The western boundary currents can be easily resolved even by the relatively crude Geos-3 radar altimeter because of the large sea level variation across the currents [ e.g., Fu et al., 1987]. A host of studies of the various western boundary currents of the world's oceans have been conducted using the much improved Geosat data. Among the numerous accomplishments is a powerful technique developed by Kelly and Gille [1990] for estimating the absolute dynamic ocean topography across strong currents from their temporal variabilities. An analytical model is used to describe the dynamic topography of an isolated jet with its kinematic parameters (i.e., location, width, and amplitude) determined from the observed temporal sea level changes as an inverse problem (also see Tai, [1990]). Comparison of the altimetry-derived velocities with simultaneous in-situ velocity measurements demonstrated the validity of the technique [ Joyce et al., 1990]. The technique was applied to the Gulf Stream and the Kuroshio for studying their spatial and temporal characteristics with results consistent with in-situ data ( Kelly, [1991]; Qiu et al., [1991]). Tai and White [1990] extended this approach to addressing the eddy-mean flow interaction of the Kuroshio Extension. Other studies based on the technique have been conducted on the recirculation flanking the main current of the Kuroshio [ Qiu, 1992] and the Gulf Stream [ Qiu, 1994]. By combining the altimetric estimate of the surface dynamic topography with historical hydrographic data, Qiu [1994] estimated the deep circulation of the Gulf Stream region with results comparing well to in-situ data. Kelly and Watts [1994] applied the technique to both altimetry data and inverted echo sounder data and obtained excellent agreement. Qiu and Kelly [1993] used altimetrically derived surface current velocities in conjunction with atmospheric wind and thermal forcing to study the heat budget of the upper ocean in the Gulf Stream and the Kuroshio regions.
Using a straightforward technique to examine the cross stream sea level differences, Zlotnicki [1991] reported that both the Gulf Stream and the Kuroshio, after entering the open ocean, had similar seasonal cycles with peak surface currents occurring in the fall, consistent with the findings of the studies cited in the preceding paragraph. The discrepancy from the finding of Fu et al. [1987] that reported a peak Gulf Stream surface velocity in the spring during the Geos-3 mission could be due to interannual variability or large systematic errors in the Geos-3 data. A downstream decrease in the time scales of the Gulf Stream sea level variability was found by Vazquez et al. [1990]. The spatial and temporal characteristics of the meandering of the Gulf Stream was investigated further by Vazquez [1993].
There have been numerous intercomparisons made between Geosat altimetry and in-situ data in the regions of the Gulf Stream and the Kuroshio [ Hallock et al., 1989; Hallock and Teague, 1993; Willebrand et al. 1990; Horton et al., 1992; Blaha and Lunde, 1992; Joyce et al., 1990], demonstrating the quality of the Geosat data for the study of the mesoscale structure of the western boundary currents. Carnes et al. [1990] investigated the feasibility of estimating subsurface temperature fields from altimeter data in the Gulf Stream region with mixed results. Using a combination of the Geosat altimeter data with in-situ hydrographic data, a number of investigators constructed for the Gulf Stream region the so-called synthetic geoid [ Mitchell et al., 1990; Porter et al., 1989, 1992; Glenn et al., 1991], which was useful for deriving approximate absolute Gulf Stream topography for research as well as operational applications. High-resolution geoids in the Gulf Stream region were also obtained by using geodetic techniques [ Rapp and Wang, 1994] and by combining altimetry, hydrography, and direct current velocities [ Kelly et al., 1991]. Rapp and Smith [1994] used the TOPEX/POSEIDON data and the geoid model of Rapp and Wang [1994] to study the characteristics of the Gulf Stream.
The Geosat data have also bee used to study the Loop Current and its eddies in the Gulf of Mexico. Johnson et al. [1992 a] tracked two Loop Current eddies which drifted southwestward across the Gulf. Jacobs and Leben [1990] reported a 10.5-month period for the shedding of eddies by the Loop Current. Leben et al. [1990] constructed maps of the mean sea surface and eddy variability of the Gulf.
Using the Geosat altimeter data with a numerical model, Matano et al. [1993] studied the seasonal variability of the Brazil and the Malvinas Currents and found that the seasonal cycles of the two currents were opposite in phase (with the maximum of the Brazil Current during the austral summer). They interpreted the result in terms of the seasonal migration of the confluence zone of the two currents. Spatial and temporal characteristics of the variability of the Brazil--Malvinas Confluence region were analyzed by Provost and Le Traon [1993]. They reported the dominance of semiannual fluctuations characterized by alternating positive and negative sea level anomalies with length scales of 400--500 km. The mesoscale variability was highly inhomogeneous and anisotropic. They also computed the Reynolds stress and discussed its relation to the dynamics of the mean flow.
The variability of the Agulhas Current was investigated using the Geosat data in a number of studies. Fu and Zlotnicki [1989] calculated the wavenumber-frequency spectrum which revealed the dominant wavelengths and periods as 300--800 km and 50--200 days. Quartly and Srokosz [1993] reported that the seasonal variation evidenced in the Geosat data was not reproduced by the Fine Resolution Antarctic Model (FRAM), suggesting possible directions for the improvement of the model [ Quartly and Srokosz, 1993]. Wakker et al. [1990] calculated the dynamic topography and the eddy variability statistics in the region.