In the field of ocean observation, we often face the challenge of the contradiction between the vast ocean environment and limited ocean sensor observations, resulting in significant sparsity in the acquired ocean sensor data. This sparse ocean sensor data typically exhibits characteristics such as discrete spatial distribution, discontinuous observation time, and vertical stratification with water depth variations. Current methods primarily employ rule-based quality control, time series modeling, or traditional graph neural networks for processing. This paper addresses the characteristics of sparse ocean sensor data, building upon these methods by further utilizing topological correlation and hierarchical feature modeling on a topological basis. It proposes a depth-aware heterogeneous spatiotemporal graph neural network (DAHSGNN) to achieve efficient anomaly detection and data correction for this type of data. DAHSGNN integrates discrete observation data along the depth axis using a local graph construction method. It employs hierarchical feature engineering to characterize the vertical stratification of the ocean. A Gaussian Hidden Markov Model is used to segment the water layers, and intra- and inter-layer trend features are extracted using a water layer probability-guided Transformer encoder. Then, a bidirectional long short-term memory deep sequence encoder captures the local dynamic context, thereby achieving fine-grained modeling of the ocean’s vertical stratification features. Finally, a heterogeneous graph autoencoder is used to reconstruct the site-level data distribution. Experiments were conducted using multiple environmental variables from the International Seabed Authority (ISA) DeepData database. Results show that DAHSGNN exhibits good cross-variable generalization ability, achieves higher reconstruction accuracy than baseline methods, and significantly improves anomaly detection performance.

Deep-sea submersibles, often featuring a symmetrical design for hydrodynamic stability, operate as safety-critical systems in extreme environments, where the tight dynamic coupling between subsystems like hydraulics and propulsion creates complex failure modes that are challenging to diagnose. A localized fault in one system can propagate, inducing anomalous behavior in another and confounding conventional single-system monitoring approaches. This paper introduces a novel unsupervised anomaly detection framework, the Dual-Stream Coupled Autoencoder (DSC-AE), designed specifically to address this cross-system fault challenge. Our approach leverages a dual-encoder, single-decoder architecture that explicitly models the normal coupling relationship between the hydraulic and propulsion systems by forcing them into a shared latent representation. This architectural design establishes a holistic and accurate baseline of healthy, system-wide operation. Any deviation from this learned coupling manifold is robustly identified as an anomaly. We validate our model using real-world operational data from the deep-sea submersible, including curated test cases of intra-system and inter-system faults. Furthermore, we demonstrate that the proposed framework offers crucial diagnostic interpretability; by analyzing the model’s reconstruction error heatmaps, it is possible to trace fault origins and their subsequent propagation pathways, providing intuitive and actionable decision support for submersible operation and maintenance. This powerful diagnostic capability is substantiated by superior quantitative performance, where the DSC-AE significantly outperforms baseline methods in detecting propagated faults, achieving higher accuracy and recall, among other performance metrics.

In this research, we examine how the Al Hoceima Marine Protected Area (MPA), located in the southwest Mediterranean Sea, can be effectively monitored using the SeaExplorer glider—an advanced autonomous underwater vehicle (AUV) designed for long-duration oceanographic missions. The study focuses on the glider’s ability to simultaneously observe a variety of environmental parameters, including temperature, conductivity, oxygen, and chlorophyll, during its deployment across multiple transects. The primary objective of the mission is to improve understanding of the vertical thermal structure and seasonal dynamics of the water column in this ecologically significant region. To achieve this, we apply Gaussian Process (GP) regression techniques to the glider-derived temperature data. This statistical method enables the smoothing and interpolation of irregularly spaced in situ measurements, thereby improving the visibility and interpretation of stratification patterns throughout the water column. Although the glider followed a predetermined course, the data-driven analysis suggests that adaptive sampling strategies—such as adjustments based on real-time outliers—could be valuable in future missions. Our results, which show distinct thermal layering and seasonal variability, are crucial for informing ecosystem function assessments and climate resilience planning. This study also discusses how integrating machine learning into glider-based monitoring could enhance MPA observation systems and promote adaptive, evidence-based management.

Deep-sea biological detection is a pivotal technology for the exploration and conservation of marine resources. Nonetheless, the inherent complexities of the deep-sea environment, the scarcity of available deep-sea organism samples, and the significant refraction and scattering effects of underwater light collectively impose formidable challenges on the current detection algorithms. To address these issues, we propose an advanced deep-sea biometric identification framework based on an enhanced YOLOv8n architecture, termed PSVG-YOLOv8n. Specifically, our model integrates a highly efficient Partial Spatial Attention module immediately preceding the SPPF layer in the backbone, thereby facilitating the refined, localized feature extraction of deep-sea organisms. In the neck network, a Slim-Neck module (GSconv + VoVGSCSP) is incorporated to reduce the parameter count and model size while simultaneously augmenting the detection performance. Moreover, the introduction of a squeeze–excitation residual module (C2f_SENetV2), which leverages a multi-branch fully connected layer, further bolsters the network’s global representational capacity. Finally, an improved detection head synergistically fuses all the modules, yielding substantial enhancements in the overall accuracy. Experiments conducted on a dataset of deep-sea images acquired by the Jiaolong manned submersible indicate that the proposed PSVG-YOLOv8n model achieved a precision of 79.9%, an mAP50 of 67.2%, and an mAP50-95 of 50.9%. These performance metrics represent improvements of 1.2%, 2.3%, and 1.1%, respectively, over the baseline YOLOv8n model. The observed enhancements underscore the effectiveness of the proposed modifications in addressing the challenges associated with deep-sea organism detection, thereby providing a robust framework for accurate deep-sea biological identification.