Abstract. A major accomplishment of the recently completed Tropical Ocean-Global Atmosphere (TOGA) Program was the development of an ocean observing system to support seasonal-to-interannual climate studies. This paper reviews the scientific motivations for the development of that observing system, the technological advances that made it possible, and the scientific advances that resulted from the availability of a significantly expanded observational database. A primary phenomenological focus of TOGA was interannual variability of the coupled oceanatmosphere system associated with E1 Nifio and the Southern Oscillation (ENSO). Prior to the start of TOGA, our understanding of the physical processes responsible for the ENSO cycle was limited, our ability to monitor variability in the troi•ical oceans was primitive, and the capability to predict ENSO was nonexistent. TOGA therefore initiated and/or supported efforts to provide real-time measurements of the following key oceanographic variables: surface winds, sea surface temperature, subsurface temperature, sea level and ocean velocity. Specific in situ observational programs developed to provide these data sets included the Tropical AtmosphereOcean (TAO) array of moored buoys in the Pacific, a surface drifting buoy program, an island and coastal tide gauge network, and a volunteer observing ship network of expendable bathythermograph measurements. Complementing these in situ efforts were satellite missions which provided near-global coverage of surface winds, sea surface temperature, and sea level. These new TOGA data sets led to fundamental progress in our understanding of the physical processes responsible for ENSO and to the development of coupled ocean-atmosphere models for ENSO prediction.And thorough this distemperature we see the seasons alter

In 1978, during the last 25 days of the SEASAT mission, the satellite ground tracks repeated within 2.5 km every 3 days. This yielded eight to nine sets of global collinear altimeter data with a cross-track grid spacing of approximately 900 km at the equator and 600 km at mid-latitude. Because the geoid is time-invariant, such observations can reveal sea surface height variability due to dynamic ocean phenomena. We have solved for variations due to deep-ocean mesoscale features by eliminating the longer wavelength deviations. Modeled tidal heights were first subtracted from the altimeter data, and linear trends were then removed from collinear segments approximately 2000 km in length. This second step eliminates relative orbit error together with any residual tide model errors. The resulting sea height profiles have a precision of a few centimeters, and the global variability map constructed from these data reveals a strikingly realistic view of mesoscale energetics. Maximum values of 20-40 cm rms variability are generated by meanders and eddies of five major current systems: Gulf Stream, Kuroshio, Agulhas, Antarctic Circumpolar, and Falkland/Brazil confluence. Perhaps a more significant discovery was the dominance of exceedingly small variability over vast regions of the oce•/ns; approximately 70% of all global values were less than 5 cm. This category included the eastern North Pacific, eastern South Pacific, and almost the entire South Atlantic where values as small as 1-2 •m were common. With such a low level of background noise, even some currents with relatively small sea height signatures could be detected. In both the Atlantic and the Pacific, for example, the North Equatorial Current systems were clearly expressed as zonal bands of higher variability. 4343 surement capability of satellite altimeters. Geoid-independent techniques have therefore been developed to determine sea height variability from altimeter data'. One of these is the 'crossover difference' approach, in which altimeter measurements are evaluated at intersections of ascending and descending ground tracks. At the crossing point, the static part of the ocean surface height caused by the geopotential contribution is the same on both tracks and can thus be eliminated by a simple difference of the two measurements. When altimetric analyses are restricted to the deep ocean, long wavelength signals due to orbit error and tides can be separated from those due to mesoscale features, and variability of the eddy field is obtained. Maps of variability produced from dense gr'fd• of altimetric data can thus be used to characterize the m•eS0scale energetics of a region or to monitor paths of currents. Application of this concept has been demonstrated by using SEASAT altimeter data in the North Pacific [Marsh et al., 1982] and GEOS 3 data in the North Atlantic [Cheney and Marsh, 1981b]. In the latter case, the altimetric map of sea height variability clearly delineated the energetic Gulf Stream• meander domain in contrast to the less variable background ...

The unprecedented accuracy of TOPEX/POSEIDON (T/P) altimeter data warrants a new evaluation of the methods typically used to form time series of sea level change. Whereas explicit removal of orbit error has always been required as a first step in altimeter data processing, the T/P analysis presented here is based simply on unadjusted, monthly averages. This approach has the advantage of retaining the large‐scale ocean signal, which would be distorted by orbit adjustment. Using 16 months of data, we have evaluated the T/P monthly means on spatial scales ranging from mesoscale to global. In the tropical Pacific, comparisons with 17 island tide gauge records and dynamic height derived from 36 thermistor moorings show that the altimeter data have an accuracy of approximately 2 cm when averaged over spatial scales of a few hundred kilometers. On basin scales in the northern hemisphere, similar agreement is found between the T/P data and the dynamic height climatology of Levitus (1982). These new altimeter observations are thus providing the first reliable view of global sea level changes on seasonal‐to‐interannual timescales.

