1. Introduction
The Atlantic meridional overturning circulation (AMOC) plays a major role in the heat budget of the North Atlantic region. Hall and Bryden (1982) showed from observations that at 26°N the Atlantic circulation carries 1.3 ± 0.3 PW of heat northward, mostly within the AMOC. Their results were subsequently confirmed by global ocean inverse analyses (e.g., Ganachaud 2003b). The heat carried by the AMOC accounts for one-quarter of the maximum global meridional heat transport required by the coupled ocean–atmosphere system to balance the global radiation budget. The Intergovernmental Panel on Climate Change considers it “very likely” that the AMOC will significantly weaken over the twenty-first century as a consequence of anthropogenic greenhouse gas emissions (Houghton et al. 2001), thus reducing the oceanic supply of heat to the North Atlantic region. Model simulations also suggest natural AMOC variability on intraseasonal to multidecadal time scales (e.g., Delworth et al. 1993; Jayne and Marotzke 2001; Latif et al. 2004; Biastoch et al. 2008; Wunsch and Heimbach 2009). A pronounced seasonal variability between the equator and midlatitudes reflects seasonally varying Ekman transports (e.g., Jayne and Marotzke 2001; Wunsch and Heimbach 2009). Multidecadal AMOC variability is thought to be linked to North Atlantic sea surface temperature changes (e.g., Delworth et al. 1993; Latif et al. 2004). However, owing to a lack of observations, the existence of this link in the real ocean remains uncertain (Kanzow and Visbeck 2009).
Daily estimates of the basinwide full-water-column-integrated AMOC became available when the U.K–U.S. Rapid Climate Change (RAPID)/meridional overturning circulation (MOC)/heat flux array (MOCHA) monitoring array (Kanzow et al. 2008a) (Fig. 2) across the Atlantic along 26.5°N became operational in April 2004. During the first year of array observations the AMOC had a mean strength of the upper cell of 18.7 Sv and an intraseasonal variability of 5.6 Sv rms (Cunningham et al. 2007).
In this study we describe seasonal AMOC transport variations as observed by the RAPID-MOC/MOCHA array at 26.5°N between April 2004 and April 2008. First, we present the dataset and the methodology underlying the computation of the strength and structure of the AMOC (section 2). We then discuss the temporal variability and vertical structure of the AMOC, with a focus on seasonal anomalies (section 3). In section 4 a forcing mechanism is proposed that accounts for a large fraction of the seasonal AMOC anomalies. Possible implications of our results are discussed in section 5. Conclusions are given in section 6.
2. Data and methods
a. Data
The three components of Ψ(z) at 26.5°N are the Gulf Stream (TGS), Ekman (TEK), and midocean (TMO) transports (Kanzow et al. 2007; Cunningham et al. 2007). The bulk of northward Gulf Stream volume transport, TGS, has been monitored using a submarine cable and repeated ship sections nearly continuously since 1982 (Larsen 1992; Meinen et al. 2010). The vertical structure of TGS(z) is inferred from TGS as described by Baringer et al. (2008).
The northward Ekman transport zonally integrated between the shelf of Abaco (Bahamas) and the African coast is estimated as the zonal integral of the zonal component of the wind stress from spaceborne Quick Scatterometer (QuikSCAT) measurements (Schlax et al. 2001). Then TEK is assumed to be distributed evenly between the surface and 100 m, to obtain a vertical profile of transport per unit depth TEK(z).
For the midocean geostrophic transport TMO, we use the RAPID–MOC/MOCHA moorings. To directly measure strong flows at the western boundary, moorings WB0–WB3 (Fig. 3) are equipped with current meters at discrete levels distributed throughout the water column, and at WBA and WB0 the velocity field in the upper 500 m is profiled by upward-looking acoustic Doppler current profilers (ADCPs) (Johns et al. 2008). All records are 40-h low-pass filtered, subsampled on a 12-hourly grid, and then interpolated onto a spatial grid of 0.5-km zonal and 20-m vertical resolution. Subsequently, profiles of zonally integrated transport (per unit depth) over the 16-km-wide western boundary wedge, TWBW(z), between the Abaco shelf and WB2 (Fig. 3) are computed (Johns et al. 2008).
The remainder of the midocean is measured by moorings near the western and eastern boundaries of the Atlantic and on both flanks of the mid-Atlantic ridge (MAR), which record temperature and salinity at discrete depths (Figs. 2a,b). These records are calibrated and subsequently 2-day low-pass filtered and subsampled at 12-hourly resolution (Kanzow et al. 2006, 2007). At the eastern boundary temperature and salinity data from several moorings have been merged into one profile from 4840 m to the shallowest available level during each deployment (Kanzow et al. 2007, 2009). The western boundary end-of-section profile uses data merged from WB2 shallower than 4000 m and from WBH1/WBH2 (or WB3 after April 2005) at depths greater than that (Fig. 2b). At the western flank of the MAR mooring, MAR1 provides temperature and salinity from the sea surface to 5000 m; on the eastern flank, MAR2 covers the 2500–5000-m interval (Fig. 2b). Filtered and subsampled temperature and salinity data at each site are vertically interpolated onto a 20-dbar grid (Kanzow et al. 2007), from which densities ρ are then computed. Vertical profiles of density at the western and eastern boundaries (ρW, ρE) and on the western and eastern flanks of the mid-Atlantic ridge (ρMARW, ρMARE) are used to compute zonally basinwide-integrated northward geostrophic internal transport per-unit-depth TINT(z) relative to a deep reference level zREF = −4740 m (appendix A). Northward transports of AABW at depths greater than 5000 m are accounted for by extending the transport profile to 6000 m using historical estimates (appendix A). Here TINT(z) and TWBW(z) are used to compute the midocean geostrophic transport (section 2b).
b. Methodology
Since each variable in this study is a function of t, the explicit mentioning of the time dependence will be dropped hereafter. Throughout this study, then, TGS(z), TEK(z), TMO(z), and so on will indicate profiles of transport per unit depth (Sv m−1), whereas TGS, TEK, TMO, etc. will represent transports (Sv) integrated over a vertical range.
c. Isolation of the different transport contributions to the AMOC
It is useful to isolate the contribution of the western and eastern boundaries of the midocean section to fluctuations in ΨMAX, so that physical mechanisms of density changes at either boundary can be studied separately (Longworth 2007). For this, 
3. Results
a. Vertical structure of the flow field across 26.5°N
The April 2004–08 mean profile of TAMOC(z) exhibits northward flow between the surface and 1025 m (Fig. 4), which is a combination of the northward transport of 31.7 ± 0.9 Sv of TGS shallower than 780 m (Beal et al. 2008), 3.5 ± 0.8 Sv of TEK shallower than 100 m, and 0.9 ± 0.2 Sv of TMO (dashed line) in the Antarctic Intermediate Water (AAIW) range between 660 and 1025 m. The bulk of northward flows are opposed by 17.5 ± 1.4 Sv of southward flow of TMO shallower than 660 m (Fig. 4a), with the latter mostly accounting for the recirculation within the subtropical gyre but also containing roughly 5 Sv of northward and western boundary flow within the Antilles Current (Bryden et al. 2005a). Each of the above error envelopes represents the sum of the standard error (SE) and the expected measurement error (appendix B).
There is 20.7 ± 1.9 Sv of southward flow of NADW between 1025 and 5200 (Fig. 4b). In this layer maximum southward transports are found near 1700 m. A time-mean northward transport of 2.1 Sv (appendix A) at depths larger than 5000 m is prescribed, to approximately account for the unobserved AABW flow (Bryden et al. 2005b), which translates in an uncertainty in the time mean ΨMAX of less than ±0.2 Sv (appendix B). Thus, the imposition of a constant AABW transport will only have a small effect on ΨMAX.
