The spatial scales of most of the significant errors in altimeter data (the orbit, the tides, the signal delay in the transmission media, the sea-state bias) are generally longer than 1000 km, making the detection of mesoscale features, whose wavelengths are less than 1000 km, relatively less prone to the measurement errors. However, certain along-track smoothing is required to reduce the random measurement noise, limiting the wavelengths of the variability examined to greater than about 60 km [ e.g., Le Traon et al., 1990]. Moreover, a single altimeter mission has only limited capability for synoptic mapping of the details of mesoscale eddies in both space and time [ Chelton and Schlax, 1994].
Maps of the statistics of the global mesoscale variability with much higher accuracy and resolution than the Seasat results have been produced from the Geosat data due to its large data volume and dense ground tracks [ Zlotnicki et al., 1989; Sandwell and Zhang, 1989; Shum et al., 1990 b]. Sandwell and Zhang [1989] computed the sea surface slope, which filtered out most of the large-scale errors, and converted it directly to the kinetic energy of ocean currents. Seasonally-varying mesoscale energy was reported in the northeast Atlantic and the northeast Pacific by Zlotnicki et al. [1989]. Energy level was found to be correlated with the strength of the wind stress curl. Shum et al. [1990 b] obtained similar results. Wavenumber spectra were estimated in a wide range of locations [ Fu and Zlotnicki, 1989; Le Traon et al., 1990; Stammer and Boning, 1992]. Different spectral slopes were found between the high-energy areas and the low-energy areas, but the interpretations were controversial [ Le Traon, 1993; Stammer and Boning, 1993].
De Mey and Menard [1989] combined the Geos-3 data with the Seasat
data in an objective analysis of the eddy field in the Polymode region
(a 500-km square centered at 29
N and 70
W). The
analyzed field was assimilated into an ocean model that yielded
realistic subsurface fields. Bisagni [1991] created time series of sea
level maps using the Geosat crossover data in a similar region,
delineating westward propagating eddy fields that were consistent with
previous in-situ data in the region. Tokmakian and Challenor [1993]
applied the Geosat data to studying the eddy field in the Azores frontal
region and the Canary basin of the North Atlantic. They reported
slightly higher eddy energy in winter and discussed evidence for
Rossby waves, although the effects of tidal aliasing was a concern for
the detection of Rossby waves of the annual period [ Schlax and
Chelton, 1994]. The Geosat data were also applied to the North
Atlantic Current [ De Mey, 1992] and the frontal zones east of Iceland
[ Robinson et al., 1989]. Significant correlations between Geosat and
in-situ observations were reported even in areas of extremely low eddy
energy such as the Iberian Basin [ Stammer et al., 1991] and the Cape
Verde Frontal Zone [ Zlotnicki et al., 1993]. A systematic study of the
relationship between the time scale and the space scale of the eddy
field in the North Atlantic was conducted by Le Traon [1991], who
identified two regimes: higher energy areas where the space and time
scales are proportional to each other, consistent with the
quasigeostrophic turbulence theory; low energy areas where the space
and time scales are inversely related, consistent with the linear Rossby
wave dynamics.
The spatial and temporal characteristics of the eddy field in the South Atlantic were investigated by Le Traon and Minster [1993] and Forbes et al. [1993]. Both reported evidence for westward propagating Rossby waves: the former emphasized the waves of the semiannual period at the subtropical latitudes west of the Walvis ridge; the latter identified the Agulhas Retroflection as a source for the waves with periods of 400--500 days. The variation of the characteristics of the eddy wavenumber spectrum in the region was extensively examined with results consistent with the findings of Fu and Zlotnicki [1989] and Le Traon et al. [1990].
The mesoscale variability of the Antarctic Circumpolar Current was investigated by Chelton et al. [1990], who reported that the eddy field was highly correlated to the mean flow and both were apparently steered by the bathymetry (also see Sandwell and Zhang, [1989]). Eddy Reynolds stress was estimated at the crossover locations of Geosat [ Morrow et al., 1992, 1994; Johnson et al., 1992 b]. The resulting horizontal eddy momentum flux tends to generally accelerate the mean flow, but the zonally averaged eddy flux is an order of magnitude too small, and in the wrong direction to balance the eastward momentum input from the wind.
Jacobs et al. [1993] fit a Rossby wave model to the Geosat data in the
entire Pacific Ocean. They found that less than 5% of the variance
could be accounted for by the model. However, most of the estimated
wave amplitude was above the estimation error. Van Woert and Price
[1993] reported evidence for Rossby waves north of the Hawaii
Islands and found close agreement of their propagation characteristics
and latitudinal dependence with the theoretical dispersion relation.
White et al. [1990 b] demonstrated that the annual fluctuations in the
California Current region were consistent with Rossby waves
propagating into the open ocean with wave fronts parallel to the
bathymetry. Kelly et al. [1993] discussed the northern limit (37
N) for Rossby wave propagation in the region. The eddy activities in
the California Current were found to be consistent with the
simulations of a primitive-equation model [ Pares-Sierra et al., 1993],
indicating that the eddies were forced by the wind adjacent to the
coast. The annual-period Rossby waves were also detected in the
Alaska Gyre [ Matthews et al., 1992]. Again the finding must be
viewed with caution due to the aliasing effects of the tides [ Schlax and
Chelton, 1994]. Evidence for topographic planetary waves in the
Bering Slope Current was reported by Okkonen [1993].
The formation, propagation, and mass transport of eddies were studied in detail by a number of investigators. Didden and Schott [1993] examined the eddies in the retroflection region of the North Brazil Current. Individual eddies were traced to the entrance of the Caribbean. Seasonal variation of the eddy energy was correlated to the seasonal cycle of the retroflection of the North Brazil Current into the North Equatorial Counter Current. The interhemispheric exchange of water mass was estimated to be 3 Sv. Gordon and Haxby [1990] were able to trace the eddies spun off from the retroflection of the Agulhas Current all the way to the western South Atlantic. They estimated as much as 15 Sv of the Indian Ocean water being transported to the South Atlantic by this process. Okkonen [1992] employed a novel eddy-tracking technique to follow and diagnose the shedding of an eddy from the Alaskan Stream. Ichikawa and Imawaki [1994] documented for the first time the complete life history of a cold-core eddy shed from the Kuroshio.