Europe's relative warmth is maintained by the poleward surface branch of the Atlantic Ocean thermohaline circulation. There is paleoceanographic evidence for significant variability and even shifts between different modes of thermohaline circulation. Coupled ocean-atmosphere climate modelling allows first insight into the relative role of the various drivers of the Atlantic thermohaline circulation variability, i.e. the North Atlantic Oscillation, the tropical Atlantic variability, the ocean basin exchanges, small scale processes like high-latitude convection, overflows and mixing as well as effects of changes in the hydrological circle, the atmospheric CO2-content and solar radiation. The strong need for continous model improvement requires concerted efforts in ocean time series observations and relevant process studies. New instrumentation and methods, both for in situ measurements and remote satellite sensing are becoming available to help on the way forward towards as improved understanding of North Atlantic climate varibility.

Using satellite altimetry time series from Topex–Poseidon and ERS-1 to estimate the annual variation of the global mean sea level, we study the global hydrological cycle at the annual frequency. The observed annual sea level signal is first corrected for steric effects and further compared to the annual change in atmospheric water vapor content and soil moisture of continents. The altimetry-derived global mean sea level variation corrected for steric effects, i.e., due to annual ocean mass change amounts to 9.5 mm with a maximum in mid-September. This observation is consistent with the signal estimated from atmospheric and continental data. Accounting for the variations of water mass stored in the superficial soil layers and in the atmosphere, one can get a reasonable phase agreement with satellite observations, with an amplitude difference of only 2.5 mm. This difference may result from uncertainties of the soil moisture estimate, or from neglecting the water storage variations in other reservoirs, such as the rivers, the underground or the ice sheets.

Along‐track sea level anomalies derived from Geosat altimeter data from November 1986 to November 1988 are assimilated by Kalman filtering into a wind‐forced second‐baroclinic vertical mode linear model of the tropical Atlantic Ocean. To save computer time, the filter is degraded, mostly by fixing the error covariance matrix of the estimate once the filter has reached its asymptotic behavior. Geosat altimeter data have been processed using improved corrections. The sea surface height variability signal is extracted using the classical along‐track technique, relative to a complete reference cycle, and using only tracks longer than 2200 km. This processing has preserved oceanic signals both on large scales (above 1000 km) and on the mesoscale (around 200 km). Sea level anomalies predicted at Principe Island are close to in situ tide gage data, though some differences can be partly related to tidal or orbit error corrections. Oceanographic signals are analyzed from two different sets of fields: one issued from anisotropic space‐time objective analysis of Geosat data and the other from the model assimilation. The latter appears as an interesting method to extract low‐frequency and propagating signals. Along the equator, eastward propagating features are consistent with Kelvin waves correlated with zonal wind stress anomalies. Upwelling in the Gulf of Guinea is 1 month earlier in 1987 than in 1988. After elimination of the annual and semiannual signals by harmonic analysis, the residual signal over the whole tropical basin, decomposed into complex empirical orthogonal functions, is found dominated by variations between the 2 years, equatorial and tropical signals being anticorrelated.

Dissolved Ba, Si and alkalinity contents are reported for 11 vertical profiles along a longitudinal section in the western Indian and Southern Oceans, sampled during 1985, 1986 and 1987 in the French expedition INDIGO. Barium concentrations increase from 30 nmol/kg at the surface to 100 nmol/kg at depth in the Southern Ocean. North of the Polar Front, Ba values range from 40 nmol/kg at the surface to 120 nmol/kg in the bottom waters. These vertical variations result from mixing of water masses as well as biologically controlled uptake/regeneration processes. By means of a water-mass mixing model, the non-conservative Ba signal was determined and compared to that of silica. The ΔBaΔSi uptake/regeneration molar ratio is 0.15 × 10−3 in the south and 0.45 × 10−3 in the north. In the south, curiously, no release of Ba is observed in Circumpolar Deep Waters, whereas Ba appears to be regenerated in the underlying Antarctic Bottom Waters. Thermodynamic calculations show that Ba is somewhat supersaturated with respect to barite in the upper 3000 m of the southern waters. Thus, whereas surface Ba uptake is probably governed by the biological activity, Ba regeneration in deep waters appears to be controlled by saturation. Barium is correlated linearly with alkalinity at each station, but the slopes of the regression line vary by a factor greater than 2 along the section. The variability of the ΔBaΔAlk ratio between the Southern Ocean and the subtropical and tropical areas raises questions about the validity of applying ΔBaΔAlk values established on a global scale to reconstruct paleo-alkalinity distributions in the Southern Ocean.