Figure 5 shows shapshots every 5 days of the meridional overturning streamfunction Ψ(z) at 26.5°N (1). The time-mean ΨMAX is 18.7 ± 2.1 Sv, with an average zero-crossing depth hZC at 1025 m, varying by 125-m rms. Note that this result illustrates why a “level of no motion” assumption associated with the mean depth of a property interface such as the AAIW–NADW interface is potentially inaccurate (Figs. 5 and 6a).
b. Time-variable meridional flow
Figure 7 shows time series of ΨMAX at 26.5°N and its components. The ΨMAX varies by 4.8 Sv rms (red line), and both it and its components display pronounced intraseasonal and seasonal variability; TGS varies by 2.9 Sv rms, a value representative of the full 1982–2008 record of continuous cable measurements (Meinen et al. 2010) (Table 1); TEK fluctuates by 3.5 Sv rms and is also representative of longer observational periods (Table 1) (Kalnay et al. 1996); TUMO, representing the vertical integral of TMO(z) between the surface and hZC [Fig. 6a, Eq. (6)], displays fluctuations of 3.2 Sv rms. Since no continuous observations of TUMO were made prior to April 2004, the representativeness of this result can only be assessed indirectly (section 5). The correlations for the transport pairs 〈TEK, TGS〉, 〈TEK, TUMO〉, and 〈TGS, TUMO〉 (0.01, −0.11, and −0.21, respectively) are insignificant at 10% error probability, and hence each of them projects on the variance of ΨMAX. The correlations for the transport pairs 〈ΨMAX, TGS〉, 〈ΨMAX, TUMO〉, and 〈ΨMAX, TEK〉 are 0.42, 0.43, and 0.62, respectively, and are all significant at 5% error probability. Although TEK, TGS, and TUMO vary by roughly the same amount, their frequency distribution displays remarkable differences (Fig. 8). The ageostrophic TEK dominates fluctuations of ΨMAX at periods between 10 and 90 days, while the seasonal variability of ΨMAX is dominated by geostrophic (density balanced) components TUMO and TGS. The contribution to ΨMAX from the compensation transport TC(z) at depths shallower than hZC (gray line in Fig. 7) is ±2.3 Sv, somewhat less than the variability of TUMO, TGS, or TEK. As TC compensates for fluctuations in TGS, TEK, and in the observed components of TUMO [i.e., TINT and TWBW; see (5)], it is negatively correlated to all of them (−0.28, −0.41, and −0.42, respectively).
On seasonal time scales, the 180-day low-pass filtered time series of TUMO, TGS, and TEK display fluctuations of 2.2, 1.7, and 1.3 Sv rms, respectively. The sum of the geostrophic upper-ocean transports that contribute to ΨMAX (i.e., TUMO plus TGS) varies by 2.7 Sv rms and clearly dominates over TEK. Moreover, Fig. 8 shows that this result is robust over a 26-yr time series of TGS and TEK (dashed blue and black lines in Fig. 8).
The separate contributions to ΨMAX from the western and eastern boundary variability of the midocean section to
c. Seasonal cycle
Does the ΨMAX or any of the three upper-ocean contributions exhibit a well-developed seasonal cycle? If so, a prediction of ΨMAX, and of its role in ocean heat storage and meridional heat transport on seasonal time scales might be possible, provided the physics of the forcing are understood. The seasonal cycle of TGS is shown as black solid lines in Fig. 10a (Meinen et al. 2010) and has an amplitude of 3.0 Sv peak to peak with a maximum in July and a minimum in November. After 4 years of measurements the seasonal cycle stands out weakly from the mean monthly standard error of ±1.1 Sv [i.e., the mean monthly standard deviation (std dev) divided by √4; Table 1]; however, both amplitude and phase are consistent with the seasonal cycle computed from the 26-yr-long time series (dashed line in Fig. 10a).
The seasonal cycle of TEK (Fig. 10b, solid line) has an amplitude of 4.1 Sv peak to peak with a maximum in December and a minimum in March (average standard error ±0.8 Sv). However, monthwise averages do not bring out a seasonal periodicity in TEK (Böning et al. 2001), and the seasonal “cycle” derived from the 4-yr record is not representative of the 26-yr-long record (dashed line in Fig. 10b), which exhibits 2.1 Sv peak to peak with a minimum in June and maximum in January. While phase and amplitude of the maximum transport obtained from the long record and the short record are in agreement, the March minimum of the 4-yr record is dominated and biased by unusually strong southward flow in March 2005 (Atkinson et al. 2008).
The TUMO shows a seasonal cycle of 5.9 Sv peak to peak, with a minimum northward transport in April and a maximum one in November (Fig. 10c), clearly significant above the mean monthly standard error of ±1.0 Sv. The seasonal cycle of TUMO is clearly stronger than that of TGS and TEK. The seasonal variability of the vertical profile associated with TUMO is illustrated in Fig. 11, where monthly-mean profiles of the TMO(z) anomaly are shown. The maximum northward flow anomaly occurs in the upper ocean in November and the minimum (relative southward) anomaly occurs in April, consistent with the seasonal cycle of TUMO. Below roughly 1000 m, the pattern is of opposite sign, and the overall variability can therefore be described fundamentally as a first-modelike internal variation of the basinwide, zonally averaged interior flow.
Figure 12 displays the contributions to Ψcycle from the western and eastern boundary fluctuations of the midocean section (as shown in Fig. 9). The western boundary signal
4. Causes of midocean seasonal transport cycle
The variability in TUMO is geostrophic and, therefore, its seasonal cycle is directly related to the difference in seasonal density anomalies between the eastern and western boundaries. Above we found that the eastern boundary density variability dominates over the western boundary (Fig. 12) and that the seasonal signal in TUMO extends to 1000-m depth (Fig. 11). The latter suggests that the seasonal cycle is not fundamentally related to buoyancy forcing at the ocean surface, but is likely a dynamical response to seasonal wind forcing. Next, we consider a simple model of the forced response of the ocean interior to seasonal wind stress curl variations, focusing on the baroclinic response, to try to attribute a mechanism to the observations.
The seasonal wind stress curl anomaly along 26.5°N (Fig. 13a) has a semiannual cycle over most of the basin. The largest signal, however, occurs at the eastern boundary, which is annual in nature with a pronounced anticyclonic curl anomaly in summer and cyclonic anomaly in winter (Figs. 13a,b). It is caused by strong summertime intensification of northerly winds adjacent to the eastern boundary and their relaxation in winter.
Figure 14a shows the resulting model-predicted variation of the midocean transport profile across the basin, 
While largely consistent with the basinwide integrated flow, the simple linear wave model has many limitations. It only allows for purely zonal propagation of northward transport anomalies. It also does not include the effect of horizontal mean flow (and vertical shear) on the anomalies or the impact of topography, as anomalies generated on the eastern boundary move westward (e.g., Killworth and Blundell 2005). It therefore cannot be expected to give an accurate description of the zonal distribution of northward flows in the basin interior along 26.5°N.
The midocean variability predicted by the model is distinct from the quasi-stationary topographic Sverdup response of the ocean interior to the wind stress curl forcing, which depends on the zonally integrated wind stress curl across the entire basin. It is well established from theory and models (Anderson and Gill 1975; Anderson and Corry 1985) and observations (Lee et al. 1996) that, on seasonal time scales, this response is carried primarily in the barotropic mode. This circulation is essentially transparent to our array and has no effect on ΨMAX. The baroclinic response of the interior depends on the density differences between the eastern and western boundaries, which instead are related to first order to the differences in wind stress curl forcing at the eastern and western boundaries, which are the fundamental dynamics expressed in the above model.
5. Discussion
a. Seasonal cycle
From a global perspective, the seasonal anomalies of ΨMAX are thought to be dominated by fluctuations of the Ekman transport, compensated for by a nearly depth-independent geostrophic return flow below the Ekman layer (Jayne and Marotzke 2001; Böning et al. 2001; Wunsch and Heimbach 2009). Jayne and Marotzke point out that the seasonal cycle of the northward Ekman transport and of the meridional overturning circulation are on average symmetric about the equator with nodes at the equator and 20°N/S. We have shown that TGS and TUMO exceed TEK in terms of both amplitude of the seasonal cycle and rms fluctuations on seasonal time scales. The mostly geostrophic seasonal cycle of TUMO of 5.9 Sv peak to peak at 26.5°N is comparable in amplitude with the maximum seasonal cycles of TEK in the North Atlantic, which are found in the tropics (10 Sv) and at midlatitudes (6 Sv). One might therefore speculate that throughout the Atlantic the contribution of geostrophic upper-ocean transports to seasonal anomalies of ΨMAX might be comparable to that of TEK (Hirschi et al. 2007). This is consistent with repeated hydrographic observations at 35°S in the Atlantic (Baringer and Garzoli 2007; Garzoli and Baringer 2007).
