Accurate observation of nearshore waves is crucial for coastal safety. In this study, the feasibility of extracting wave information from wave video images captured by shore-based cameras using deep learning methods was explored, focusing on inverting nearshore significant wave height (SWH) from instantaneous wave video images. The accuracy of deep learning models in classifying wind wave and swell wave images was investigated, providing reliable classification results for SWH inversion research. A classification network named ResNet-SW for wave types with improved ResNet was proposed. On this basis, the impact of instantaneous wave images, meteorological factors, and oceanographic factors on SWH inversion was evaluated, and an inversion network named Inversion-Net for SWH that integrates multiple factors was proposed. The inversion performance was significantly enhanced by the specialized models for wind wave and swell. Additionally, the inversion accuracy and stability were further enhanced by improving the loss function of Inversion-Net. Ultimately, time series inversion results were synthesized from the outputs of multiple models; the final inversion results yielded a mean absolute error of 0.04 m and a mean absolute percentage error of 8.52%. Despite certain limitations, this method can still serve as a useful alternative for wave observation.

The safety of winter activities in the Bohai Sea requires more detailed information on ice characteristics and a more refined ice zone division. In the present study, 1/12°-resolution sea ice characteristic data were obtained based on the NEMO-LIM2 ice–ocean coupling model that assimilated MODIS satellite sea ice observations from the years of 2005 to 2022 to acquire new sea ice hindcasting data. On this basis, the ice period, ice thickness, ice concentration, ice temperature, ice salinity, and design ice thickness for different return periods in the 1/4°-resolution refined zoning were analyzed, which were then compared with the sea ice characteristics in the previous 21-ice-zone standard. The distribution of ice temperature and ice salinity was closely related to the distribution of ice thickness. The results of ice period, ice thickness, and ice concentration, as well as design ice thickness for different return periods, and the comparison with the previous 21-ice-zone standards, showed that the ice condition on the west coast of the Bohai Sea has significantly reduced.

Effective risk assessment can reduce the economic losses and physical damage caused by marine dynamic processes, such as storm surges. Most risk assessments of marine disasters are based on regional parameters and discrete hazard grades. Targeted, multilevel, and multiangle risk assessments are urgently needed. This study focuses on specific types of affected infrastructure. We established a sensitivity matrix by considering the effects of different disaster causal factors on different types of affected infrastructure. Through this matrix, hazards, vulnerability, and emergency response and recovery capability were effectively combined in a risk assessment framework. We completed the risk calculation for multiple concurrent effects of disasters in areas with superimposed key infrastructure using complementary risk superposition. The hazard grade, vulnerability grade, and coefficient of emergency response and recovery capability were established based on the means of return period, characteristics of disaster distribution, types of affected infrastructure and disaster relief distance, and were continuous by solving functions, normal cumulative distributions, and analytic functions. On the basis of reasonable MIKE21 numerical simulation and abstract spatial distribution of vulnerable assets, we tested the rationality of the assessment system in the Lekima typhoon storm surge process. The results showed that the assessment system accurately reflected the risk of damage to the important infrastructure in terms of spatial distribution. Therefore, this risk assessment framework was suitable for the assessment of a marine dynamic disaster process in the lower Laizhou Bay coastal area of the Bohai Sea, China. Moreover, it provided a reference for disaster prevention and reduction, guided the way for decision making, and effectively reduced disaster losses.

Deep-learning post-processing can correct systematic errors in numerical wave forecasts, but its reliability under high-wave conditions remains unclear. We develop an error-correction framework in which deep networks predict the full future error trajectory of operational forecasts, and compare three architectures, a pure LSTM, a pure Transformer and a hybrid LSTM–Transformer, tuned identically with Bayesian optimization, using hourly observations from six buoys in the Bohai Sea and Yellow Sea over 2023–2025. On the full test sample all models improve the 72 h forecasts by 15.37% to 23.93% in RMSE, with the hybrid model attaining the lowest pooled RMSE; on the high-wave subset with SWH above 1.5 m, however, all three degrade the forecast, increasing RMSE by 7.58% to 9.05%. An exact decomposition of the squared-error change shows that fluctuation-capturing skill remains intact while the degradation is driven by the mean term: trained on samples dominated by low and moderate waves, the correctors learn a globally averaged adjustment whose sign opposes the reversed raw bias at high waves. Neither loss reweighting nor filtered training resolves this degradation at the 72 h horizon. Shortening the horizon to 24 h restores positive skill, with RMSE reductions of 4.43% to 7.88%, and reverses the ranking in favor of the pure LSTM. These findings indicate that architecture rankings are horizon-dependent, and they are consistent with the interpretation that extreme-condition skill is limited by the information content of the predictors rather than by model capacity.

The development of wind wave (i.e., sea state) inside an intense tropical cyclone (TC) is the dominant contributor to the sea surface roughness and thus significantly impacts the air–sea interaction. The sea state is known to vary with TC characteristics (intensity, size, and translation speed); however, comprehensive knowledge of the influence of TC characteristics on the sea state and sea surface roughness is quite limited, largely because of the lack of observations. In this study, numerical experiments are performed to investigate the influence of TC characteristics on the sea state and the sea surface roughness under a range of idealized TCs. The numerical results indicate that the sea states are systematically younger for a more intense or smaller TC, and their azimuthal variation is predominantly determined by the TC translation. The dependence of the sea surface roughness on wind speed shows systematic variations with the TC characteristics, which are most significant for a moderately moving TC.

The wave spectrum, as a key statistical feature describing wave energy distribution, is crucial for understanding wave propagation mechanisms and supporting ocean engineering applications. This study, based on ERA5 reanalysis spectrum data, proposes a model combining CNN and xLSTM for rapid gridded wave spectrum prediction over the Bohai and Yellow Seas domain. It uses 2D gridded spectrum data rather than a spectrum at specific points as input and analyzes the impact of various input factors at different time lags on wave development. The results show that incorporating water depth and mean sea level pressure significantly reduces errors. The model performs well across seasons with the seasonal spatial average root mean square error (SARMSE) of spectral energy remaining below 0.040 m²·s and RMSEs for significant wave height (SWH) and mean wave period (MWP) of 0.138 m and 1.331 s, respectively. At individual points, the spectral density bias is near zero, correlation coefficients range from 0.95 to 0.98, and the peak frequency RMSE is between 0.03 and 0.04 Hz. During a typical cold wave event, the model accurately reproduces the energy evolution and peak frequency shift. Buoy observations confirm that the model effectively tracks significant wave height trends under varying conditions. Moreover, applying a frequency-weighted loss function enhances the model’s ability to capture high-frequency spectral components, further improving prediction accuracy. Overall, the proposed method shows strong performance in spectrum prediction and provides a valuable approach for regional wave spectrum modeling.