ROC-AUC scores of 0.966 and 0.940, and precision–recall AUCs of 0.952 and 0.916, respectively. Manual verification shows over 90% consistency, indicating high sensitivity and robust generalization capability under challenging scenarios such as weak anomalies and structural profile shifts. Compared to existing fully supervised models, the proposed semi-supervised QC framework exhibits superior practical value in terms of labeling efficiency, anomaly modeling capacity, and cross-platform adaptability.

As sound speed is a fundamental parameter of ocean acoustic characteristics, its prediction is a central focus of underwater acoustics research. Traditional numerical and statistical forecasting methods often exhibit suboptimal performance under complex conditions, whereas deep learning approaches demonstrate promising results. However, these methodologies fall short in adequately addressing multi-spatial coupling effects and spatiotemporal weighting, particularly in scenarios characterized by limited data availability. To investigate the interactions across multiple spatial scales and to achieve accurate predictions, we propose the STA-ConvLSTM framework that integrates spatiotemporal attention mechanisms with convolutional long short-term memory neural networks (ConvLSTM). The core concept involves accounting for the coupling effects among various spatial scales while extracting temporal and spatial information from the data and assigning appropriate weights to different spatiotemporal entities. Furthermore, we introduce an interpolation method for ocean temperature and salinity data based on the KNN algorithm to enhance dataset resolution. Experimental results indicate that STA-ConvLSTM provides precise predictions of sound speed. Specifically, relative to the measured data, it achieved a root mean square error (RMSE) of approximately 0.57 m/s and a mean absolute error (MAE) of about 0.29 m/s. Additionally, when compared to single-dimensional spatial analysis, incorporating multi-spatial scale considerations yielded superior predictive performance.

The low-wavenumber components in the gradient of full waveform inversion (FWI) play a vital role in the stability of the inversion. However, when FWI is implemented in some high frequencies and current models are not far away from the real velocity model, an excessive number of low-wavenumber components in the gradient will also reduce the convergence rate and inversion accuracy. To solve this problem, this paper firstly derives a formula of scattering angle weighted gradient in FWI, then proposes a hybrid gradient. The hybrid gradient combines the conventional gradient of FWI with the scattering angle weighted gradient in each inversion frequency band based on an empirical formula derived herein. Using weighted hybrid mode, we can retain some low-wavenumber components in the initial low-frequency inversion to ensure the stability of the inversion, and use more high-wavenumber components in the high-frequency inversion to improve the convergence rate. The results of synthetic data experiment demonstrate that compared to the conventional FWI, the FWI based on the proposed hybrid gradient can effectively reduce the low-wavenumber components in the gradient under the premise of ensuring inversion stability. It also greatly enhances the convergence rate and inversion accuracy, especially in the deep part of the model. And the field marine seismic data experiment also illustrates that the FWI based on hybrid gradient (HGFWI) has good stability and adaptability.

Seafloor sediment surveys play a crucial role in providing invaluable information and guidance for ocean monitoring and management. In recent years, the combination of multibeam echosounder (MBES) data and field samples has become one of the most widespread approaches for seafloor sediment supervised classification. However, owing to the limitations of low efficiency, high complexity and high cost of seabed in situ sampling, ground truth sample data tend to have a small size, which impedes the training and deployment of classifiers. In light of the aforementioned issues, this paper proposed a seabed sediment sample enhancement method on the basis of superpixel segmentation and active learning. First, we introduce superpixel multi-resolution segmentation to effectively expand the training sample set, which takes into account the different weights of the input features. Subsequently, poor samples in the new sample set are removed using an active learning technique to improve the quality of labels. In the end, 33 original sample points are augmented to a total of 6097 valid sample points. To validate the effectiveness of our strategy, the new sample set is employed for the supervised classification of MBES data in the Southern North Sea, UK. The experimental results show that the classification accuracy of our sample enhancement method reaches 86.67%, which is a significant improvement over that of the original sample set and traditional sample enhancement methods.

Seabed sediment classification is meaningful for seafloor habitat mapping and marine resources exploration. Multibeam Echo-sounding System can acquire backscatter and topographic information, which has become the mainstream for detecting seabed sediment. Because topographic features are closely correlated with sediment distribution, and sediment types tend to be distributed in a continuous pattern, we propose a seabed sediment classification based on topographybased image partitioning. Multi-resolution segmentation and K-means clustering are utilized for image partition based on topographic features, then the classification maps of different subregions are integrated as one result map. The experimental results show that the partitioned methods considerably outperform the global methods in terms of accuracy. This method provides ideas for achieving high-resolution seafloor mapping in the future.