Our measurements suggest that the largest part of the seasonal cycle of TUMO is driven by density anomalies at the eastern boundary of the Atlantic. Chidichimo et al. (2010) find coherent seasonal anomalies in density in the depth range between 100 and 1400 m at the mooring sites on the upper continental slope of the eastern boundary, while 1000 km offshore no significant seasonal density anomalies are found at depths in excess of 100 m. It is therefore plausible that the transport anomalies that dominate the seasonal cycle of TUMO do not correspond to basin-scale coherent flows but, rather, are concentrated in a narrow band along the eastern boundary. This concept is consistent with the observed near-eastern-boundary intensification of the seasonal wind stress curl anomalies.
b. Wind stress curl forcing of seasonal anomalies of ΨMAX
The response of TUMO to the seasonal cycle in wind stress curl along 26.5°N has been simulated in a linear “Rossby wave model,” which implies that the seasonal variation of TUMO is almost entirely attributable to changes in stratification at the eastern boundary, caused by local wind stress curl variations. Orography, sea surface temperature gradients, and ocean currents are known to affect wind stress curl (Chelton et al. 2004) as (i) constrictions due to sloping continental orography, island tips, and interisland gaps create jet winds and (ii) differential heating of the marine atmospheric boundary layer across an SST front accelerates wind over warm waters and decelerates it over cold waters. In the annual mean fields there is a narrow band of coherent positive wind stress curl along the eastern margin of the Atlantic from south of Cape Vert near 15°N to Cape Finisterre near 43°N (Fig. 2 of Chelton et al. 2004).
Ongoing studies based on QuikSCAT data suggest that seasonal anomalies in wind stress curl in the tropical/subtropical Atlantic are meridionally coherent (1000-km scale) along the eastern boundary (not shown), but with rather small zonal scales, and may be related to a seasonal pattern with alternating signs in the zonal direction extending westward from the coast to about 19.5°W, resulting from orographic jet winds induced by the Canary islands and Cape Yubi (at 28°N on the Moroccan coast). Hence, we expect the seasonality in TUMO at 26.5°N to have a large meridional coherence scale, modulated locally by jet winds. It has been demonstrated that orography, SST gradient, and ocean current effects on the wind stress curl along continental margins are poorly represented in datasets, such as the National Centers for Environmental Prediction–National Center for Atmospheric Research reanalysis (Chelton et al. 2004), that are routinely used to drive ocean models. Yet our Rossby wave model suggests that these effects may drive the seasonal cycle in the circulation. Kanzow et al. (2009) showed that an eddy-resolving 
c. Is the variability in ψMAX observed between April 2004 and April 2008 representative of longer periods?
There is not, nor has there been, any other AMOC observing system in place for comparing our results. Therefore, the long-term representativeness (particularly of TUMO) can only be assessed indirectly. A large body of literature exists on hydrographic variability on intraseasonal to decadal periods in the North Atlantic (e.g., Roemmich and Wunsch 1985; Joyce and Robbins 1996; Joyce et al. 1999; Johnson and Gruber 2007; Cunningham and Alderson 2007; Kieke et al. 2009). However, there is no straightforward link between changes in hydrographic properties and changes in ΨMAX. For example, the strength of the AMOC-related Labrador Sea outflow along the western boundary appears to have been stable despite a decade-long warming trend in the outflow waters (e.g., Schott et al. 2006). A further complication for the interpretation of historical hydrographic data in terms of ΨMAX is that density measurements away from the ocean boundaries (even few tens of kilometers away) do not provide a strong constraint on AMOC transport variability at 26.5°N (Kanzow et al. 2009) because of eddy noise.
Unfortunately, a similar limitation applies to satellite altimetry data that otherwise could be considered as a promising way to extend our time series back in time. Kanzow et al. (2009) have shown that sea surface height (SSH) differences between the eastern and western boundary cannot be used to infer the temporal variability of TUMO at 26.5°N. They argue that this is primarily due to the more complex vertical structure of the flow close to the ocean boundaries, which inhibits a simple projection of SSH on the first baroclinic mode (in contrast to the offshore ocean). The results are in agreement with simulations based on a numerical model by Hirschi et al. (2009).
Currently, simulations from numerical models are probably the only source for long, daily AMOC time series exceeding our 4-yr measurement period. In their ocean state estimate Wunsch and Heimbach (2009) find a dominant seasonal tropical Ekman transport response, which is in line with the results of Jayne and Marotzke (2001). Whether the state estimate successfully captures the observed seasonal anomaly in TUMO at 26.5°N is unclear. In general, the degree of realism of fluctuations in TUMO in assimilation products will depend (among other things) on purposeful observations that provide strong constraints on the basinwide integrated northward flow. Hydrographic measurements away from the ocean boundaries or SSH do not fall in this category (Kanzow et al. 2009). This view is supported by the findings of Smith et al. (2009), who show that the assimilation of hydrographic data from Argo floats into a numerical model fails to improve the temporal variability of ΨMAX, when compared with the RAPID–MOC/MOCHA time series. They conclude that density measurements across the ocean margins are required to constrain the flow. The scarcity of such observations might also explain why today’s state-of-the-art ocean state estimates (even when carefully constrained by the same observations) remarkably differ from one another in terms of the strength and temporal variability of ΨMAX (Lee 2009).
In this study wind stress curl at the eastern boundary has been identified as a possible driving mechanism of the seasonal cycle of TUMO. Assuming that this relationship is robust, the representativeness of the seasonal cycle in TUMO (derived from the 4-yr measurement period) of longer measurement intervals will be linked to the representativeness of the seasonal cycle of the wind stress curl. The QuikSCAT high-resolution wind measurements started in 1999. From daily gridded wind stress data (horizontal resolution of 0.25° × 0.25°), the monthly mean wind stress curl was computed close to upper-ocean density moorings (EBH4, EBH5). Figure 15 reveals that the seasonal cycle is a rather regular feature at the eastern boundary over the 12-yr interval, in both phase and amplitude. It clearly dominates the variability at this location, as each of the January values in the 1999–2009 interval is larger than each of the July ones. In addition, the seasonal cycle in wind stress curl from the 2004–08 interval (bold dashed line) is almost identical to that from the 1999–2009 interval (bold solid line). The observed seasonal cycle of TUMO may therefore be representative of the last decade and even longer periods.
If the seasonal cycle of TUMO is a long-term persistent feature of the ocean circulation at 26.5°N, it is likely that the inference of decadal trends in ΨMAX based on hydrographic snapshots might suffer from seasonal biases. Bryden et al. (2005b) deduced a decline in ΨMAX of 8 Sv between 1957 and 2004 using the five hydrographic sections shown in Fig. 16 (filled squares and Table 2). They used constant values for TEK and TGS, leaving TUMO as the only time-variable component of ΨMAX. Based on our analysis, the months of the first and last cruises (October and April) correspond to the maximum and minimum in the seasonal cycle of TUMO (Fig. 10c) such that the 1957 and 2004 estimates are likely to be biased high and low, respectively. If we subtract the seasonal anomalies of TUMO (shown in Fig. 10c) from the hydrographic estimates, by taking into account the months in which the cruises were conducted (Table 2), the resulting “de-seasoned” time series of ΨMAX (open diamonds, Fig. 16) exhibits a reduction in variance of more than 80% and does not show a persistent decline. The efficiency of the seasonal bias correction in removing variance implies that aliasing due to seasonal anomalies possibly accounts for a large part of the trend found by Bryden et al. (2005b).
d. What are the meridional scales associated with the seasonal anomalies?
In the climate context, it would be instructive to know what the meridional scales of the seasonal anomalies in ΨMAX (and of the associated meridional heat transport) are. Are the seasonal anomalies a local phenomenon [i.e., associated with an eddy decorrelation scale of O(100 km) or less] or is their meridional extent of O(1000 km)? To answer this question, simultaneous continuous measurements of density along the ocean margins at different latitudes and depth levels would be required. As mentioned above, such observations are very rare, and this represents a major gap in today’s ocean observing system.
A handle on the meridional scales of anomalies in TUMO (or ΨMAX) may indirectly be obtained from numerical models and/or plausibility arguments. Kanzow et al. (2009) concluded from a combination of RAPID–MOC/MOCHA observations, altimetry, and a high-resolution numerical model that the impact of local eddies on TUMO at 26.5°N was rather small. Numerical model results from Hirschi et al. (2007), relying on monthly values, suggest that anomalies of the thermal wind component of ΨMAX (i.e., the Ekman and external component subtracted) at 26.5° display a meridional decorrelation scales of roughly 1000 km.
