Intelligent detection of internal waves based on SWOT satellite observations

Internal waves are key phenomena that impact ocean dynamics, material transport, and energy dissipation. This study presents an automated method for detecting internal waves using SWOT (surface water and ocean topography) altimetry data. A novel deep learning-based framework, SWOT_IWD, leverages high-resolution sea surface height (SSH) measurements for internal wave detection. A comprehensive dataset of internal waves from SWOT has been constructed, and the results are cross-validated using multi-source remote sensing data to ensure accuracy. The method was applied to internal wave detection in the Sulu Sea, achieving a true positive rate (TPR) of 70.00%, accuracy (ACC) of 96.08%, false discovery rate (FDR) of 3.92%, and false negative rate (FNR) of 30.00%. The results highlight the significant spatiotemporal variability of internal waves in the Sulu Sea. Statistical analysis shows that while there is general consistency between the number of SWOT data containing internal waves and those detected, the relationship is not strictly linear. This study demonstrates the feasibility of using SWOT data combined with deep learning for internal wave monitoring, showcasing its potential for advancing oceanography and marine remote sensing applications.