Research Of Sea Surface Height Prediction Based On LSTM And CNN

Sea Surface Height is an elemental factor in ocean environment and marine engineering.The research and accurate prediction of the sea surface height play a major role in the research,production and life of marine.Due to the complexity and uncertainty of the marine environment,the accuracy of surface height prediction needs to be improved.In order to improve the prediction accuracy of sea surface height,This paper proposes two prediction models in terms of time prediction and space prediction.Physical oceanography models have been developed to forecast sea surface height,but the accuracy decreases heavily when it needs to predict a little long time ahead.In time-scale forecasting,a novel data driven deep learning method is proposed to predict SSHA.Specifically,SSHA prediction is treated as a time series forecasting problem,and our model can mine the discipline hidden in short time series,and tackle long-term dependence of series changes.Data experiments conducted on SSHA dataset of China Ocean Reanalysis in the South China Sea show that our method achieves average predicting accuracy plus/minus standard deviation of coming 24 h,48 h,72 h,96 h,and 120 h by 90.99±10.56%,85.49±13.93%,79.99±16.08%,74.23±18.05%,68.15±18.84%,respectively.The proposed method performs better than several state-of-the-art machine learning methods,including artificial neural network,merged-recurrent neural network,time convolutional network,merged-gate recurrent unit,and one-dimensional convolutional neural network in predicting SSHA.In the research of sea surface height spatial prediction,this paper proposes a spatial prediction model based on self-attention mechanism and convolutional neural network.The data of each grid point is correlated with the data of other grid points.The spatial prediction of the sea surface height is to simulate the predicted point data based on the data around the predicted point to improve the spatial resolution of the original data.Traditional interpolation only takes into account the information of spatial dimensions.Considering that the sea surface height data has temporal and spatial correlation,the model proposed in this paper simulates the mapping of low-resolution data to high-resolution data,and uses historical data for training.The model was verified on the Reanalysis Dataset of the South China Sea.The SSIM,PSNR,and MSE of our model reached 0.9999,155.6459,and 0.000089,respectively.The experiment is also compared with other interpolation and machine learning models,including cubic interpolation,nearest neighbor interpolation,bilinear interpolation,SRCNN,FSRCNN,DRCN.The experimental results show that our proposed model achieves the highest spatial prediction accuracy.