## Tidal Prediction

If tides in the ocean were in equilibrium with the tidal potential, tidal prediction would be much easier. Unfortunately, tides are far from equilibrium. The shallow-water wave which is the tide cannot move fast enough to keep up with sun and moon. On the equator, the tide would need to propagate around the world in one day. This requires a wave speed of around 460m/s, which is only possible in an ocean 22km deep. In addition, the continents interrupt the propagation of the wave. How to proceed?

We can separate the problem of tidal prediction into two parts. The first deals with the prediction of tides in ports and shallow water where tides can be measured by tide gauges. The second deals with the prediction of tides in the deep ocean where tides cannot be easily measured.

Tidal Prediction for Ports and Shallow Water
Two methods are used to predict future tides at a tide-gauge station using past observations of sea level measured at the gauge.

The Harmonic Method This is the traditional method, and it is still widely used. The method uses decades of tidal observations from a coastal tide gauge from which the amplitude and phase of each tidal constituent (the tidal harmonics) in the tide-gage record are calculated. The frequencies used in the analysis are specified in advance from the basic frequencies given in Table 17.1.

Despite its simplicity, the technique had disadvantages compared with the response method described below.

1. More than 17.6 years of data are needed to resolve the modulation of the lunar tides.
2. Amplitude accuracy of 10-3 of the largest term require that at least 39 frequencies be determined. Doodson found 400 frequencies were needed for an amplitude accuracy of 10-4 of the largest term.
3. Non-tidal variability introduces large errors into the calculated amplitudes and phases of the weaker tidal constituents. The weaker tides have amplitudes smaller than variability at the same frequency due to other processes such as wind set up and currents near the tide gauge.
4. At many ports, the tide is non-linear, and many more tidal constituents are important. For some ports, the number of frequencies is unmanageable. When tides propagate into very shallow water, especially river estuaries, they steepen and become non-linear. This generates harmonics of the original frequencies. In extreme cases, the incoming waves steepens so much the leading edge is nearly vertical, and the wave propagates as a wall of water. This is a tidal bore.

The Response Method This method, developed by Munk and Cartwright (1966), calculates the relationship between the observed tide at some point and the tidal potential. The relationship is the spectral admittance between the major tidal constituents and the tidal potential at each station. The admittance is assumed to be a slowly varying function of frequency so that the admittance of the major constituents can be used for determining the response at nearby frequencies. Future tides are calculated by multiplying the tidal potential by the admittance function.

1. The technique requires only a few months of data.
2. The tidal potential is easily calculated, and a knowledge of the tidal frequencies is not needed.
3. The admittance is Z ( f ) = G ( f ) / H ( f ). G ( f ) and H ( f ) are the Fourier transforms of the potential and the tide gage data, and f is frequency.
4. The admittance is inverse transformed to obtain the admittance as a function of time.
5. The technique works only if the waves propagate as linear waves.

Tidal Prediction for Deep-Water
Prediction of deep-ocean tides has been much more difficult than prediction of shallow-water tides because tide gauges were seldom deployed in deep water. All this changed with the launch of Topex/Poseidon. The satellite was placed into an orbit especially designed for observing ocean tides (Parke et al., 1987) , and the altimetric system was sufficiently accurate to measure many constituents of the tide. Data from the satellite have now been used to determine deep-ocean tides with an accuracy of ± 2cm. For most practical purposes, the tides are now known accurately for most of the ocean.

Several approaches have led to the new knowledge of deep-water tides using altimetry.

Prediction Using Hydrodynamic Theory Purely theoretical calculations of tides are not very accurate, especially because the dissipation of tidal energy is not well known. Nevertheless, theoretical calculations provide insight into processes influencing ocean tides. Several processes must be considered:

