The current ocean tide version is calculated from the NASA GSFC, GOT00.2 ocean tide model. Earlier versions of the GOT ocean tide model are described below.
Version 2.0
The ocean tide `correction' to the altimetry is the largest of all standard corrections. For example, in an analysis of collinear differences of sea-surface heights, Ray, Koblinsky, & Beckley [International Journal of Remote Sensing, 12, 1979, 1991] found that the ocean tides were responsible for more than 80% of the signal variance.
The Pathfinder model used for removing the ocean-tide signal from the altimeter data is actually a combination of many models. The primary deep-ocean model is an updated version of the one described by Schrama and Ray [Journal of Geophysical Research, 99, 24799, 1994]. This is version 960104 (hereafter SR960104), but it is nearly identical to version 950308, which was studied extensively by Shum et al. [JGR, in press, 1996]. The model is a long-wavelength adjustment to the FES94.1 hydrodynamic model of Le Provost et al [JGR, 99, 24777, 1994] (this is also the case for Richard Eanes's CSR3.0 model). Cycles 9 to 71 of both TOPEX and POSEIDON data were used in the model development; an adjustment for the JGM3 ephemerides was done so that the model is consistent with the second-generation orbits available for TOPEX/POSEIDON. Model SR960104 differs from SR950308 only in some slight adjustments toward higher latitudes which were necessary to extend the model into latitudes above 66 degrees. The model is given on a 0.5-degree grid.
The SR960104 model has 7 primary tidal constituents: O1, P1, K1, N2, M2, S2, and K2. The Q1 constituent has been adopted directly from model FES95.2.1 of Le Provost et al.; this Q1 is an assimilation solution, based on TOPEX/POSEIDON data. In addition to these 8 constituents, 16 minor consituents are accounted for in the tidal-height predictions; these terms are inferred from the major tides by the admittance method [Munk & Cartwright, Philosophical Transactions of the Royal Society, A259, 533-581, 1966]. Nodal corrections for lunar tides are, of course, incorporated. Finally, the long-period tides are handled similarly to the T/P GDR's: they are assumed to be in equilibrium; the 15 largest spectral lines in the Cartwright-Tayler-Edden tables [Geophysical Journal of the Royal Astronomical Society, 33, 253-264, 1973] have been used, including the 18.6-year nodal tide but not the permanent tide.
In the polar oceans (latitudes above 66 degress), including Hudson Bay, the ocean-tide model is FES94.1, although with some slight adjustments near the transition to Schrama-Ray in order to allow for (relatively) smooth corrections without boundary jumps. Therefore, all latitudes overflown by the ERS-1 altimeter have ocean-tide corrections available. However, they should be considered less accurate than latitudes overflown by TOPEX/POSEIDON where the SR960104 model directly incorporates the T/P altimetry.
To see the difference between this Pathfinder model (including the adopted regional models--see discussion below) and the CSR3.0 model produced by Richard Eanes (University of Texas), click here.
No available global T/P-based tide model is likely to be accurate in shallow seas. In these areas, the tides are spatially complex and the spatial averaging that is necessary to overcome tidal aliasing problems in altimetry cannot be performed without considerable distortion of the tidal signal. The use of local hydrodynamic models may therefore often be preferable.
The Pathfinder project has begun an effort to supplement the above global tide model with a number of local high-resolution hydrodynamic models for various marginal and inland seas. For the ocean-tide corrections in current use, the chart below shows which model has been used in which location.
The red areas denote locations where local ocean-tide models have been adopted. These models are:
In addition, inland seas as well as the Black Sea and Baltic Sea are assumed to have no tides, which is not strictly true, but no accurate models are currently available. The one body of water with known large tides but with no available (and reasonably accurate) model is the Red Sea.
We are accumulating additional local models for this project, and new versions of our altimeter products will gradually reflect this.
The utility of local models for handling the tide corrections in marginal and shallow seas can be seen from the following experiments. Two-cycle differences of TOPEX collinear sea-surface heights were processed along the three tracks shown in the diagrams below (track 101 across the Gulf of Maine and tracks 3 and 54 across the Persian Gulf). The rms of the height differences were tabulated using (a) no tide correction at all (b) the tide correction adopted for the second release of the GDRs (i.e. CSR3.0) and (c) the local models adopted in this project. Results are:
___________________________________________________________
RMS of collinear differences (cm)
-----------------------------------------
track no tide new GDR Pathfinder
correction correction correction
----- ---------- ---------- ----------
101 133 95 18.7
3 39 34 15.3
54 68 75 20.4
----------------------------------------------------------
The reduced variances resulting from using the local models provides
convincing evidence of the value of these models.
Location of track 101:
Location of tracks 3 and 54:
To see an earlier version (Version 1) of this page, click here.