Variations of the 10‐day‐averaged sea level extracted from TOPEX/POSEIDON (T/P) altimeter data are examined from October 1992 to February 1995. Data are corrected from instrumental drifts, and recent precise orbit and ocean tide corrections are used. When data from the less accurate first eight cycles are excluded and atmospheric pressure effects are not corrected, the 10‐day mean sea level varies by 5.0 mm rms. Its variations can be described by a 4.1±1 mm/yr drift, a small annual signal of 2.7 mm amplitude and a residual signal of 3.5 mm rms amplitude, and Gaussian statistics close to formal error estimates. The mean sea level rise is difficult to explain by drifts of geophysical corrections, though the case for the wet tropospheric correction and that for the orbit are uncertain. Pressure effects should be corrected relative to the global pressure averaged over the ocean. The latter shows a seasonal signal of 1 cm amplitude and apparent fluctuations of 30‐ to 70‐day periods. These fluctuations could result from errors in the atmospheric pressure fields, but they seem to be mostly related to large regional pressure evolutions. Evidence is also found for a sampling effect by the T/P orbit of pressure signals at shorter than 10‐day periods. Pressure effects on sea level are then assumed to be an inverse barometer relative to 10‐day global atmospheric pressure over the ocean, which allows analysis of the geographical repartition of the mean sea level evolution due to other effects. The mean sea level rise is comparable in the northern (4.2±2.5 mm/yr) and southern (3.7±1.9 mm/yr) hemispheres. It seems mostly related to interannual variations of the tropical ocean (41% of the global signal) and of the northern hemisphere subtropical gyres (30% of the global signal); residual noise on these regional values seems inversely proportional to the square root of the number of data points in each area. The hemispheric seasonal pressure variation relative to the globally averaged pressure over the ocean is less than 3 mm amplitude; in our calculation it is implicitely assumed that it induces a small interhemispheric water mass exchange. The amplitudes of the seasonal hemispheric sea level variations are 27 and 18 mm in the northern and southern hemispheres, respectively. These amplitudes are very close to being inversely proportional to the ocean surface in each hemisphere. The maxima of sea level in these hemispheres are reached in the last weeks of September and March, respectively. At subpolar latitudes, semiannual components are observed in each hemisphere, which are consistent with those of sea surface temperature.

In order to validate the satellite altimeter measurements over Antarctica, we analyze the Seasat scatterometer backscatter coefficient which has been measured at different incidence and azimuth angles and for the two polarizations. We show that the signal at large incidence angles is a mixed signal of volume, backscattered by subsurface layering and surface scattering whose proportions depend on physical characteristics of the snow. Only the effect of the volume scattering by spherical grains on the altimetric signal can be estimated: It is of the order of a few percents of the total altimetric signal and the induced height error is lower than 25 +/- 20 cm, if the echo is properly retracked. Other effects, such as internal density boundary are not estimated but a mean to detect the areas of great effect is given.

