## 1. Introduction

[2] High-precision satellite altimetry has allowed getting very accurate solutions for ocean tides at the global scale [*Le Provost et al.*, 1995; *Le Provost*, 2001]. The main difficulty encountered in extracting the tidal characteristics from the altimetric signal is related to the repeat cycle period of the altimeter satellites. Their low-frequency sampling of the ocean topography results in aliasing of the high-frequency signals into much longer periods. The TOPEX/Poseidon (T/P) repeat cycle period of 9.9156 days results in an aliasing of the semidiurnal M_{2} lunar and S_{2} solar tide at 62.10 days and 58.74 days, respectively, and the diurnal K_{1} component at 173.19 days. Strict application of Rayleigh's criterion for separating the characteristics of these waves in the T/P spectrum requires 3 years to separate M_{2} from S_{2} and 9 years to isolate K_{1} from the semiannual oceanic signal. This difficulty was overcome at the very beginning of the T/P data analysis by estimating tides in bins with a typical size of 3°, which is a little larger than the T/P longitudinal sampling at low latitude. This allows one to combine information from neighboring tracks and thus increase the number of observations per T/P repeat cycle [*Schrama and Ray*, 1994]. Note that this procedure corresponds to differential improvements to a reference tidal model, allowing one to preserve the scales smaller than 3° bins included in the reference.

[3] After a few years of the T/P mission, it became possible to analyze the altimetric data along track (one point every 7 km). These analyses revealed that altimetry detects surface wave features in M_{2} that are too high frequency in space to be barotropic tide features. Except for areas subject to ocean mesoscale activity linked to strong ocean currents and over strong topographic features (continental shelves and ridges), it is likely that we are detecting the part of the internal tides that maintains coherence with the astronomical forcing [*Ray and Mitchum*, 1997].

[4] Now, with more than 10 years of T/P data, determination of the tidal contribution to sea level variability through direct along-track analysis is even more accurate, because of the increase of the signal-to-noise ratio, and more effective separation of the different tidal components.

[5] The aim of this paper is to investigate the convergence and stability of the tidal characteristics directly deduced from along-track analysis of the T/P data and to locate the areas where these characteristics are the most stable. These investigations will consider the barotropic and baroclinic tides. The focus will be on the major M_{2} tidal component.