为了提高有雾场景下船只检测识别的准确率,本文运用四分法计算求解大气光值实现对暗通道先验去雾算法的优化,引入空洞卷积方法和K-means++聚类算法改进YOLOv4 算法,提出改进的暗通道先验去雾算法和改进YOLOv4 深度学习的船只检测方法.通过与不同去雾算法和船只识别算法进行对比实验分析,改进后的方法更好地实现了海面船只的实时检测及分类识别.实验结果表明该方法解决了原去雾算法中去雾图像亮度偏暗等问题,提高了船只识别的准确率与实时性,对海上有雾环境条件下的船只实时检测研究具有一定的理论指导意义.

Seafloor habitat mapping plays an important role in marine environment monitoring and marine geological research. Optic and acoustic remote sensing are becoming common survey tools in seafloor habitat mapping. However, a single acoustic or optic technique may have a limited detection range and be more susceptible to the impact of image quality. Additionally, it is challenging to satisfy the requirements for accurate detection since single-source data cannot fully reflect the substrate distribution characteristics. This article developed a method for detecting coastal seafloor habitats through the fusion of multiscale optics and acoustics data. First, the original feature set was composed of multispectral satellite data and bathymetric data by multibeam echo sounder and airborne light detection and ranging at different scales, which improved the capacity to represent feature information. Then, a ReliefF–mRMR method was implemented to select optimal features with appropriate scales and remove redundant features. Finally, the optimal features were employed in model training and classification of several supervised classifiers to verify the effectiveness of the strategy. The developed method was applied to the Ganquan Island survey in the South China Sea. The results demonstrated that, after integrating multisource data, the accuracies were up to 3.31% and 17.28% higher than those obtained using multispectral data or bathymetric data alone, respectively. ReliefF–mRMR exhibited better performance than other feature selection methods. The average coral coverage in the study area was estimated to range from 70.85% to 80.33%. This research highlights the greater potential of multisource data for precisely detecting seafloor habitats.

Hemipelagic calcareous ooze is a mixture of biogenic carbonate ooze and detrital terrigenous materials from continental weathering. Understanding its early diagenetic features is critical for geophysical measurement and geo-hazard assessment. However, the early diagenesis of hemipelagic calcareous ooze has rarely been studied due to a lack of samples and in-situ measurement in shallow strata. During IODP Expedition 368, hemipelagic calcareous oozes, up to hundreds of meters thick, were encountered in the shallow subsurface (< 400 m) at two sites (U1501 and U1505) close to the continent-ocean transition zone in the northern South China Sea margin. Here, we use coring and well logging data to perform rock physics diagnostics to characterize early diagenetic processes in hemipelagic calcareous ooze. The calcareous ooze in the study area is primarily chalky marl with the ratio of carbonate to carbonate plus siliciclastic minerals ranging 30

为提高岛礁海区声呐目标探测能力,对实际海洋环境中目标辐射噪声在岛礁海区的声传播现象进行研究.考虑岛礁区域斜坡地形对声呐作用距离的影响,采用抛物方程RAM方法,针对不同季节的声速剖面,以声呐优质因子为门限,仿真计算了岛礁斜坡背景下声呐在对不同深度目标探测时的有效作用距离.结果表明:声呐作用距离与水文条件密切相关,不同声呐在不同海洋地形环境中对不同位置目标的探测性能相差悬殊;由于冬季海洋环境的特殊性,表面声道的存在使得声呐在探测海面附近目标时,作用距离提高2～6倍;在斜坡顶部的目标,只能在距离相对较短的岛礁浅海区域进行探测,斜坡外缘目标深度越大,声呐探测效果越好;工作于表面波导内的声呐在300～1 800 Hz频带内,平均作用距离较小.

High-precision habitat mapping can contribute to the identification and quantification of the human footprint on the seafloor. As a representative of seafloor habitats, seabed sediment classification is crucial for marine geological research, marine environment monitoring, marine engineering construction, and seabed biotic and abiotic resource assessment. Multibeam echo-sounding systems (MBES) have become the most popular tool in terms of acoustic equipment for seabed sediment classification. However, sonar images tend to consist of obvious noise and stripe interference. Furthermore, the low efficiency and high cost of seafloor field sampling leads to limited field samples. The factors above restrict high accuracy classification by a single classifier. To further investigate the classification techniques for seabed sediments, we developed a decision fusion algorithm based on voting strategies and fuzzy membership rules to integrate the merits of deep learning and shallow learning methods. First, in order to overcome the influence of obvious noise and the lack of training samples, we employed an effective deep learning framework, namely random patches network (RPNet), for classification. Then, to alleviate the over-smoothness and misclassifications of RPNet, the misclassified pixels with a lower fuzzy membership degree were rectified by other shallow learning classifiers, using the proposed decision fusion algorithm. The effectiveness of the proposed method was tested in two areas of Europe. The results show that RPNet outperforms other traditional classification methods, and the decision fusion framework further improves the accuracy compared with the results of a single classifier. Our experiments predict a promising prospect for efficiently mapping seafloor habitats through deep learning and multi-classifier combinations, even with few field samples.

