DualSeq: A Novel Ensemble Method for Predictive Analysis of Sea Surface Temperature Using Remote Sensing Data

This research brings an innovative dual sequence ensemble approach on predictions of Sea Surface Temperature Prediction (SSTP) across the Arabian Sea region. The proposed DualSeq model combines Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) networks using weighted predictions. The proposed DualSeq model uses the five-year dataset of SEVIRI-IO-SST retrieved from the PODAAC, NASA. It utilizes remote sensing data to provide SST forecasts for up to 30 days. The DualSeq model outperforms existing models, including ANN, CNN, LSTM, GRU, LSTM-Adaboost, GRU encoder-decoder, LSTM-Attention, GRU-FL, Numerical-NN, RBFN, FC-LSTM, and CFCC-LSTM. Its advanced ensemble approach demonstrates superior performance in SST prediction. The coefficient of determination for the DualSeq model came at 0.9833, while RMSE came at 0.328 degrees C, MAPE at 0.11%, and MAE of 0.0216 degrees C. The DualSeq model represents a significant advancement in SST prediction, with potential applications in marine science, environmental monitoring, and climate studies. This combines both LSTM and GRU with the result being improved forecasted predictions along with higher resolution of SST variability, thus showing that this system is capable of bringing high-resolution predictive power to oceanography.