Mesoscale Eddy Fitting from Satellite Along-Track Data in the Antarctic Circumpolar Current

Abstract Accurate fitting of mesoscale eddies is crucial for understanding the transport of oceanic kinetic energy and its impact on climate systems. By utilizing satellite along-track sea level anomaly data, mesoscale eddies in localized regions can be fitted precisely using methods such as B-spline surface fitting or Legendre orthogonal polynomial surface fitting. However, as the spatial extent of the region increases, both methods necessitate progressively higher fitting orders, which significantly escalate computational demands. In this study, we address this challenge by dividing a large area into several 15° × 15° subregions with overlapping boundaries. Each subregion is fitted independently, and a distance-weighted averaging technique is applied in the overlapping areas to ensure a smooth transition between adjacent regions. This approach enables efficient fitting of large areas while substantially reducing computational load. The method’s robustness is thoroughly validated across different seasons and ocean basins, confirming its stability under varying oceanic conditions. Results show that expanding the overlapping area does not notably affect the mean absolute error (MAE) or root-mean-square error (RMSE) at the overlap, with error increases remaining under 1 cm. Changing the spatial extent of overlapping areas in one direction alone does not significantly affect the fitting results in the orthogonal direction. Among the three distance weighting schemes (linear, cosine, Gaussian), linear weighting is preferred due to its computational efficiency and competitive accuracy, particularly when the overlapping area is less than 2.5°. By decomposing the fitting process into smaller, manageable tasks, this method maintains fitting accuracy and provides a practical solution for the efficient dynamic monitoring of global mesoscale eddies.