Numerical models have shown that ΨMAX at low latitudes in the Atlantic (including 26.5°N) is highly correlated with the advective meridional heat transport (e.g., Böning et al. 2001; Kanzow et al. 2008b), and this has been confirmed from an analysis of the RAPID–MOC/MOCHA measurements (Johns et al. 2010, manuscript submitted to J. Climate). Further, it has been shown that the meridional divergence of advective meridional heat transport nearly balances upper-ocean heat storage on seasonal time scales at low latitudes, whereas toward higher latitudes, air–sea heat fluxes are of primary importance (e.g., Jayne and Marotzke 2001). This study suggests that seasonal geostrophic upper-ocean transport fluctuations are stronger than previously thought. Therefore, the possible meridional divergence of these fluctuations might represent an important contribution to low-latitude, seasonal heat storage anomalies.
Johns et al. (2010, manuscript submitted to J. Climate) find that a change in ΨMAX of 1 Sv at 26.5°N corresponds to a change in advective heat transport of 0.06 × 1015 W. A simple calculation shows that meridional divergence in upper-ocean geostrophic flow of 2 Sv between two transatlantic sections separated by 1000 km over the course of 6 months would lead to a net temperature change of 0.2°C in the upper 500 m (if there is no exchange with the atmosphere). Since heat storage will not be spatially uniform, local changes (near the ocean margins) larger than this on seasonal periods are likely. In contrast anomalous TUMO associated with an eddy scale of O(100 km) would correspond to a 2°C anomaly, which is far more than we observe at the various measurement sites. From these considerations we assume that the meridional extent of the seasonal anomalies is likely to be O(1000 km) rather than being set by localized eddy processes.
Possible seasonal storage of heat by large-scale divergences of geostrophic upper-ocean transport may be important for regional oceanic and of near-boundary continental climates, if the heat is (partly) released to the atmosphere. Near-surface seasonal heat storage at low latitudes may represent a nonnegligible source of energy for tropical cyclones. However, given that wind stress curl forcing along coastal margins may be unrealistically small in OGCMs, simultaneous, continuous observations of upper-ocean geostrophic transport across two or more zonal transects would be needed to observe the possible existence of strong upper-ocean meridional geostrophic transport divergence.
6. Conclusions
Between April 2004 and April 2008, the strength of the AMOC, ΨMAX, at 26.5°N has a mean of 18.7 ± 2.1 Sv and rms fluctuations of 4.8 Sv. At periods shorter than 100 days, TEK variability dominates over TGS and TUMO, while at seasonal time scales ΨMAX variability is dominated by TUMO and TGS.
The total western boundary contribution ΨWMAX (i.e., TGS plus the western boundary contribution of the upper-midocean component
The best estimate of the long-term peak-to-peak amplitude of the seasonal cycle of ΨMAX is 6.7 Sv. From the three transport components TUMO has the most pronounced seasonal cycle of 5.9 Sv peak to peak with a maximum northward upper-ocean transport in autumn and a minimum in spring. The TUMO cycle is dominated by the density contribution from the eastern boundary, which has a peak-to-peak amplitude of 5.4 Sv.
The response of TUMO to the seasonal cycle in wind stress curl along 26.5°N has been simulated in a linear “Rossby wave model.” The modeled and observed seasonal cycle of TUMO agree both in phase and amplitude. The model implies that the seasonal variation of TUMO is almost entirely attributable to changes in stratification at the eastern boundary, caused by local wind stress curl variations that uplift (depress) density surfaces in the spring (fall) that follow, in quadrature, the winter (summer) periods of enhanced cyclonic (anticyclonic) curl at the eastern boundary.
Acknowledgments
The authors thank the captains and crews of the R/Vs Charles Darwin, Discovery, Ronald H. Brown, Knorr, Poseidon, and Seward Johnson; and the UKORS, RSMAS, and AOML mooring and hydrography teams. The mooring operations have been supported by the Natural Environment Research Council (NERC) RAPID program, the U.S. National Science Foundation (NSF) under Grant 0728108, and the U.S. National Oceanic and Atmospheric Administration Western Boundary Time Series program. The Florida Current cable data are made freely available by the Atlantic Oceanographic and Meteorological Laboratory (available online at www.aoml.noaa.gov/phod/floridacurrent/) and are funded by the NOAA Office of Climate Observation. The wind stress data were obtained from CERSAT at IFREMER, Plouzané (France). We thank Darren Rayner (NOCS) for coordinating the U.K. seagoing activities and for leading the mooring data recovery. Two of the authors (TK and JJMH) were supported in the framework of the NERC-funded Rapid Climate Change Program (Grant 880 NER/T/S/2002/00481). We thank the two anonymous reviewers for their detailed criticism, which led to significant improvement of the manuscript.
REFERENCES
Anderson, D. L. T., and A. E. Gill, 1975: Spin-up of a stratified ocean with application to upwelling. Deep-Sea Res. Oceanogr. Abstr., 22 , 583–596.
Anderson, D. L. T., and R. A. Corry, 1985: Seasonal transport variations in the Florida Current: A model study. J. Phys. Oceanogr., 15 , 773–786.
Atkinson, C. P., H. L. Bryden, and T. Kanzow, 2008: Ekman transport variability at 26°N in the Atlantic. Proc. Rapid Climate Change 2008 Annual General Meeting, Cambridge, United Kingdom.
Baringer, M. O., and S. L. Garzoli, 2007: Meridional heat transport determined with expendable bathythermographs—Part I: Error estimates from model and hydrographic data. Deep-Sea Res. I, 54 , 1390–1401. doi:10.1016/j.dsr.2007.03.011.
Baringer, M. O., W. E. Johns, C. S. Meinen, D. Shoosmith, and H. L. Bryden, cited. 2008: On the structure of Florida current variability. [Available online at http://www.noc.soton.ac.uk/rapid/sci/viewabs1.php?keyword1=gwa806].
Barnier, B., 1988: A numerical study on the influence of the mid-Atlantic Ridge on nonlinear first-mode baroclinic Rossby waves generated by seasonal winds. J. Phys. Oceanogr., 18 , 417–433.
Beal, L. M., J. M. Hummon, E. Williams, O. B. Brown, W. Baringer, and E. J. Kearns, 2008: Five years of Florida Current structure and transport from the Royal Caribbean cruise ship Explorer of the Seas. J. Geophys. Res., 113 , C06001. doi:10.1029/2007JC004154.
Biastoch, A., C. W. Böning, and J. Getzlaff, 2008: Causes of interannual–decadal variability in the meridional overturning circulation of the midlatitude North Atlantic Ocean. J. Climate, 21 , 6599–6615.
Böning, C. W., C. Dieterich, B. Barnier, and Y. Jia, 2001: Seasonal cycle of meridional heat transport in the subtropical North Atlantic: A model intercomparison in relation to observations near 25°N. Prog. Oceanogr., 48 , 231–253.
Bryden, H. L., W. E. Johns, and P. M. Saunders, 2005a: Deep western boundary current east of Abaco: Mean structure and transport. J. Mar. Res., 63 , 35–57.
Bryden, H. L., H. R. Longworth, and S. A. Cunningham, 2005b: Slowing of the Atlantic meridional overturning circulation at 25°N. Nature, 438 , 655–657.
Bryden, H. L., A. Mujahid, S. A. Cunningham, and T. Kanzow, 2009: Adjustment of the basin-scale circulation at 26°N to variations in Gulf Stream, deep western boundary current and Ekman transports as observed by the RAPID array. Ocean Sci., 5 , 421–433.
Cabanes, C., T. Lee, and L-L. Fu, 2008: Mechanisms of interannual variations of the meridional overturning circulation of the North Atlantic Ocean. J. Phys. Oceanogr., 38 , 467–480.
Chelton, D. B., M. G. Schlax, M. H. Freilich, and R. F. Milliff, 2004: Satellite measurements reveal persistent small-scale features in ocean winds. Science, 303 , 978–983.
Chidichimo, M. P., T. Kanzow, S. A. Cunningham, and J. Marotzke, 2010: The contribution of eastern-boundary density variations to the Atlantic meridional overturning circulation at 26.5°N. Ocean Sci., 6 , 475–490.
Cunningham, S. A., and S. Alderson, 2007: Transatlantic temperature and salinity changes at 24.5°N from 1957 to 2004. Geophys. Res. Lett., 34 , L14606. doi:10.1029/2007GL029821.
Cunningham, S. A., and Coauthors, 2007: Temporal variability of the Atlantic meridional overturning circulation at 26.5°N. Science, 317 , 935–938.
