Short-Term Prediction of Global Sea Surface Temperature Using Deep Learning Networks

The trend of global Sea Surface Temperature (SST) has attracted widespread attention in several ocean-related fields such as warming, marine environmental protection and biodiversity. surface temperature is influenced by climate change; with the accumulation data from ocean remote sensing observations year year, many scholars have started to use deep learning methods for SST prediction. In this paper, we a dynamic region partitioning approach process big design framework applied short-term prediction system. On architecture Long Short-Term Memory (LSTM) network, two multi-region models are proposed, which extract temporal spatial information encoding, using feature transformation decoding predict future multi-step states. tested OISST model performance evaluated different metrics. proposed MR-EDLSTM MR-EDConvLSTM obtained best results prediction, RMSE ranging 0.2712 °C 0.6487 accuracies 97.60% 98.81% ten consecutive days show that better coastal areas, while performs predicting sea area near equator. addition, smaller compared forecasting system based on model, indicating method certain advantages SST.