1. The tides in one ocean basin perturb Earth's gravitational field, and the mass in the tidal bulge attracts water in other ocean basins. The self- gravitational attraction of the tides must be included.
2. The weight of the water in the tidal bulge is sufficiently great that it deforms the seafloor. The Earth deforms as an elastic solid, and the deformation extends thousands of kilometers.
3. The ocean basins have a natural resonance close to the tidal frequencies.
4. The tidal bulge is a shallow-water wave on a rotating ocean, and it propagates as a high tide rotating around the edge of the basin. Thus the tides are a nearly resonant sloshing of water in the ocean basin. The actual tide heights in deep water can be higher than the equilibrium values noted in Table 17.2.
5. Tides are dissipated by bottom friction especially in shallow seas, by the flow over seamounts and mid-ocean ridges, and by the generation of internal waves over seamounts and at the edges of continental shelves. If the tidal forcing stopped, the tides would continue sloshing in the ocean basins for several days.
6. Because the tide is a shallow-water wave everywhere, its velocity depends on depth. Tides propagate more slowly over mid-ocean ridges and shallow seas. Hence, the distance between grid points in numerical models must be proportional to depth with very close spacing on continental shelves (LeProvost et al., 1994).
7. Internal waves generated by the tides produce a small signal at the sea surface near the tidal frequencies, but not phase-locked to the potential. The noise near the frequency of the tides causes the spectral cusps in the spectrum of sea-surface elevation first seen by Munk and Cartwright (1966). The noise is due to deep-water, tidally generated, internal waves.

Altimetry Plus Response Method Several years of altimeter data from Topex/Poseidon have been used with the response method to calculate deep-sea tides almost everywhere equatorward of 66 (Ma et al., 1994). The altimeter measured sea-surface heights in geocentric coordinates at each point along the subsatellite track every 9.97 days. The temporal sampling aliased the tides into long frequencies, but the aliased periods are precisely known and the tides can be recovered (Parke et al., 1987). Because the tidal record is shorter than 8 years, the altimeter data are used with the response method to obtain predictions for a much longer time.

Recent solutions by ten different groups, have accuracy of ± 2.8cm in deep water (Andersen, Woodworth, and Flather, 1995). Work has begun to improve knowledge of tides in shallow water.

Maps produced by this method show the essential features of the deep-ocean tides (Figure 17.13). The tide consists of a crest that rotates counterclockwise around the ocean basins in the northern hemisphere, and in the opposite direction in the southern hemisphere. Points of minimum amplitude are called amphidromes. Highest tides tend to be along the coast. Figure 17.13 Global map of M2 tide calculated from Topex/Poseidon observations of the height of the sea surface combined with the response method for extracting tidal information. Full lines are contours of constant tidal phase, contour interval is 30°. Dashed lines are lines of constant amplitude, contour interval is 10cm. From Richard Ray, NASA Goddard Space Flight Center.
Altimetry Plus Numerical Models Altimeter data can be used directly with numerical models of the tides to calculate tides in all areas of the ocean from deep water all the way to the coast. Thus the technique is especially useful for determining tides near coasts and over seafloor features where the altimeter ground track is too widely spaced to sample the tides well in space. Tide models use infinite-element grids similar to the one shown in Figure 15.4. Recent numerical calculations by (LeProvost et al., 1994; LeProvost, Bennett, and Cartwright, 1995) give global tides with ± 2-3cm accuracy and full spatial resolution.

Further improvements will lead to solutions at the ultimate limits of practical accuracy, which is about ± 1-2cm. The limit is set by noise from internal waves with tidal frequency, and the small, long-term variations of depth of the ocean. Changing heat content of the ocean produces changes in oceanic topography of a few centimeters, and this changes ever so slightly the velocity of shallow-water waves.

Tidal Dissipation
Tides dissipate 3.75 ± 0.08TW of power (Kantha, 1998), of which 3.5TW are dissipated in the ocean, and much smaller amounts in the atmosphere and solid Earth. The dissipation increases the length of day by about 2.07 milliseconds per century, it causes the semimajor axis of moon's orbit to increase by 3.86cm/yr, and it mixes water masses in the ocean.

The calculations of dissipation from Topex/Poseidon observations of tides are remarkably close to estimates from lunar-laser ranging, astronomical observations, and ancient eclipse records. Our knowledge of the tides is now sufficiently good that we can begin to use the information to study mixing in the ocean. Remember, mixing drives the abyssal circulation in the ocean as discussed in §13.2 (Munk and Wunsch, 1998). Who would have thought that an understanding of the influence of the ocean on climate would require accurate knowledge of tides?
Department of Oceanography, Texas A&M University
Robert H. Stewart, [email protected]