Delworth, T., S. Manabe, and R. J. Stouffer, 1993: Interdecadal variations of the thermohaline circulation in a coupled ocean–atmosphere model. J. Climate, 6 , 1993–2011.
Fennel, W., and H. U. Lass, 2007: On the impact of wind curls on coastal currents. J. Mar. Syst., 68 , 128–142.
Ganachaud, A., 2003a: Error budget of inverse box models: The North Atlantic. J. Atmos. Oceanic Technol., 20 , 1641–1655.
Ganachaud, A., 2003b: Large-scale mass transports, water mass formation, and diffusivities estimated from World Ocean Circulation Experiment (WOCE) hydrographic data. J. Geophys. Res., 108 , 3213. doi:10.1029/2002JC001565.
Garzoli, S. L., and M. O. Baringer, 2007: Meridional heat transport determined with expandable bathythermographs—Part II: South Atlantic transport. Deep-Sea Res. I, 54 , 1402–1420. doi:10.1016/j.dsr.2007.04.013.
Hall, M. M., and H. L. Bryden, 1982: Direct estimates and mechanisms of ocean heat transport. Deep-Sea Res., 29A , 339–359.
Herrmann, P., and W. Krauss, 1989: Generation and propagation of annual Rossby waves in the North Atlantic. J. Phys. Oceanogr., 19 , 727–744.
Hirschi, J. J-M., and J. Marotzke, 2007: Reconstructing the meridional overturning circulation from boundary densities and the zonal wind stress. J. Phys. Oceanogr., 37 , 743–763.
Hirschi, J. J-M., J. Baehr, J. Marotzke, J. Stark, S. Cunningham, and J-O. Beismann, 2003: A monitoring design for the Atlantic meridional overturning circulation. Geophys. Res. Lett., 30 , 1413. doi:10.1029/2002GL016776.
Hirschi, J. J-M., P. D. Killworth, and J. R. Blundell, 2007: Subannual, seasonal, and interannual variability of the North Atlantic meridional overturning circulation. J. Phys. Oceanogr., 37 , 1246–1265.
Hirschi, J. J-M., P. D. Killworth, J. R. Blundell, and D. Cromwell, 2009: Sea surface height signals as indicators for oceanic meridional mass transports. J. Phys. Oceanogr., 39 , 581–601.
Houghton, J. T., Y. Ding, D. J. Griggs, M. Noguer, P. J. van der Linden, X. Dai, K. Maskell, and C. A. Johnson, Eds. 2001: Climate Change 2001: The Scientific Basis. Cambridge University Press, 881 pp.
Jayne, S. R., and J. Marotzke, 2001: The dynamics of ocean heat transport variability. Rev. Geophys., 39 , 385–411.
Johns, W. E., T. Kanzow, and R. Zantopp, 2005: Estimating ocean transports with dynamic height moorings: An application in the Atlantic Deep Western Boundary Current at 26°N. Deep-Sea Res. I, 52 , 1542–1567.
Johns, W. E., L. M. Beal, M. O. Baringer, J. R. Molina, S. A. Cunningham, T. Kanzow, and D. Rayner, 2008: Variability of shallow and deep western boundary currents off the Bahamas during 2004–05: Results from the 26°N RAPID–MOC array. J. Phys. Oceanogr., 38 , 605–623.
Johnson, G. C., and N. Gruber, 2007: Decadal water mass variations along 20°W in the Northeastern Atlantic Ocean. Prog. Oceanogr., 73 , 277–295.
Joyce, T. M., and P. Robbins, 1996: The long-term hydrographic record at Bermuda. J. Climate, 9 , 3121–3131.
Joyce, T. M., R. S. Pickart, and R. C. Millard, 1999: Long-term hydrographic changes at 52° and 66°W in the North Atlantic subtropical gyre and Caribbean. Deep-Sea Res. II, 46 , 245–278.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77 , 437–471.
Kanzow, T., 2000: Integrale Erfassung langperiodischer Transporte: Simulation und Optimierung eines verankerten Systems. M.S. thesis, Department of Mathematics and Natural Sciences, Christian-Albrechts-University, 95 pp.
Kanzow, T., and M. Visbeck, 2009: Ocean currents as indicator of climate change. Climate Change: Observed Impacts on Planet Earth, T. M. Letcher, Ed., Elsevier, 349–366.
Kanzow, T., U. Send, W. Zenk, A. D. Chave, and M. Rhein, 2006: Monitoring the integrated deep meridional flow in the tropical North Atlantic: Long-term performance of a geostrophic array. Deep-Sea Res. I, 53 , 528–546.
Kanzow, T., and Coauthors, 2007: Observed flow compensation associated with the MOC at 26.5°N in the Atlantic. Science, 317 , 938–941.
Kanzow, T., and Coauthors, 2008a: A prototype system for observing the Atlantic meridional overturning circulation: Scientific basis, measurement and risk mitigation strategies, and first results. J. Oper. Oceanogr., 1 , 19–28.
Kanzow, T., U. Send, and M. McCartney, 2008b: On the variability of the deep meridional transports in the tropical North Atlantic. Deep-Sea Res. I, 55 , 1601–1623. doi:10.1016/j.dsr.2008.07.011.
Kanzow, T., H. Johnson, D. Marshall, S. A. Cunningham, J. J-M. Hirschi, A. Mujahid, H. L. Bryden, and W. E. Johns, 2009: Basinwide integrated volume transports in an eddy-filled ocean. J. Phys. Oceanogr., 39 , 3091–3110.
Kieke, D., B. Klein, L. Stramma, M. Rhein, and K. P. Koltermann, 2009: Variability and propagation of Labrador Sea Water in the southern subpolar North Atlantic. Deep-Sea Res. I, 56 , 1656–1674.
Killworth, P. D., and J. R. Blundell, 2005: The dispersion relation for planetary waves in the presence of mean flow and topography. Part II: Two-dimensional examples and global results. J. Phys. Oceanogr., 35 , 2110–2133.
Köhl, A., 2005: Anomalies of meridional overturning: Mechanisms in the North Atlantic. J. Phys. Oceanogr., 35 , 1455–1472.
Larsen, J. C., 1992: Transport and heat flux of the Florida Current at 27°N derived from cross-stream voltages and profiling data: Theory and observations. Philos. Trans. Roy. Soc. London, 338A , 169–236.
Latif, M., and Coauthors, 2004: Reconstructing, monitoring, and predicting multidecadal-scale changes in the North Atlantic thermohaline circulation with sea surface temperature. J. Climate, 17 , 1605–1614.
Lee, T., cited. 2009: Consistency of AMOC estimates from global ocean data assimilation (ODA) products. [Available online at http://www.atlanticmoc.org/AMOC2009agenda.php].
Lee, T. N., W. E. Johns, R. J. Zantopp, and E. R. Fillenbaum, 1996: Moored observations of western boundary current variability and thermocline circulation at 26.5°N in the subtropical North Atlantic. J. Phys. Oceanogr., 26 , 962–983.
Levitus, S., 1982: Climatological Atlas of the World Ocean. NOAA Prof. Paper 13, 173 pp. and 17 microfiche.
Longworth, H. R., 2007: Constraining variability of the Atlantic meridional overturning circulation at 26.5°N from historical observations. Ph.D. thesis, University of Southampton, 198 pp.
Lumpkin, R., and K. Speer, 2007: Global ocean meridional overturning. J. Phys. Oceanogr., 37 , 2550–2562.
McCartney, M. S., and R. A. Curry, 1993: Transequatorial flow of Antarctic bottom water in the western Atlantic Ocean: Abyssal geostrophy at the equator. J. Phys. Oceanogr., 23 , 1264–1276.
Meinen, C. S., M. O. Baringer, and R. F. Garcia, 2010: Florida Current transport variabilty: An analysis of annual and longer-period signals. Deep-Sea Res. I, 57 , 835–846. doi:10.1016/j.dsr.2010.04.001.
Mercier, H., and K. Speer, 1998: Transport of bottom water in the Romanche Fracture Zone and Chain Fracture Zone. J. Phys. Oceanogr., 28 , 779–790.
Orsi, A. H., W. M. Smethie Jr., and J. L. Bullister, 2002: On the total input of Antarctic waters to the deep ocean: A preliminary estimate from chlorofluorocarbon measurements. J. Geophys. Res., 107 , 3122. doi:10.1029/2001JC000976.
Osychny, V., and P. Cornillon, 2004: Properties of Rossby waves in the North Atlantic estimated from satellite data. J. Phys. Oceanogr., 34 , 61–76.