中建海底峡谷具有分段性,但分段的关键地貌特征、各段沉积充填及其控制因素缺乏精细描述和系统论证.综合利用高分辨率二维和三维地震资料,结合水深地貌数据,对中建海底峡谷地貌及沉积特征进行了详细分析,总结了其南北段沉积过程的主控因素.中建海底峡谷呈NW向顺直展布于广乐隆起与西沙隆起之间,以华光礁附近的地貌高点为拐点被分为南北两段.中建海底峡谷北段沉积体系包括重力流沉积(水道、席状沉积、滑塌体)和底流沉积(漂积体、环槽、谷槽),南段以重力流水道和海底扇为主.北段沉积体系受底流和重力流交互作用控制,底流自中中新世开始出现,改造重力流水道,使其出现侧向迁移或翼部不对称现象,上新世以后重力流作用减弱,底流作用增强,沉积物波和漂积体广泛发育;峡谷南段水道表现出侵蚀-沉积-废弃的沉积旋回,未见底流沉积现象.相对海平面变化导致碳酸盐生产率变化影响物源供应,从而控制水道沉积演化,碳酸盐台地的"高位溢流"作用决定水道在高水位时发育.

从岛礁斜坡地形条件下的声信号衰减和地形阻断效应分析出发,重点针对水下声场分布规律及其对声传播造成的影响开展研究.利用水声模型理论,结合某礁实测地形以及水文数据,建立岛礁斜坡地形下的多途声信道模型,基于Bellhop与RAM声学仿真方法,对不同地形下的声线轨迹、声传播损失以及信号时延等声场特性进行仿真分析,得出岛礁斜坡地形下的声场分布特征.研究结果表明:(1)岛礁斜坡地形是影响其声传播模式的关键因素;(2)斜坡外缘浅海区域的目标不易被岛礁斜坡顶端的声呐所探测;(3)陡坡地形对浅海声源的声传播有利,当声源深度足够大时,缓坡地形下的本征声线数目能够达到在陡坡地形下的5倍,对声传播有利.以上研究结果可为岛礁区水下声场的特性分析以及水下声学对抗等实践应用提供理论基础和技术参考.

High-resolution seismic and borehole data were used to investigate the sedimentological characteristics of carbonate accumulations in the Pearl River Mouth Basin (PRMB) of the South China Sea. Based on seismic morphologies and petrologic analysis, three scale types of oligo-mesophotic ICBs were identified to have developed ephemerally in the early Early Miocene. The first ICB type is represented by small ICBs with lateral extension in the kilometer range, characterized by mounded geometry, complex internal architectures and a high-frequency alternation of carbonates and volcaniclastics. The second ICB type has lateral extensions of tens of kilometers and convex-up reflections, featuring obvious vertical thickening compared to the surrounding siliciclastic deposits. The third ICB type consist of extensive platforms with hundreds of kilometers of lateral extension, onlapped by surrounding siliciclastic deposits. The correlation of seismic and well data allows us to reconstruct the evolutionary history of the three types of ICBs in the PRMB: All the ICBs built by large benthic foraminifera and/or coralline algae biota, occurred on or around submarine topographic highs such as volcanic edifices or inherited from basement rises, developed under stable subsidence rates between 120 m/Myr and 180 m/Myr, and extinguished suddenly in the late Early Miocene with an abrupt decrease in the subsidence rate or a change from subsidence to tectonic uplift. It is proposed that post-rift subsidence of the study area controlled the evolution of the ICBs. Except for ICBs, mixed carbonate-siliciclastic deposits are identified in well succession, characterized by two types of mixing, namely, compositional mixing (core-plug scale) and strata mixing (stratigraphic scale). These mixed systems generated high-frequency alternations of carbonate, siliciclastic and/or mixed deposits sequence stratigraphy. This work reveals in detail the typical features of seismic morphologies, lithofacies, internal architectures and evolutionary mechanism of ICBs and mixed systems in the PRMB, and thus has implications for predicting reservoirs in hydrocarbon exploration and understanding carbonate sedimentology in a siliciclastic-dominated passive marginal basin.

本文围绕海上无人值守平台周界安防技术需求,以800 m范围内的快艇、渔船、客轮和商船为航行目标构建运动目标跟踪监控系统,并集成开发了基于水下被动声学检测和水上云台控制的光学视频系统为一体的声光一体化监控系统,设计了一套基于深度学习的目标检测算法,实现了对水面航行目标的跟踪检测与分类识别.本文在目前基于单一方法进行水面目标跟踪的基础上进行了突破,系统综合了水听器远距离目标监控跟踪和近距离的水面视频监控目标识别,当两者同时确认目标,由声警器发出警报驱离可疑目标,系统将水下声学监控和水面视频监控有效结合,极大地降低了平台的误报概率,同时实验测试结果表明该系统可有效实现对4类水面航行目标的跟踪检测和特征提取,且改进后的算法可有效提升目标检测效果与识别精度,具有较好的鲁棒性和有效性.