No abstract

Two years of GEOSAT altimeter observations are used to investigate the response of sea level to anomalous westerly wind bursts in the tropical Pacific Ocean before and during the 1986-87 El Niño. Sea level time series along the equator show examples of both positive and negative anomalies of 10-centimeter amplitude and 2- to 4-week time scale propagating across the Pacific with phase speeds of 2.4 to 2.8 meters per second, suggesting downwelling and upwelling Kelvin waves, respectively. A comparison of island wind observations with sea level indicates one instance (May 1986) in which a positive sea level anomaly can be related to westerly winds caused by a cross-equatorial cyclone pair in the western Pacific. This episode was followed by additional wind bursts later in the year, and finally by sustained westerlies in the western Pacific during November-December 1986, at the height of El Niño. The GEOSAT observations reveal the sea level response to these meteorological events and provide a synoptic description of the El Niño oceanographic phenomenon.

Geosat altimeter data are used to obtain the first detailed basin‐wide measurement of the meridional transport of warm surface water in the tropical Pacific during El Niño. Using a combination of crossover and collinear difference techniques, continuous sea level time series are constructed on a 2°×1° grid covering the Pacific between 20°N and 20°S for the 4‐year period 1985–1989. Zonal integrations of these data are performed over three latitude bands to examine large‐scale sea level changes. Comparison with tide gauge data suggests that these zonal averages have an accuracy of better than 1 cm. The Geosat analyses show a clear pattern of water exchange involving principally the equatorial and north equatorial regions. From the onset of the warm event in late 1986 to the mature phase in mid‐1987, mean sea level in the equatorial region dropped nearly 5 cm while simultaneously in the north equatorial region it increased by about the same amount. These anomalies, equivalent to 2×104 m3 of upper layer water, persisted for nearly 2 years before both regions returned to normal. A similar pattern and amplitude of north‐south water exchange is also observed on the seasonal time scale, consistent with annual variation of the wind stress curl. Water moves northward across 8°N during boreal winter and southward during summer with a net transport of approximately 30 Sv. Thus, in terms of meridional transport, the 1986–1987 El Niño is seen as a low‐frequency modulation of the normal seasonal cycle. In contrast to the widely held view that a surplus of equatorial water is required prior to El Niño, no such buildup was observed prior to the 1986–1987 event.

Repeated Geosat altimeter measurements of sea level at ground track intersections (crossovers) and along collinear nests of profiles were used to evaluate Geosat data by (1) construction of a regional, precise surface made up of intersecting sea level profiles, (2) determination of the magnitude of the electromagnetic (EM) bias, and (3) generation of sea level time series and anomaly maps. The first procedure, after elimination of radial ephemeris error and other long‐wavelength signals, established the data's internal precision at a few centimeters. With regard to the EM bias, comparing repeated sea level profiles for different values of significant wave height (SWH) gave a value of about 1% of SWH. The final process yielded a record of sea level whose accuracy was tested by comparison with in situ measurements and wind‐driven model results. On the basis of these analyses we conclude that Geosat data can determine sea level changes with an rms accuracy of approximately 4 cm for time scales of a month or longer. For determining sea level variability, Geosat geophysical data records are a high‐quality, long‐term data set of significant value for the study of ocean dynamics.

Individual Seasat altimeter profiles in the western North Atlantic have been differenc•l with the best available geoid model to remove the gravitational component. The resulting sea surfac• height profiles compare remarkably well with independent oceanographic observations. The Gulf Stream is clearly apparent in each profile as a 1-to 2-m step, and known positions of cyclonic and anticyclonic rings correspond with depressions and elevations, respectively, with amplitudes as large as 95 cm. Some of the most important altimeter data for analyzing the dynamic ocean signal were gathered during the last month of the mission when the same ground track was repeated every 3 days. These data allow detailed examination of time-varying ocean phenomena, since the gravitational component is time-invariant. One set of collinear passes dearly shows a cyclonic ring as it moves out from underneath the spac•craft's track, an event which was simultaneously observed with the aid of a satellite-tracked surfac• buoy. Another striking feature seen in the altimeter data set is the apparent variability exhibited by the Gulf Stream. On time scales of a few days, surfac• transport indicated by the sea surfac• height differenc• across the stream varied by nearly 30%, and over the entire 3-month period much larger fluctuations were observed, suggesting significant changes in total mass transport. Altimetry may provide an effective means of determining the time and spac• scales associated with these variations.