Risien, C. M., and D. B. Chelton, 2008: A global climatology of surface wind and wind stress fields from eight years of QuikSCAT scatterometer data. J. Phys. Oceanogr., 38 , 2379–2413.
Roemmich, D., and C. Wunsch, 1985: Two transatlantic sections: Meridional circulation and heat flux in the subtropical North Atlantic Ocean. Deep-Sea Res., 32 , 619.
Schlax, M. G., D. B. Chelton, and M. H. Freilich, 2001: Sampling errors in wind fields constructed from single and tandem scatterometer datasets. J. Atmos. Oceanic Technol., 18 , 1014–1036.
Schott, F. A., J. Fischer, M. Dengler, and R. Zantopp, 2006: Variability of the deep western boundary current east of the Grand Banks. Geophys. Res. Lett., 33 , L21S07. doi:10.1029/2006GL026563.
Smith, G. C., K. Haines, T. Kanzow, and S. Cunningham, 2009: Impact of hydrographic data assimilation on the Atlantic meridional overturning circulation. Ocean Sci. Discuss., 6 , 2667–2715. doi:10.5194/osd-6-2667-2009.
Sturges, W., B. G. Hong, and A. J. Clarke, 1998: Decadal wind forcing of the North Atlantic subtropical gyre. J. Phys. Oceanogr., 28 , 659–668.
Talley, L. D., J. L. Reid, and P. E. Robbins, 2003: Data-based meridional overturning streamfunctions for the global ocean. J. Climate, 16 , 3213–3226.
Tennekes, H., and J. L. Lumley, 1972: A First Course in Turbulence. The MIT Press, 300 pp.
Wijffels, S. E., 2001: Ocean transport of fresh water. Ocean Circulation and Climate: Observing and Modelling the Global Ocean, G. Siedler, J. Church, and J. Gould, Eds., International Geosphysics Series, Vol. 77, Academic Press, 475–488.
Woodgate, R. A., and K. Aagaard, 2005: Revising the Bering Strait freshwater flux into the Arctic Ocean. Geophys. Res. Lett., 32 , L02602. doi:10.1029/2004/GL021747.
Wunsch, C., and P. Heimbach, 2009: The globally integrated ocean circulation (MOC), 1992–2006: Seasonal and decadal variability. J. Phys. Oceanogr., 39 , 351–368.
APPENDIX A
Computation of Internal Transport TINT
APPENDIX B
Error Bars
Uncertainties in time-mean transports can come from three sources: (i) measurement errors (temperature, conductivity, velocity, wind stress, cable voltage), (ii) model errors (e.g., geostrophic approximation; compensation), and (iii) the time variability of the transport signal (standard errors). The measurement errors consist of two parts: random errors and possible bias errors. A detailed error analysis for a precursor experiment near 26.5°N (Johns et al. 2005; Kanzow et al. 2006) has an error in baroclinic transports (i.e., TINT) of 2.5 Sv rms. The precursor experiment used a lower number of vertical density sampling levels, less precise temperature sensors, and very few pressure measurements and no conductivity measurements. On the basis of this, we estimate the error in instantaneous (i.e., 10-day low-pass filtered) measurements of top-to-bottom integrated TINT(z) is less than 2.0 Sv rms. The contribution of the instantaneous uncertainty in TINT to that of ΨMAX will be less than 1.5 Sv rms as a result of the application mass compensation (reduction of 25%). The errors in TINT arise from both uncertainties in the T, C, and P sensors and vertical interpolation between the discrete measurement levels. As the sensors are replaced and carefully calibrated each year, potential biases in the 4-yr average sensor-related uncertainties are expected to be small. The interpolation-related uncertainty is mostly random, even if the sampling levels do not change over time [see Fig. 18 of Johns et al. (2005)]. Therefore, the uncertainty of the 4-yr mean will be substantially smaller than the instantaneous uncertainty. We therefore expect the remaining 4-yr-mean bias of ΨMAX (and TUMO) resulting from uncertainties in TINT to be not larger than ±1.0 Sv.
Errors in daily mean instantaneous (3-day low-pass filtered) measurements of TGS amount to 1.7 Sv rms with the errors being mostly random (Larsen 1992; Meinen et al. 2010). Here TGS is regularly corrected for potential biases using independent estimates of TGS from calibration cruises (relying on velocity measurements from dropsondes). A conservative estimate of the 4-yr-averaged uncertainty in TGS is ±0.5 Sv. This is based on the fact that six or more cable calibration cruises per year are performed on average, yielding more than 24 independent calibration points over the 4-yr record, each with ±1.7-Sv accuracy, and therefore a mean transport bias of 1.7/√24 = 0.3 Sv. We estimate the possible 4-yr mean bias in TEK is ±0.5 Sv (resulting from uncertainties in both wind measurements and the drag coefficient). We consider this estimate to be rather conservative as (i) it amounts to 15% of the observed time mean of TEK and (ii) rms differences in TEK between instantaneous values from QuikSCAT and the NCEP–NCAR reanalysis amount to 0.5 Sv rms. The instantaneous error in TWBW yields 0.4 Sv rms (including possible mean biases), based on comparisons between lowered acoustic Doppler and the moored current measurements at the western boundary for TWBW. The 4-yr-mean bias of TWBW should be of O(±0.2 Sv).
Instantaneous measurements of TAABW do not exist, and so the uncertainty of both instantaneous and time-mean values is difficult to estimate. The standard deviation of the five snapshot estimates of AABW transport at 26°N (Bryden et al. 2005b) is 0.5 Sv. The true uncertainty instantaneous measurements of TAABW might be somewhat larger because of possible undersampling of the deep transport signal over rough bathymetry (Ganachaud 2003a), say, less than 1.0 Sv rms. The effect of a 1.0-Sv uncertainty in TAABW to that in ΨMAX is less than 0.2 Sv rms. This is because compensation transport TC(z) (see [4]) is essentially barotropic, so that the compensation for the error contribution of TAABW is distributed almost uniformly in the vertical. As ΨMAX is an integral over approximately the upper 1000 m and the average depth of the section is around 5000 m, the errors contributing to ΨMAX amount to only about 20% of the uncertainty in TAABW. Therefore the 4-yr-average bias in ΨMAX from this contribution should be less than 0.2 Sv. If there is a mean 1-Sv net transport across 26.5°N resulting from the inflow into the Arctic through the Bering Strait, this can only show up in our array as a barotropic component (since all vertically sheared flow is accounted for in TINT). If one added this to the observed (mass balanced) profile, and integrated from the surface downward to hZC, it would add an uncertainty to ΨMAX of about 0.2 Sv (or 20% of 1 Sv).
Combining the above error estimates as root-sum-square, we estimate the measurement error for the 4-yr average of ΨMAX is ±1.3 Sv (from rms errors of TINT 1.0 Sv, TGS 0.5 Sv, TEK 0.5 Sv, TWBW 0.2 Sv, TAABW 0.2 Sv, and the Bering Strait imbalance 0.2 Sv).
It is beyond the scope of this paper to discuss the model-related errors. Scaling arguments imply that the errors in using the geostrophic and Ekman approximations for our application are on the order of 3% (Kanzow 2000). The standard error of a time series can be estimated as the standard deviation divided by the square root equivalent degrees of freedom (DOF). To estimate the DOF, we divide the time series length by the integral time scale. We define the integral time scale as the sum of the autocorrelation from minus zero crossing to plus zero crossing (Tennekes and Lumley 1972). The results are summarized in Table 1. The division of the observational period of 1450 days by the integral time scale then gives the DOF. The standard errors of ΨMAX, TGS, TEK, and TUMO amount to 0.8, 0.4, 0.3, and 0.6 Sv, respectively.
Therefore the total uncertainty of the 4-yr mean ΨMAX (representing the sum of the measurement error and standard errors) amounts to 1.3 Sv + 0.8 Sv = 2.1 Sv. In principle the two errors could also be combined randomly, since they have arbitrary signs; however, we choose to add them linearly.
