Abstract
Three methods of extraction of mesoscale variability from altimetry are compared in the Somali current area: crossover residuals after minimization to reduce orbit errors, along-track variability using quasi-repetitive tracks, and difference relative to the altimetric mean sea surface of Marsh et al. The signals are filtered using an optimal estimation technique with a zero-crossing of the spatial correlation function of 250 km and an e-folding decay time of 30 days. The comparison is made by spatial correlation between the variation maps produced by the three techniques and with XBT data along the tanker sea lane. Probably because of insufficient data defining the reference mean sea surface in the area, the third method leaves large topographic features related to the geoid. Similarly, the horizontal distance between the quasi-repetitive tracks is too large and their difference is contaminated by the geoid signal. The comparison between the topographic heights deduced from crossover residuals and those estimated from the XBT sections is excellent down to 7 cm rms. These maps reveal an eastward extension of the “Great Whirl” in August. The variation of topography is so swift that it is impossible to sample the system correctly with a single oceanographic vessel.
