Accurate wave prediction can help avoid disasters. In this study, the significant wave height (SWH) prediction performances of the recurrent neural network (RNN), long short-term memory network (LSTM), and gated recurrent unit network (GRU) were compared. The 10 m u-component of wind (U10), 10 m v-component of wind (V10), and SWH of the previous 24 h were used as input parameters to predict the SWHs of the future 1, 3, 6, 12, and 24 h. The SWH prediction model was established at three different sites located in the Bohai Sea, the East China Sea, and the South China Sea, separately. The experimental results show that the performance of LSTM and GRU networks based on the gating mechanism was better than that of traditional RNNs, and the performances of the LSTM and GRU networks were comparable. The EMD method was found to be useful in the improvement of the LSTM network to forecast the significant wave heights of 12 and 24 h.

A coastal inundation simulation system was developed for the coast of the Pearl River estuary (PRE), which consists of an assimilation typhoon model and the coupled ADCIRC (Advanced Circulation) + SWAN (Simulating Waves Nearshore) model. The assimilation typhoon model consists of the Holland model and the analysis products of satellite images. This is the first time an assimilation typhoon model has been implemented and tested for coastal inundation via case studies. The simulation results of the system agree well with the real measurements. Three observed typhoon paths (Hope, Nida, and Hato) were chosen to be the studied paths based on their positions relative to the PRE, China. By comparing the results of experiments with different forcing fields, we determined that the storm surge and the coastal inundation were mainly induced by wind forcing. By simulating coastal inundation for different typhoon center speeds, the Hato3 path most easily causes coastal inundation in the PRE. Moreover, the moving speed of the typhoon’s center significantly affects the coastal inundation in the PRE. The inundation becomes very serious as the movement of the typhoon center was slow down. This study provides a new reference for future predictions of coastal inundations.

Characteristics of near-inertial waves (NIWs) induced by the tropical storm Noul in the South China Sea are analyzed based on in situ observations, remote sensing, and analysis data. Remote sensing sea level anomaly data suggests that the NIWs were influenced by a southwestward moving anticyclonic eddy. The NIWs had comparable spectral density with internal tides, with a horizontal velocity of 0.14–0.21 m/s. The near-inertial kinetic energy had a maximum value of 7.5 J/m³ and propagated downward with vertical group speed of 10 m/day. Downward propagation of near-inertial energy concentrated in smaller wavenumber bands overwhelmed upward propagation energy. The e-folding time of NIWs ranged from 4 to 11 days, and the larger e-folding time resulted from the mesoscale eddies with negative vorticity. Modified by background relative vorticity, the observed NIWs had both red-shifted and blue-shifted frequencies. The upward propagating NIWs had larger vertical phase speeds and wavelengths than downward propagating NIWs. There was energy transfer from the mesoscale field to NIWs with a maximum value of 8.5 × 10⁻⁹ m² s⁻³ when total shear and relative vorticity of geostrophic currents were commensurate. Our results suggest that mesoscale eddies are a significant factor influencing the generation and propagation of NIWs in the South China Sea.

Based on satellite and analysis data and in situ observations acquired during May 23, 2017 to May 19, 2018, the spatiotemporal variations of the along-slope counter-flow off northeastern Taiwan were investigated. It was observed that the along-slope counter-flow in the subsurface layer was uplifted and lowered significantly during the study period. The counter-flow was significantly uplifted (lowered) while the sea surface was during an interval of positive (negative) geostrophic velocity anomaly (GVA) curl. The vertical migration of the counter-flow was also found closely linked with the Kuroshio intrusion (KI) to the northeast of Taiwan. The depths of both the upper boundary and the axis of the counter-flow were found proportional to the KI variance along the western continental slope off northeastern Taiwan. More importantly, it was established that the variation of the KI to the northeast of Taiwan had better correlation with the counter-flow than the Kuroshio derived from altimetry data. Thus, further study of the variation and mechanism of the along-slope counter-flow is needed to improve the understanding and prediction of the KI in the area of northeastern Taiwan, as well as the biochemical systems and marine economy in the East China Sea in the future.

In the era of marine big data, making full use of multi-source satellite observations to accurately retrieve and predict the temperature structure of the ocean subsurface layer is very significant in advancing the understanding of oceanic processes and their dynamics. Considering the time dependence and spatial correlation of marine characteristics, this study employed the convolutional long short-term memory (ConvLSTM) method to retrieve the subsurface temperature in the Western Pacific Ocean from several types of satellite observations. Furthermore, considering the temperature’s vertical distribution, the retrieved results for the upper layer were iteratively used in the calculation for the deeper layer as input data to improve the algorithm. The results show that the retrieved results for the 100 to 500 m depth temperature using the 50 m layer in the calculation resulted in higher accuracy than those retrieved from the standard ConvLSTM method. The largest improvement was in the calculation for the 100 m layer, where the thermocline was located. The results indicate that our improved ConvLSTM method can increase the accuracy of subsurface temperature retrieval without additional input data.