Atlantic meridional overturning streamfunction Ψ(y, z) from observations (from Talley et al. 2003), with a 2-Sv contour interval. The observations reveal two interhemispheric overturning cells, with the deep one involving AABW and the shallower one NADW.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Atlantic meridional overturning streamfunction Ψ(y, z) from observations (from Talley et al. 2003), with a 2-Sv contour interval. The observations reveal two interhemispheric overturning cells, with the deep one involving AABW and the shallower one NADW.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Atlantic meridional overturning streamfunction Ψ(y, z) from observations (from Talley et al. 2003), with a 2-Sv contour interval. The observations reveal two interhemispheric overturning cells, with the deep one involving AABW and the shallower one NADW.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
(top) The distribution of moorings along 26.5°N in the subtropical North Atlantic. (bottom) Section of density (and bottom pressure) moorings along 26.5°N. The current meter moorings west of WB2 are not shown here for clarity (see Fig. 3).
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
(top) The distribution of moorings along 26.5°N in the subtropical North Atlantic. (bottom) Section of density (and bottom pressure) moorings along 26.5°N. The current meter moorings west of WB2 are not shown here for clarity (see Fig. 3).
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
(top) The distribution of moorings along 26.5°N in the subtropical North Atlantic. (bottom) Section of density (and bottom pressure) moorings along 26.5°N. The current meter moorings west of WB2 are not shown here for clarity (see Fig. 3).
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Moorings near the western boundary (off Abaco, the Bahamas): density sensors (crosses), bottom pressure recorders squares), and current meters (circles). The dots at WBA and WB0 indicate the part of the water column covered by ADCP measurements. WBH1 and WBH2 were only deployed during the period from April 2004 to 2005.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Moorings near the western boundary (off Abaco, the Bahamas): density sensors (crosses), bottom pressure recorders squares), and current meters (circles). The dots at WBA and WB0 indicate the part of the water column covered by ADCP measurements. WBH1 and WBH2 were only deployed during the period from April 2004 to 2005.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Moorings near the western boundary (off Abaco, the Bahamas): density sensors (crosses), bottom pressure recorders squares), and current meters (circles). The dots at WBA and WB0 indicate the part of the water column covered by ADCP measurements. WBH1 and WBH2 were only deployed during the period from April 2004 to 2005.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Zonally integrated northward transport (Sv m−1) across 26.5°N (top) shallower than 1000 m and (bottom) deeper than 600 m. The bold solid line represents the April 2004–April 2008 time mean of TAMOC(z); the dashed line is the time mean of TMO(z). The abyssal transport structure (below the gray line) is estimated based on the synthetic approximation to historical estimates from Fig. A1.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Zonally integrated northward transport (Sv m−1) across 26.5°N (top) shallower than 1000 m and (bottom) deeper than 600 m. The bold solid line represents the April 2004–April 2008 time mean of TAMOC(z); the dashed line is the time mean of TMO(z). The abyssal transport structure (below the gray line) is estimated based on the synthetic approximation to historical estimates from Fig. A1.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Zonally integrated northward transport (Sv m−1) across 26.5°N (top) shallower than 1000 m and (bottom) deeper than 600 m. The bold solid line represents the April 2004–April 2008 time mean of TAMOC(z); the dashed line is the time mean of TMO(z). The abyssal transport structure (below the gray line) is estimated based on the synthetic approximation to historical estimates from Fig. A1.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Overturning streamfunction 
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Overturning streamfunction
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Overturning streamfunction
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Midocean transport (Sv m−1) TMO(z) (a) shallower than 1200 m and (b) deeper than 600 m (note that the two panels overlap in the 600–1200-m depth range). The data are 10-day low-pass filtered. Note that the transport scale in (a) is much broader than in (b). The interface depth between the upper and lower branches of the upper (NADW) overturning cell hZC is shown as a white dotted line.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Midocean transport (Sv m−1) TMO(z) (a) shallower than 1200 m and (b) deeper than 600 m (note that the two panels overlap in the 600–1200-m depth range). The data are 10-day low-pass filtered. Note that the transport scale in (a) is much broader than in (b). The interface depth between the upper and lower branches of the upper (NADW) overturning cell hZC is shown as a white dotted line.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Midocean transport (Sv m−1) TMO(z) (a) shallower than 1200 m and (b) deeper than 600 m (note that the two panels overlap in the 600–1200-m depth range). The data are 10-day low-pass filtered. Note that the transport scale in (a) is much broader than in (b). The interface depth between the upper and lower branches of the upper (NADW) overturning cell hZC is shown as a white dotted line.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
The thin lines denote the time series of ΨMAX (red), TGS (blue), TEK (black), and TUMO (magenta) for the period between April 2004 and April 2008. The data have been 10-day low-pass filtered. Also shown is the contribution of the compensation transport to ΨMAX [i.e., TC(z) integrated between the sea surface and the level of no motion]. The bold lines represent the best estimates of the long-term seasonal cycles of each transport component (see section 3c and Fig. 10).
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
The thin lines denote the time series of ΨMAX (red), TGS (blue), TEK (black), and TUMO (magenta) for the period between April 2004 and April 2008. The data have been 10-day low-pass filtered. Also shown is the contribution of the compensation transport to ΨMAX [i.e., TC(z) integrated between the sea surface and the level of no motion]. The bold lines represent the best estimates of the long-term seasonal cycles of each transport component (see section 3c and Fig. 10).
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
The thin lines denote the time series of ΨMAX (red), TGS (blue), TEK (black), and TUMO (magenta) for the period between April 2004 and April 2008. The data have been 10-day low-pass filtered. Also shown is the contribution of the compensation transport to ΨMAX [i.e., TC(z) integrated between the sea surface and the level of no motion]. The bold lines represent the best estimates of the long-term seasonal cycles of each transport component (see section 3c and Fig. 10).
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Solid lines denote power spectra of ΨMAX (red), TGS (blue), TEK (black), and TUMO (magenta) for the period from April 2004 to April 2008 (as shown in Fig. 7). Also shown for reference purposes as dashed lines are the transport spectra of TGS (blue) and TEK (black) based on time series between March 1982 and January 2008. The long TEK time series is based on NCEP–NCAR reanalysis data (Kalnay et al. 1996). The spectra are based on Welch’s periodogram method using a 365 (730)-days-wide Hamming window and 182 (365) days’ overlap between consecutive data segments for periods ≤ (>) 365 days.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Solid lines denote power spectra of ΨMAX (red), TGS (blue), TEK (black), and TUMO (magenta) for the period from April 2004 to April 2008 (as shown in Fig. 7). Also shown for reference purposes as dashed lines are the transport spectra of TGS (blue) and TEK (black) based on time series between March 1982 and January 2008. The long TEK time series is based on NCEP–NCAR reanalysis data (Kalnay et al. 1996). The spectra are based on Welch’s periodogram method using a 365 (730)-days-wide Hamming window and 182 (365) days’ overlap between consecutive data segments for periods ≤ (>) 365 days.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Solid lines denote power spectra of ΨMAX (red), TGS (blue), TEK (black), and TUMO (magenta) for the period from April 2004 to April 2008 (as shown in Fig. 7). Also shown for reference purposes as dashed lines are the transport spectra of TGS (blue) and TEK (black) based on time series between March 1982 and January 2008. The long TEK time series is based on NCEP–NCAR reanalysis data (Kalnay et al. 1996). The spectra are based on Welch’s periodogram method using a 365 (730)-days-wide Hamming window and 182 (365) days’ overlap between consecutive data segments for periods ≤ (>) 365 days.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Ten-day low-pass filtered time series of the western (
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Ten-day low-pass filtered time series of the western (
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Ten-day low-pass filtered time series of the western (
