Wave prediction in the Yellow Sea and Bohai Sea region based on convolutional neural network
JIANG Feifei
WANG Zhifeng
LI Haoqian
LI Rui
Abstract：Accurate prediction of wave height is of great significance for engaging in various marine ac-tivities,and the existing prediction methods are mainly numerical prediction methods.In terms of deep learning,most of the forecasts are single-point forecasts for the measured sites,and there are fewer predictions for regional wave fields.Based on three years of ERA5 data,the wave height in the Yellow sea and Bohai sea region was forecast every 3 h for the next 3～96 h by using the developed convolution-al neural network(CNN)model.The point evaluation method and the field evaluation method were used to compare the forecast values and the target values respectively.The results showed that the de-veloped CNN model was extremely accurate in short-time prediction.Comparing the results of 12 h prediction,the correlation coefficient was basically 0.93,the mean absolute percentage error was about 20％,and it ran fast.It also had good applicability in the long-time prediction,the correlation coeffi-cient was mostly above 0.8,and the mean absolute percentage error was generally below 30％.It showed that the model can provide reference for marine rescue,marine engineering construction and other activities.
Keywords：convolutional neural network(CNN)modelwave height area forecastthe Yellow Sea and Bohai Sea regiondeep learning
Publication Date：2025-10-30
Online Publishing Date：2026-05-22(First online date of this platform, not the publication date of the document)
Pages：10( 46-55 )