| description abstract | hen a drifter is trapped in an eddy, it makes either a cycloidal or a looping trajectory. The former case takes place when the translating speed is larger than the eddy spinning speed. When the background mean velocity is removed, drifter trajectories make loops. Thus, eddies can be detected from a drifter trajectory by identifying looping segments. In this paper, an automated scheme is developed to identify looping segments from Lagrangian trajectories, based on a geometric definition of a loop, that is, a closing curve with its starting point overlapped by its ending point. The scheme is to find the first returning point, if it exists, along a trajectory of a surface drifter with a few other criteria. To further increase the chance that detected loops are eddies, it is considered that a loop identifies an eddy only when the loop?s spinning period is longer than the local inertial period and shorter than the seasonal scale, and that at least two consecutive loops with the same polarity that stay sufficiently close are found. Five parameters that characterize an eddy are estimated by the scheme: location (eddy center), time (starting and ending time), period, polarity, and intensity. As an example, the scheme is applied to surface drifters in the Kuroshio Extension region. Results indicate that numbers of eddies are symmetrically distributed for cyclonic and anticyclonic eddies, mean eddy sizes are 40?50 km, and eddy abundance is the highest along the Kuroshio path with more cyclonic eddies along its southern flank. | |
