Research on Ocean Temperature Field Reconstruction Based on Convolutional Neural Network

This study proposes a method for reconstructing the ocean temperature field based on convolutional neural network (CNN). By combining multi-source information such as satellite remote sensing data and buoy data, deep learning technology is used to reconstruct the global ocean temperature field. This method constructs a deep neural network architecture by preprocessing the input data, and optimizes the model parameters during the training process to improve the accuracy of temperature field prediction. Experimental results show that the ocean temperature field reconstruction method based on CNN can effectively improve the accuracy of temperature field reconstruction compared with traditional interpolation methods, especially when processing a large range of ocean temperature data, it can capture complex spatial features. By comparing with traditional methods, the CNN model showed higher accuracy and robustness in multiple experiments. This study provides new technical support for climate change prediction, ocean monitoring and resource management, and lays a foundation for the future application of deep learning in the field of marine science.