Weather routing is a critical guarantee for the safe and economical navigation of ships. Existing methods for weather routing still face challenges in selecting the appropriate planning granularity. A granularity that is overly coarse may result in routes passing through coastal and island-rich waters, such as coastal zones and reefs, thus compromising navigational safety. Conversely, a granularity that is excessively fine leads to an exponential increase in computational complexity, rendering the problem intractable. To address this issue, this paper proposes an efficient method for weather routing in coastal and island-rich waters, guided by ship trajectory big data: First, an adaptive quadtree is used to partition the navigable space into an adaptive grid, based on which a route network is constructed using ship trajectory big data. Next, a ship motion model is introduced to build both static and dynamic marine environmental fields, which are used to dynamically update the time weights of the route network. Finally, using the updated route network as a guide, the method aims to minimize voyage time and employs an improved time-varying A* algorithm for weather routing. Experimental results show that the proposed method effectively adapts to coastal and island-rich waters, outperforming the baseline SIMROUTE in safety, optimization, and efficiency. Unlike SIMROUTE, which crosses restricted areas, it avoids such risks entirely. It achieves average reductions of 6.8% in route length and 4.3% in navigation time and is 5.8 times faster than SIMROUTE for fine-grained planning. This balances voyage time, safety, and efficiency, offering a practical weather routing solution.

Marine researchers rely heavily on ocean sound velocity, a crucial hydroacoustic environmental metric that exhibits large geographical and temporal changes. Nowadays, spatio-temporal series prediction algorithms are emerging, but their prediction accuracy requires improvement. Moreover, in terms of ocean sound speed, most of these models predict an ocean sound speed profile (SSP) at a single coordinate position, and only a few predict multi-spatial-scale SSPs. Hence, this paper proposes a new data-driven method called STA-Conv-LSTM that combines convolutional long short-term memory (Conv-LSTM) and spatio-temporal attention (STA) to predict SSPs. We used a 234-month dataset of monthly mean sound speeds in the eastern Pacific Ocean from January 2004 to June 2023 to train the prediction model. We found that using 24 months of SSPs as the inputs to predict the SSPs of the following month yielded the highest accuracy. The results demonstrate that STA-Conv-LSTM can achieve predictions with an accuracy of more than 95% for both single-point and three-dimensional scenarios. We compared it against recurrent neural network, LSTM, and Conv-LSTM models with optimal parameter settings to demonstrate the model’s superiority. With a fitting accuracy of 95.12% and the lowest root-mean-squared error of 0.8978, STA-Conv-LSTM clearly outperformed the competition with respect to prediction accuracy and stability. This model not only predicts SSPs well but also will improve the spatial and temporal forecasts of other marine environmental factors.

Addressing the shortcomings of existing ship trajectory compression methods that rely on the empirical setting of a fixed threshold and face challenges in controlling the spatial similarity before and after compression, this paper proposes a self-adaptive compression method for ship trajectories that does not require threshold setting. Initially, a hierarchical and sequential tree structure based on the trajectory characteristics is constructed to determine the importance of a ship’s trajectory points. Subsequently, the dynamic time warping (DTW) operator is introduced to assess the spatial similarity of the trajectory before and after compression, exploring the laws governing similarity variations in the step changes during the compression process from lower to higher levels of the hierarchical and sequential sequence. Finally, the trajectory point that causes the largest step change in similarity within the hierarchical and sequential sequence is identified, and points at lower levels than this point are discarded, thus achieving the objective of self-adaptive adjustment of the level of compression. Our case study results demonstrate that, compared with existing ship trajectory compression methods based on empirically set thresholds, the proposed method has the following advantages: (1) it does not require presetting a fixed threshold and adaptively determines the degree of compression by identifying the trajectory point that leads to the largest step change in similarity, and (2) under the condition of a similar data compression rate, the DTW calculation value is reduced by approximately 25%, significantly enhancing the similarity of the trajectory before and after compression.

Shallow water bathymetry data underpins maritime shipping and marine resource survey/protection, but its accuracy is constrained by water heterogeneity and spectral interference. To address this, this study proposes a Spatio-Temporal Coupling and Correlation Adaptive Spectral (STCCAS) inversion method, integrating four machine learning models: Random Forest (RF), XGBoost, Support Vector Regression (SVR), and Multi-Layer Perceptron (MLP). Experiments were conducted in Tampa Bay’s nearshore waters, using Sentinel-2 imagery and Airborne LiDAR Bathymetry (ALB) data. Core to STCCAS, the Temporal Stability Index (TSI) quantifies spectral temporal consistency, while the Normalized Difference Turbidity Index (NDTI) characterizes water turbidity, and the two indices synergistically form a dual-scale “spectral reliability-turbidity stability” evaluation system for pixel-level feature quality assessment—coupled with spectral fusion features and spatial location, they jointly realize pixel-level feature reliability weighting and dynamic filtering to build a water condition-adaptive input set. Comparative analysis of inversion performance under the original spectral features (OSFs) inversion method vs. STCCAS inversion method confirms STCCAS significantly boosts accuracy. XGBoost outperforms others, achieving a coefficient of determination (R²) of 0.93, root mean square error (RMSE) of 0.16 m, and mean absolute error (MAE) of 0.12 m. STCCAS breaks the limitations of traditional fixed feature combinations, effectively adapting to nearshore water heterogeneity. It provides a novel method for high-frequency, high-precision shallow water bathymetry inversion, with important practical value for marine environmental monitoring and resource management.

Coastal geomorphology is a key component of nearshore representation on nautical charts. However, its cartographic generalization remains constrained by incomplete knowledge-system modeling, limited task-oriented guidance, and insufficient quantitative descriptions of generalization quality. To address these limitations, this study proposes a knowledge-graph-based approach for cartographic generalization of coastal geomorphology in nautical charts. A three-layer graph framework, consisting of a concept layer, an entity layer, and a rule layer, was constructed around shorelines, coastline categories, and generalization rules. Textual knowledge extraction and spatial-feature extraction were then conducted using cartographic specifications, historical chart samples, and expert-experience documents to support the unified organization and population of multi-source knowledge. A reasoning mechanism integrating spatial constraints and semantic associations was further developed to generate quantitative reference indicators and evaluate shoreline generalization quality. The results show that the proposed knowledge graph can organize domain knowledge in a structured and traceable form, identify rule-compliance differences among different shoreline generalization results, and provide descriptive reference benchmarks for quality evaluation. The method offers a knowledge-driven basis for improving the objectivity and explainability of coastal geomorphology generalization in nautical-chart production.

To address the challenges posed by significant variations in bend morphology and spatial relationships within complex shoreline segments—where single generalization operators and uniform parameters struggle to balance node compression with morphological preservation—this paper proposes an adaptive shoreline generalization method based on bend sequence attention learning. Treating bend units as local decision objects, the method integrates geometric morphology, proximity distance, and shoreline type features. It employs bidirectional long short-term memory (BiLSTM) and self-attention mechanisms to learn contextual relationships among bends within each shoreline and predict local generalization operators and their parameters. Proximity links between shorelines support spatial consistency checks during constraint-based correction. After operator execution, candidate geometries that violate cartographic constraints undergo parameter rollback or operator reselection; unresolved bends are restored to their original geometry before output. Under the evaluated datasets, scoring procedure, and compression settings, the final generalized shorelines achieved a vertex-reduction ratio of 86.43% and a bend-retention ratio of 94.84%. These results support the feasibility of bend-level adaptive operator selection at the evaluated compression–preservation operating point. Performance across other compression settings and chart series remains to be established.