We analyze the Geosat altimeter data off the mouth of the Amazon river. Variations of the sea level of up to 6 m are observed, which are correlated with bathymetry, and dominated by a complex tidal pattern. M2, N2 and O1 altimetric cotidal maps are constructed. Their amplitudes decrease strongly off the river mouth, probably by dissipation in the estuary. Strong maxima are found on both sides of the river. In particular, a M2 amplitude of 2.9 m is found near 2°N–50°W, just north of Cape Norde. These results are in very good agreement with in situ tide gage data as well as with a preliminary analysis of Topex-Poseidon altimeter data. The models of Schwiderski (1980, Review of Geophysics and Space Physics, 18, 243–268) and Cartwright et al. (1991, NASA Technical memorandum, No. 104544) are out of phase with these observations, probably because of their low resolution. The M2 phase signal is rather well reproduced by the high resolution finite element tidal model of the north Atlantic Ocean, developed for the correction of Topex-Poseidon data. However, in this model, amplitude is too low off the river mouth, and the maximum north of Cape Norde is absent. This is probably due to inaccurate bathymetry in the model.

ERS 1 three‐day repeat altimeter Ocean Products have been analyzed from September 3 to December 9, 1991. In the high mesoscale energy areas of the world ocean the obtained height variability is comparable to that derived from 2 years of Geosat altimeter data (17‐day repeat), while it is slightly lower elsewhere. The data reveal some high‐frequency variability of western boundary currents, but not the seasonal signals, nor the low‐frequency mesoscale signal which dominates elsewhere. Isocorrelation maps show that the e ‐folding time is around 10 to 15 days in strong currents. The sea height variance, integrated over all wavelengths in the frequency‐wavenumber spectrum, corresponds to less than 5% of the total 3‐month variance for periods shorter than 20 days, up to 15% of this variance for periods shorter than 34 days, which are respectively the Nyquist frequencies of the TOPEX/POSEIDON and Geosat data. This signal covers a whole range of wavelengths that cannot be obtained from current meter data. In the Gulf Stream area the total dynamic height signal is estimated by adding a climatological dynamic topography to the mesoscale variability. Near 60°W, the height variations apparently correspond to meanders with periods of 30 to 40 days and latitudinal extension about 200 km. They are consistent with westward propagation at about 10 cm/s, and their surface transport varies by about 50% on this timescale, which is in agreement with in situ observations.

Using the altimeter data from the Geosat Exact Repeat Mission, we have produced yearly averaged mean profiles and a global mean sea surface. The radial error of each 6‐day orbital arc computed with the GEM‐T2 geopotential was first estimated by calculating the amplitude and phase of the nine dominant frequencies of the difference between the altimetric profiles and the mean sea surface obtained when adding the permanent sea surface topography (computed from the Levitus' Climatological Atlas) to the GEM‐T2 geoid. We show that this operation is little affected by the choice of the geoid or by its formal error. The resulting correction has been subtracted from each individual arc. Yearly mean profiles were then obtained by averaging the corrected altimetric data of each repeat cycle on a yearly basis. Their noise level is 1 to 2 cm and their resolution is 20 km, but the differences of the altimetric heights at crossovers of ascending and descending tracks are still 30 cm nns. The latter can be reduced to 7 cm rms by a crossover analysis. In addition to the mean values, standard deviations were computed at each point of the repeat cycle. This “yearly along‐track variability” is of the order of 10 cm rms and is dominated by the ocean mesoscale variability. A global yearly mean sea surface has been derived by bilinear interpolation. Its resolution ranges approximately from 160 km to 80 km, depending on the latitude. It is shown to be much less noisy than those deduced from GEOS 3 and Seasat data.

