This research introduces a deep learning method for ocean wave height estimation utilizing a Convolutional Neural Network (CNN) based on the VGGNet. The model is trained on a dataset comprising buoy wave heights and radar images, both critical for marine engineering. The dataset features X-band radar images sourced from Sokcho, Republic of Korea, spanning from June 1, 2021, to August 13, 2021. This collection amounts to 72,180 three-dimensional images, gathered at intervals of approximately 1.43 seconds. The data collected was highly unbalanced in terms of wave heights, with images of lower wave heights being more common. To deal with data imbalances in the wave height datasets, we categorized the data into three groups based on wave heights and applied stratified random sampling at each level. This approach balances the data patches for each training iteration, reducing the risk of overfitting and promoting learning from diverse data. We also implemented a system to protect data in groups with fewer instances, ensuring fair representation across all categories. This study presents a deep learning regression model for predicting wave height values from radar images. The model extracts features from sequences of 64 radar images using three-dimensional convolutions for both temporal and spatial learning. Using three-dimensional convolutions, the model captures temporal features in radar image sequences and provides accurate wave height estimates with an RMSE of 0.3576 m. The study derived results using radar images under different wave height conditions for 74 days to ensure reliability.

Modeling a nonlinear ocean wave is one of the primary concerns in ocean engineering and naval architecture to perform an accurate numerical study of wave-structure interactions. The high-order spectral (HOS) method, which can simulate nonlinear waves accurately and efficiently, was investigated to see its capability for nonlinear wave generation. An open-source (distributed under the terms of GPLv3) project named "HOS-ocean" was used in the present study. A parametric study on the "HOS-ocean" was performed with three-hour simulations of long-crested ocean waves. The considered sea conditions ranged from sea state 3 to sea state 7. One hundred simulations with fixed computational parameters but different random seeds were conducted to obtain representative results. The influences of HOS computational parameters were investigated using spectral analysis and the distribution of wave crests. The probability distributions of the wave crest were compared with the Rayleigh (first-order), Forristall (second-order), and Huang (empirical formula) distributions. The results verified that the HOS method could simulate the nonlinearity of ocean waves. A set of HOS computational parameters was suggested for the long-crested irregular wave simulation in sea states 3 to 7.

Ocean surface waves have been utilized as fundamental information in various fields of oceanic research. In this paper, we suggest a simulation and modelling technique for generating an ocean surface wave using an inverse Fast Fourier Transform (iFFT), and we subsequently verify the wave’s accuracy. The conventional method, linear superposition, requires recursive calculation because of the double summation and the time variable; to circumvent this issue, the new algorithm is presented. The Joint North Sea Wave Project (JONSWAP) spectrum is utilized for the ocean surface wave simulation example, and the parameters are the significant wave height (HS) and the zero-crossing wave period (TZ). A coordinate transform for the wavenumber domain was used to apply the inverse FFT algorithm. To verify the accuracy of the simulation result, the relative error between the input condition and the analysis result was calculated. The result for TZ is below 4% relative error, and the maximum relative error for HS is 7%. To avoid the Nyquist frequency for wave-field analysis and simulation, the minimum grid size was calculated by twice applying the maximum wavenumber.