Regional identification in the tropical Atlantic Ocean of residual tide errors from an empirical orthogonal function analysis of TOPEX/POSEIDON altimetric data

Two years (October 1992 to September 1994) of TOPEX/POSEIDON data have been processed over the tropical Atlantic Ocean to produce monthly sea level anomalies. Four different oceanic tide corrections have been applied in order to quantify the effect of this correction on the large-scale altimetrically determined sea level variability and to identify potential residual tide errors. Two tide corrections are drawn from the U.S. University of Texas tide model, and two others are drawn from the French Grenoble tide model. Comparison of the altimetric sea level anomalies with tide gauge data gives RMS differences of between 2 and 5 cm, equivalent to those already observed by several authors during recent TOPEX/POSEIDON studies. However, this paper demonstrates that the 2-monthly oscillations, which were previously identified in investigations using old versions of tide corrections, still remain, to some extent independently of the new tide models being used. These a-monthly oscillations are not only located along the western boundary and therefore cannot be attributed to 40- to 60 day oscillations observed in several mooring records in that area. An empirical orthogonal function analysis of the altimetric data clearly shows that these oscillations are due to tide aliasing. This approach also emphasizes the improvement finalized by the more recent versions of the University of Texas and the Grenoble tide models. However, the most convincing result of this analysis is that it allows regional quantification of the impact of tidal residual error aliasing which persists even when the more sophisticated tide models are used.