Mathematical Morphology Approach for Finding Mesoscale Ocean Features

摘要

The ocean is poorly sampled by conventional techniques. Satellite sensors are alogical choice to fill the gaps. This paper describes a technique based on mathematical morphology to aid in interpreting satellite infrared images of the North Atlantic Ocean's Gulf Stream region. The algorithm used to find rings (eddies) and the Gulf Stream's North Wall performs successive openings and closings (defined in terms of erosion and dilation) at increasing thresholds until a stable division into objects and background is found. This method delineates the North Wall clearly. It finds it in approximately the same place as human analysts, with less smoothing of small irregularities, and finds it more precisely than another automated procedure. Performance in locating warm-core eddies is comparable to the other methods. The present procedure does not find cold-core rings well, but we expect to be able to improve that performance and to reduce the effects of clouds and other interference. Remote sensing, artificial intelligence, data assimilation, satellite data.