Image stitching can extend the limited field of view of ROV imagery and provide continuous visual records for aquaculture netting inspection. However, the repetitive, weak-texture characteristics of netting and deformations induced by water currents make it difficult for conventional stitching methods to generate complete panoramas. This paper proposes a factor graph-based sequential image stitching framework for visual mapping of aquaculture netting. Specifically, a motion prior-based image registration algorithm incorporating a planar netting model is introduced to improve the reliability of image correspondences. Furthermore, a multi-model switching framework uses the spatial distribution and tracking quality of optical-flow correspondences to trigger covariance inflation of the optical-flow constraint and introduce ORB and odometry constraints for homography estimation. The proposed framework is evaluated using simulated, controlled land-based, and sea-trial datasets. The results show robust and efficient sequential stitching in netting scenes. The resulting panoramic visual records provide a basis for ROV-assisted inspection of offshore aquaculture cages.
Coastal regions increasingly require adaptable protection systems due to sea-level rise and the increasing occurrence of hydrodynamic extremes. Floating breakwaters (FB) offer certain advantages over conventional structures, including reduced dependence on water depth and potential integration with wave energy converters. This study investigates the wave attenuation capacity and hydrodynamic response of a box-type FB with vertical and inclined taut moorings in an experimental test campaign in the Large Wave Current Flume GWK+. With a focus on long waves (FB breadth to wavelength ratio (B/L) from 0.05 to 0.16), wave reflection, transmission, energy dissipation, motion response (surge, heave, pitch), and mooring loads were analyzed. The findings reveal significant differences between the mooring configurations. The vertical mooring exhibits minimal wave attenuation capacity (CT=0.84±0.02 at B/L=0.15) due to a predominantly horizontal movement of the FB. Conversely, the inclined mooring enhances energy dissipation and wave attenuation capacity (CT=0.55 at B/L=0.06) due to a complex motion response leading to considerably higher mooring loads, which may limit integration with offshore wave energy converters (OWC). These results underscore the critical role of hydrodynamic response and wave attenuation in FB design, highlighting the need for further investigation into varied mooring angles. Additionally, experimental studies at a larger scale are advantageous regarding the co-use with integrated oscillating water columns, thereby optimizing FB-OWC integration for enhanced coastal resilience.
Highly nonlinear soliton-dominated sea states in deep water have been explored in previous studies through nonlinear theory and physical experiments. Deep water solitons are known to retain their characteristic properties during evolution within one-dimensional wave fields that are described by the nonlinear Schrödinger (NLS) equation, which is the most important model for their study. However, realistic ocean wave fields are typically multidirectional, a condition that can destabilize solitons and challenge the direct application of NLS-based interpretations. To address this gap, we generate directional JONSWAP wave fields with different directional spreadings that can be directly compared to field measurements to discuss the directional effects on soliton-dominated seas. The soliton energy ratio derived from the nonlinear Fourier spectrum is used as an indicator to assess dominant components of directional sea states. We investigate the relation between the soliton energy ratio obtained from the original time series, where the full directionality is present, to that obtained from the time series with reduced directionality using a directional filtering approach. For small directional spreading without filtering, the soliton energy ratio in the main direction is only slightly lower compared to initial wave fields, as expected. With increasing directional spreading, the soliton energy ratio decreases more significantly in most cases. However, in highly nonlinear soliton-dominated directional sea states, soliton energy is found to be concentrated in the main propagation direction, as the soliton ratio remains large after the directional interference has been removed.
Route-pattern information is important for shore-side analysis of North Pacific container shipping, as trans-Pacific route patterns differ in sailing distance, metocean exposure, and the spatial distribution of traffic. At departure, however, the route pattern ultimately followed is not yet known. In this study, a two-stage framework was developed for departure-time prediction of route-pattern probabilities using voyage attributes and metocean forecasts available at departure. First, three geographically interpretable route patterns, denoted High, Medium, and Low, were identified from post-voyage Automatic Identification System (AIS) trajectories in the training set using enhanced DBTCAN with adaptive consolidation. The coherence and reproducibility of the resulting route-pattern labels were assessed through trajectory-distance sensitivity and subsampling analyses. Second, a late-fusion 3D convolutional neural network (3D CNN) integrated 20 static voyage attributes with a 10-day ensemble-mean metocean forecast tensor comprising significant wave height and 10-m wind components to estimate route-pattern probabilities. On an independent one-year test set from the 2021–2025 voyage dataset, the 3D CNN achieved 85.16% accuracy, 81.48% Macro-F1, and an overall expected calibration error of 1.76%. Compared with XGBoost, the largest improvements occurred for the Low label, where recall and F1 score increased by 22.33 and 13.43 percentage points, respectively. Ablation analysis revealed that metocean inputs complemented static voyage attributes and increased Low-label F1 by 18.12 percentage points compared with the static-only 3D CNN. Permutation importance identified significant wave height as the dominant metocean input group for overall discrimination, while Low-label recall depended strongly on both wave and wind inputs.
Predicting surface-drifter trajectories is vital for material transport studies, marine hazard response, and analyses of upper-ocean circulation. However, this task remains challenging because winds, tides, mesoscale currents, and waves interact nonlinearly in ways that are difficult to represent in conventional drift-forecasting frameworks.This study developed a multi-source machine learning framework for drifter trajectory prediction around the Korean Peninsula. The framework uses oceanic model outputs from HYCOM, TPXO, and SCHISM, together with atmospheric and wave forcings from ERA5, as environmental input variables. An Extreme Gradient Boosting (XGBoost)-based surrogate model was trained to estimate the zonal and meridional components of drifter velocity, and the resulting velocity estimates were time-integrated to reconstruct 24-h trajectories. Model performance was evaluated against independent drifter observations and compared with a SCHISM-based OpenDrift simulation, while SHapley Additive exPlanations (SHAP) analysis was used to examine regional variations in input-variable attributions.Applied to the marginal seas around the Korean Peninsula, the proposed approach improved 24-h trajectory prediction, reducing distance RMSE from 9.73 to 7.49 km (23%) and increasing the normalized cumulative Lagrangian separation skill score (NCLS-SS) from 0.47 to 0.57 (21%) over the independent test set. Performance gains were observed across the tide-dominated western, mixed southern, and eddy-influenced eastern subregions. The SHAP results further revealed systematic regional contrasts in the attribution patterns of the environmental inputs across these dynamical regimes. These results indicate that the proposed framework offers a practical and interpretable approach for short-term surface-drift forecasting in heterogeneous coastal and marginal-sea environments.
