Eddies play a crucial role in the global transport of heat, salt, and nutrients. They significantly influence ocean circulation and regional climate patterns, making their detection and analysis one of the key areas of oceanographic and climate research. In particular, the marginal ice zone, where ocean water meets sea ice, is a region of intense eddy activity, and understanding eddy variability in this region is essential for predicting changes in the Earth’s climate system. Spaceborne SAR data have become a primary source of information for monitoring eddies in the MIZ, where other sources fail to provide relevant information. The high-resolution capabilities of SAR allow for the detailed observation of ocean and sea ice surface features, including eddies on both the mesoscale and submesoscale levels. However, the sheer volume and complexity of SAR data necessitate the development of automated methods for accurate and efficient eddy detection.
In this study, we propose and verify an optimal approach for the automatic detection of eddies in the MIZ, specifically in the Fram Strait, a crucial gateway between the Arctic and Atlantic Oceans. Our  approach employs the YOLOv8 a state-of-the-art computer vision model for its speed and accuracy in object detection tasks. By applying this model to high-resolution SAR data, we aim to enhance the detection and classification of eddies in the region of interest. Based on the acquired time-series, we investigate the intensity of eddy generation, their properties and their seasonal variability in the MIZ regions of Fram Strait.
 

eddies,ocean mesoscale eddies, mesoscale eddies, marginal ice zone, Arctic