First results with 16 months of Topex/Poseidon data are presented. The unprecedent accuracy (of the order of 5 cm including orbit) of this data set makes possible the study of large-scale signals. The global-averaged sea level exhibit seasonal and near-60-day signals at the sub-centimetric level, only partly explainable by errors or mismodelling of the altimeter corrections and which are strongly correlated with pressure fluctuations. The sea level variations outside the tropics are dominated by the steric height cycle of 5 to 8 cm amplitude. Variations of the major current positions may however be the causes for strong signals in the high eddy energy areas. The authors exhibit also interannual signals of that kind. A first assimilation experiment into a North Atlantic model is described. The dynamics of the Gulf Stream are well reproduced and this approach will help decorrelating the complexity of the ocean circulation in this region.< >

The concentration and isotopic composition of Nd in water and particles collected in the western Mediterranean Sea are studied by two complementary approaches. The first examines local vertical profiles and time series; the second considers the global Nd budget of the whole western Mediterranean Sea. These two approaches are used to quantify the Nd inputs and the dissolved/particulate exchange processes in the water column.Two profiles of Nd in seawater in the Ligurian Sea taken in May and October 1992 show an average epsilon(Nd)(0) = -9.6 +/- 0.5. Seawater from the Strait of Sicily, representative of the eastern waters flowing into the western basin, is more radiogenic [epsilon(Nd)(0) = -7.7 +/- 0.6]. Profiles of particulate matter collected in sediment traps in coastal (Gulf of Lions) and offshore (Ligurian Sea) environments are also shown. Particles are enriched in Nd and are more radiogenic near the coast than offshore. Measurements of Nd concentration and epsilon(Nd)(0) of external sources to the western Mediterranean Sea compared with the literature data demonstrate that particulate flux of atmospheric Saharan origin are more rich ([Nd] = 38 +/- 10 mug/g) and less radiogenic [epsilon(Nd)(0) = -13.0 +/- 1.0] than riverine particulate discharge ([Nd] = 21.5 +/- 4.4 mug/g; epsilon(Nd)(0) = -10.1 +/- 0.5), allowing to trace Nd particulate inputs in the water column. Nd atmospheric flux appears to be the major source into the whole western basin, although lateral advection of riverine material is the prevailing process in the coastal environment.Offshore, the vertical propagation of an important Saharan dust event has been recorded for two months in sediment traps at 80, 200 and 1000 m. The evolution of the resulting negative epsilon(Nd)(0) peak along depth and time shows that the particles reach 200 m on a time scale of one week.For the first time, the Nd budget in the western Mediterranean basin is constrained by both concentrations and isotopic compositions measured in particles and seawater. Surface budget requires a remobilization of 30 +/- 20% of particulate Nd input. In deep water, dissolved Nd concentrations are balanced by a scavenging of 10 +/- 20% of the sinking particulate flux. On the other hand, the deep isotopic compositions suggest an exchange between 30 +/- 20% of the sinking particles and the deep waters. The hypothesis of a non-stationary regime for the surface waters in the Ligurian Sea is also considered.

Altimeter data are helpful in monitoring the evolution of polar caps as well as constraining their dynamics. In the recent years, our group has quantified the impact of the so-called volume echo on the determination of the surface height, analyzed the effect of surface roughness on the intensity of the radar signal, adapted an inverse technique for mapping the topography of the ice sheets and its error, including correction of the surface slope effect and finally used the Antarctica ice sheet topography deduced from Seasat data to estimate the rheological parameters of the ice flow.

We compare the simulation of the O2 concentration in the mixed layer at Sta. P using two different mixed layer models with the same biological production-consumption function. One is the integral mixed layer model already used by Thomaset al. (Deep-Sea Research, 37, 463–491, 1990). The other is the eddy kinetic energy (EKE) model of Gasparet al. (Journal of Geophysical Research, 95, 16179–16194, 1990). The latter simulates better both the seasonal and the short time evolution of the oxygen concentration. The submixed layer summer supersaturation is also closer to observations, by about 25%, though a factor of 2 too high: this could be improved by adjustment of the production function. The net annual gas exchange flux with the atmosphere is always a degassing of the order of 10% of the total biological production. In general, the model values are smaller (within a factor of 3 for the EKE model) than the data. However, the latter may not be robust as 96 measurements per year are necessary to estimate this flux within a factor of 10.

