On the Exploitation of Multimodal Remote Sensing Data Combination for Mesoscale/Submesoscale Eddy Detection in the Marginal Ice Zone

Abstract

The detection and analysis of ocean eddies via remote sensing have become a hot topic in physical oceanography during the last few decades. However, eddy identification and tracking via remote sensing can be a challenging task, since each sensor has some limitations. In order to overcome potential challenges, it is crucial to exploit the complementary information provided by different sensing systems. As one of the steps toward this aim, we have investigated the pertinence of applying the scheme, including a texture features extraction and a superpixel segmentation method, in order to distinguish eddies in the marginal ice zone (MIZ) using multisensor remote sensing data. Nevertheless, not all the images available from various sensors are of actual importance, since they can be corrupted, redundant, or simply unnecessary for a particular task. Therefore, we are additionally exploring the relevance of different sensors separately and simultaneously as well as with extracted texture features for eddy monitoring.