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Seasonal cycles (black solid lines) of (a) TGS, (b) TEK, (c) TUMO, and (d) ΨMAX, as obtained from month-wise averages of the time series between April 2004 and April 2008. The gray envelopes represent the standard error of each month (as obtained from the four realizations of monthly averages that are available for each month). The dashed lines in (a) and (b) represent seasonal cycles of TGS and TEK based on the 26-yr-long time series (October 1982–January 2008) used for the computation of the spectra in Fig. 8. The dashed line in (d) represents the best guess of the long-term seasonal cycle of ΨMAX (see text). Positive values denote northward flow.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Seasonal cycles (black solid lines) of (a) TGS, (b) TEK, (c) TUMO, and (d) ΨMAX, as obtained from month-wise averages of the time series between April 2004 and April 2008. The gray envelopes represent the standard error of each month (as obtained from the four realizations of monthly averages that are available for each month). The dashed lines in (a) and (b) represent seasonal cycles of TGS and TEK based on the 26-yr-long time series (October 1982–January 2008) used for the computation of the spectra in Fig. 8. The dashed line in (d) represents the best guess of the long-term seasonal cycle of ΨMAX (see text). Positive values denote northward flow.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Seasonal cycles (black solid lines) of (a) TGS, (b) TEK, (c) TUMO, and (d) ΨMAX, as obtained from month-wise averages of the time series between April 2004 and April 2008. The gray envelopes represent the standard error of each month (as obtained from the four realizations of monthly averages that are available for each month). The dashed lines in (a) and (b) represent seasonal cycles of TGS and TEK based on the 26-yr-long time series (October 1982–January 2008) used for the computation of the spectra in Fig. 8. The dashed line in (d) represents the best guess of the long-term seasonal cycle of ΨMAX (see text). Positive values denote northward flow.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Monthly-mean midocean transport profile (Sv m−1) for the period April 2004–08, after removal of the annual mean and the barotropic (vertical mean) flow for each month.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Monthly-mean midocean transport profile (Sv m−1) for the period April 2004–08, after removal of the annual mean and the barotropic (vertical mean) flow for each month.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Monthly-mean midocean transport profile (Sv m−1) for the period April 2004–08, after removal of the annual mean and the barotropic (vertical mean) flow for each month.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Seasonal cycles of the (left) western and (right) eastern boundary contributions to the midocean section of the overturning strength 
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Seasonal cycles of the (left) western and (right) eastern boundary contributions to the midocean section of the overturning strength
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Seasonal cycles of the (left) western and (right) eastern boundary contributions to the midocean section of the overturning strength
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
(a) Seasonal wind stress curl anomaly (107 N m−3) along 26.5°N relative to annual mean value based on SCOW climatology (Risien and Chelton 2008). This field is used to force the midocean response analysis illustrated in Fig. 14. (b) Wind stress curl anomaly extracted at 26.5°N, 16.1°W from the data shown in (a). The wind stress curl variability is largest near the eastern boundary, where strong seasonal variations in southward winds along the African coast (maximum in boreal summer) produce a large seasonal cycle.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
(a) Seasonal wind stress curl anomaly (107 N m−3) along 26.5°N relative to annual mean value based on SCOW climatology (Risien and Chelton 2008). This field is used to force the midocean response analysis illustrated in Fig. 14. (b) Wind stress curl anomaly extracted at 26.5°N, 16.1°W from the data shown in (a). The wind stress curl variability is largest near the eastern boundary, where strong seasonal variations in southward winds along the African coast (maximum in boreal summer) produce a large seasonal cycle.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
(a) Seasonal wind stress curl anomaly (107 N m−3) along 26.5°N relative to annual mean value based on SCOW climatology (Risien and Chelton 2008). This field is used to force the midocean response analysis illustrated in Fig. 14. (b) Wind stress curl anomaly extracted at 26.5°N, 16.1°W from the data shown in (a). The wind stress curl variability is largest near the eastern boundary, where strong seasonal variations in southward winds along the African coast (maximum in boreal summer) produce a large seasonal cycle.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
The midocean response from the forced Rossby wave model [Eqs. (8)–(10)], using the SCOW seasonal wind stress curl anomaly climatology (Fig. 13), summed over the first two baroclinic modes. (a) The resulting anomaly of 
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
The midocean response from the forced Rossby wave model [Eqs. (8)–(10)], using the SCOW seasonal wind stress curl anomaly climatology (Fig. 13), summed over the first two baroclinic modes. (a) The resulting anomaly of
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
The midocean response from the forced Rossby wave model [Eqs. (8)–(10)], using the SCOW seasonal wind stress curl anomaly climatology (Fig. 13), summed over the first two baroclinic modes. (a) The resulting anomaly of
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Monthly anomalies of the wind stress curl (107 N m−3) at 27.25°N, 14.50°W near the eastern boundary density moorings EBH4 and EBH5. Each thin line represents one year of monthly averaged data between 1999 and 2009. The bold lines represent seasonal cycles [2004–08 average (dashed) and 1999–2009 average (solid)]. Data source is the 0.25° × 0.25° gridded QuikSCAT scatterometer wind stress from the Jet Propulsion Laboratory (available online at http://podaac.jpl.nasa.gov/DATA_CATALOG/quikscatinfo.html).
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Monthly anomalies of the wind stress curl (107 N m−3) at 27.25°N, 14.50°W near the eastern boundary density moorings EBH4 and EBH5. Each thin line represents one year of monthly averaged data between 1999 and 2009. The bold lines represent seasonal cycles [2004–08 average (dashed) and 1999–2009 average (solid)]. Data source is the 0.25° × 0.25° gridded QuikSCAT scatterometer wind stress from the Jet Propulsion Laboratory (available online at http://podaac.jpl.nasa.gov/DATA_CATALOG/quikscatinfo.html).
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Monthly anomalies of the wind stress curl (107 N m−3) at 27.25°N, 14.50°W near the eastern boundary density moorings EBH4 and EBH5. Each thin line represents one year of monthly averaged data between 1999 and 2009. The bold lines represent seasonal cycles [2004–08 average (dashed) and 1999–2009 average (solid)]. Data source is the 0.25° × 0.25° gridded QuikSCAT scatterometer wind stress from the Jet Propulsion Laboratory (available online at http://podaac.jpl.nasa.gov/DATA_CATALOG/quikscatinfo.html).
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
The ΨMAX inferred from five hydrographic snapshot estimates between 1957 and 2004 (solid diamonds), as reproduced from Bryden et al. (2005b). The hydrography cruises were carried out in different seasons, namely, in October 1957, August–September 1982, July–August 1991, February 1998, and April 2004. The open squares represent the historical estimates of ΨMAX with seasonal anomalies of TUMO (Fig. 10c; Table 2) subtracted.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
The ΨMAX inferred from five hydrographic snapshot estimates between 1957 and 2004 (solid diamonds), as reproduced from Bryden et al. (2005b). The hydrography cruises were carried out in different seasons, namely, in October 1957, August–September 1982, July–August 1991, February 1998, and April 2004. The open squares represent the historical estimates of ΨMAX with seasonal anomalies of TUMO (Fig. 10c; Table 2) subtracted.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
The ΨMAX inferred from five hydrographic snapshot estimates between 1957 and 2004 (solid diamonds), as reproduced from Bryden et al. (2005b). The hydrography cruises were carried out in different seasons, namely, in October 1957, August–September 1982, July–August 1991, February 1998, and April 2004. The open squares represent the historical estimates of ΨMAX with seasonal anomalies of TUMO (Fig. 10c; Table 2) subtracted.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Abyssal, zonally integrated transport across 26.5°N from hydrographic cruises in 1957, 1981, 1992, 1998, and 2004 (as presented by Bryden et al. 2005b). The synthetic approximation of the transports below 5000 m (bold black line) represents the transport shear profile used in this study to extend the AMOC transport profile TAMOC(z) into the AABW range.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Abyssal, zonally integrated transport across 26.5°N from hydrographic cruises in 1957, 1981, 1992, 1998, and 2004 (as presented by Bryden et al. 2005b). The synthetic approximation of the transports below 5000 m (bold black line) represents the transport shear profile used in this study to extend the AMOC transport profile TAMOC(z) into the AABW range.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Abyssal, zonally integrated transport across 26.5°N from hydrographic cruises in 1957, 1981, 1992, 1998, and 2004 (as presented by Bryden et al. 2005b). The synthetic approximation of the transports below 5000 m (bold black line) represents the transport shear profile used in this study to extend the AMOC transport profile TAMOC(z) into the AABW range.
Citation: Journal of Climate 23, 21; 10.1175/2010JCLI3389.1
Basic statistics of the different transport components (Sv) discussed in this study for the period between April 2004 and April 2008 (the values in brackets in columns 2 and 4 refer to the period between October 1982 and January 2008) based on 10-day low-pass filtered data. The standard error (SE) in column 4 represents the average of the 12 monthly standard errors. The integral time scale of the 4-yr-long time series in column 4 has been computed according to appendix B. The asterisk indicates calculations based on (7).
Seasonal bias correction of the Bryden et al. (2005b) historical estimates of ΨMAX (see Fig. 16). Corrections have only been applied to the upper midocean transport (TUMO), as Bryden et al. used constant values for TEK and TGS (see text). Columns 1–4 give the cruise dates, the historical estimates of ΨMAX (solid line in Fig. 16, Bryden et. al), the seasonal anomalies of TUMO (from Fig. 10c) corresponding to the months in which the measurements cruises were conducted, and the seasonal-anomaly-corrected estimates of ΨMAX (dashed line in Fig. 16), respectively.