In order to estimate the accuracy of altimetric height measurements over ice sheets, an altimeter wave-form simulator has been developed, and different tracking methods have been tested. A large range of surface features, including large-scale and medium-scale features and micro-roughness have been taken into account for modeling of either surface- or volume-scattering. A large set of parameters affects the trailing edge of the radar wave form, so that re-tracking algorithms based on the detection of its leading edge provide better retrievals of the surface height than those based on the analysis of the whole wave form. A volume component is clearly present in the radar wave forms; its effect on the leading edge depends mostly on the snow grain-size (and therefore on the snow temperature) and on the pointing angle. However, on average, the induced error on the snow-surface height estimation should only be around 25 cm.

We derived the sea level variability in the South Atlantic subtropical gyre between 15° and 35°S from 2 years of the Geosat exact repeat mission altimeter data. On the mesoscale, along‐track wavenumber spectra are those of nonlinear dynamics with a steeper slope and stronger energy along the eastern boundary. Frequency/wavenumber spectra reveal a significant semiannual signal at a wavelength of about 500 km. West of the Walvis ridge (5°E), this signal appears to correspond to semiannual Rossby waves of 2‐ to 3‐cm amplitude, propagating westward at 3 cm s−1, and showing latitude refraction effects. The pattern is very similar to that of annual Rossby waves as modeled by Reason et al. (1987). These waves are probably related to the semiannual wind component. However, understanding the exact mechanism for their excitation requires dynamical modeling. The large‐scale variability also shows a significant semiannual component (about 6‐cm amplitude and 15% of the yearly variance) in phase with climatological winds.

The precision of radar altimetry above an ice sheet can improve glaciological studies such as mass balance surveys or ice-sheet flow models, the first by comparing altimetric data at different times (see this issue), the second by testing or constraining models with data. This paper is a first step towards the latter. From a precise topography deduced by inversion of altimetric data (Remy and others, 1989), we calculate ice-flow direction, balance velocity and basal shear stress. The rheological parameters involved in the relation linking velocity, stress and temperature are then derived by least-squares regression. Ice flow is well represented by setting the Glen parameter,nto 1 ± 0.25 and the activation energy as 70 ± 10 kJ mol−1.

The effect of a poorly constrained repetitivity of the orbit of an altimetric satellite is analyzed. From existing data, 35% of the marine geoid slopes are found to exceed 1.5 cm km-1. This may be due either to short-distance-scale features (seamounts, fracture, and subduction zones) or to the large-scale geoid (in the Indian and North Atlantic oceans). A geoid cross-track slope (CTS) can be calculated locally from the tracks inside the repetitivity band. Assuming that the various measurement errors and ocean variability signals are decorrelated, it has a precision of 0.2-2 cm km-1, depending on the orbit cycle (which constrains the number of repeat passes per year) and on the width of the band (from 1 to 10 km in these calculations). This can be used as a correction but increases the noise level by at least 50%. Alternatively, the CTS can be derived from a mean sea surface. This adequately corrects for the large-scale signals but, with present mean sea surfaces, it is inadequate for the short-distance-scale features. Future high-density altimetric samplings such as that of the ERS-1 176-day orbit should improve this precision to about +/-0.1-0.2 cm km-1.Above continental ice, larger than 0.3% along-track slopes were encountered for more than 10% of the time above an altitude of 500 m. These slopes result mostly from undulations of the ice topography. Over one year, a median height profile inside the repetitivity band can be derived at 8-16-cm precision, depending on the number of tracks used and assuming that the measurement noise is 50 cm. From one year to the next, a CTS correction needs to be applied to compare the yearly median height profiles. The latter can be estimated at a 13-130 cm km-1 precision, depending on the repetitivity (from 0.5 to 2.5 km at a latitude of 70-degrees) and the number of tracks.In each case, the precision is comparable with the expected signals (e.g., mesoscale variability of the ocean dynamic topography or climatic variation of the snow accumulation rate). These signals can, however, be recovered by space-time analysis of the data. A more elaborate analysis of the covariances of these corrections is thus required.