Mesoscale eddy identification in the North Pacific based on DeepLabV3+

Abstract In recent years, the application of artificial intelligence (AI) in Marine feature detection has become the focus of research. Ocean eddies are ubiquitous in the global ocean and play a crucial role in the transport of key tracers. The detection scheme of eddy current based on ocean dynamics has been extensively explored. Recently, eddy detection based on artificial intelligence has also received much attention In this study, the DeepLabV3+ algorithm was employed to identify mesoscale eddies. Sea surface height (SSH) data from the Copernicus L4 satellite, covering the region between 140°E-180°E and 10°N-40°N, were selected for analysis The distribution of mesoscale eddies during January-June, 2024 was obtained using DeepLabV3+. The shape, propagation, generation and dissipation information of mesoscale eddies identified here are potentially useful for future studies about eddy prediction and identification.