34 Pierre-Yves Le Traon
seas (e.g. Mediterranean, Black Sea, Caspian Sea). The ability to forecast the mean
sea level change in these high density regions is of outmost importance.
One of the main achievements of TIP has been its ability to measure large-scale
sea level variability. Seasonal variations are mainly due to steric effects related to
heat fluxes (see section 2.3.3). Theyare 180 0 phase-shifted between the two hemispheres and have a typical amplitude of 3 to 5 cm. These variations are correlated
with sea surface temperature (expansion and contraction of surface waters) but
with a phase lag of 1 to 2 months. The amplitude of the signal is smaller in the
southem hemisphere, because of its greater expanse of ocean, and because of the
role of continental air masses in the northem hemisphere. In the tropical oceans,
seasonal variations mainly result from the seasonal variation of the circulation in
response to wind foreing. Interannual variations can also be as large as seasonal
variations. El Nifio events and associated Kelvin waves are also well observed by
TIP (Fall1992 and Winter 1993, Fall1994 and Winter 1995 and mainly Fall1997
and Winter 1998). Several studies oftropieal Kelvin and Rossby waves using TIP
data have shown that they are a potentially important factor in understanding the
coupled ocean/atmosphere system related to the El Nifio Southem Oscillation
(ENSO). Rossby waves and their propagation eharacteristics have also been studied in the mid-latitudes. One of the main results was to show that the propagation
was faster than that of free linear waves (Chelton and Schlax, 1996). This may
have major effects on the way the ocean adjusts to atmospheric forcing.
TIP also provides a unique data set for validating global ocean models and understanding their main deficiencies (Stammer et al., 1996; Fu and Smith, 1997). A
detailed model/data eomparison is a fundamental step before assimilation. Comparison of TIP results with global high-resolution (1/4 or 1/6 degree) simulations
shows that the models still underestimate variability in mid- and high-Iatitude
regions. The seasonal signal is qualitatively weB reproduced. At high frequencies
(higher than 100 days-l), the barotropie response ofthe ocean to wind forcing has
been eharacterized globally, for the first time, by comparing TIP and model results.
We have a1so leamed a lot from altimetry about the mesoscale ocean circulation
variations (e.g. Gulf Stream, Kuroshio, Antarctic Circumpolar Current, Canary
Basin, Mediterranean sea). TIP alone eannot provide a good mapping of mesoseale
signals: at least two satellites are needed. The eontribution ofmerged TIP and ERS1/2 data in providing a detailed description of mesoscale dynamics has been illustrated in, for example, the Mediterranean sea (Ayoub et al., 1998). The ability to
map mesoscale signals was also very well demonstrated by the comparison of TIP
and ERS-l data with in-situ measurements in the Canary basin during the Semaphore experiment (Hemandez et al., 1995). The agreement was better than 3 em
rms, whieh is about the accuracy we expect from in-situ measurements. Satellite
altimetry is actually the only way of precisely monitoring and possibly foreeasting
mesoscale signa1s.
seas (e.g. Mediterranean, Black Sea, Caspian Sea). The ability to forecast the mean
sea level change in these high density regions is of outmost importance.
One of the main achievements of TIP has been its ability to measure large-scale
sea level variability. Seasonal variations are mainly due to steric effects related to
heat fluxes (see section 2.3.3). Theyare 180 0 phase-shifted between the two hemispheres and have a typical amplitude of 3 to 5 cm. These variations are correlated
with sea surface temperature (expansion and contraction of surface waters) but
with a phase lag of 1 to 2 months. The amplitude of the signal is smaller in the
southem hemisphere, because of its greater expanse of ocean, and because of the
role of continental air masses in the northem hemisphere. In the tropical oceans,
seasonal variations mainly result from the seasonal variation of the circulation in
response to wind foreing. Interannual variations can also be as large as seasonal
variations. El Nifio events and associated Kelvin waves are also well observed by
TIP (Fall1992 and Winter 1993, Fall1994 and Winter 1995 and mainly Fall1997
and Winter 1998). Several studies oftropieal Kelvin and Rossby waves using TIP
data have shown that they are a potentially important factor in understanding the
coupled ocean/atmosphere system related to the El Nifio Southem Oscillation
(ENSO). Rossby waves and their propagation eharacteristics have also been studied in the mid-latitudes. One of the main results was to show that the propagation
was faster than that of free linear waves (Chelton and Schlax, 1996). This may
have major effects on the way the ocean adjusts to atmospheric forcing.
TIP also provides a unique data set for validating global ocean models and understanding their main deficiencies (Stammer et al., 1996; Fu and Smith, 1997). A
detailed model/data eomparison is a fundamental step before assimilation. Comparison of TIP results with global high-resolution (1/4 or 1/6 degree) simulations
shows that the models still underestimate variability in mid- and high-Iatitude
regions. The seasonal signal is qualitatively weB reproduced. At high frequencies
(higher than 100 days-l), the barotropie response ofthe ocean to wind forcing has
been eharacterized globally, for the first time, by comparing TIP and model results.
We have a1so leamed a lot from altimetry about the mesoscale ocean circulation
variations (e.g. Gulf Stream, Kuroshio, Antarctic Circumpolar Current, Canary
Basin, Mediterranean sea). TIP alone eannot provide a good mapping of mesoseale
signals: at least two satellites are needed. The eontribution ofmerged TIP and ERS1/2 data in providing a detailed description of mesoscale dynamics has been illustrated in, for example, the Mediterranean sea (Ayoub et al., 1998). The ability to
map mesoscale signals was also very well demonstrated by the comparison of TIP
and ERS-l data with in-situ measurements in the Canary basin during the Semaphore experiment (Hemandez et al., 1995). The agreement was better than 3 em
rms, whieh is about the accuracy we expect from in-situ measurements. Satellite
altimetry is actually the only way of precisely monitoring and possibly foreeasting
mesoscale signa1s.