From 2004 to 2014, the Line W moorings measured a 0.7 Sv yr⁻¹ slowing of the deep western boundary current (DWBC) offshore of Cape Cod. Here, we combine these deep mooring observations with float and satellite altimeter data and find that this DWBC change corresponded to a slowing of the cross-basin Atlantic Meridional Overturning Circulation (AMOC) of about 0.3 Sv yr⁻¹. Our AMOC transport time series corresponds well with the Estimating the Circulation and Climate of the Ocean state estimate, particularly when the Line W mooring data influences our volume closure. We compare our 35°N time series with a similar time series at 41°N as well as the 26°N RAPID AMOC, and find AMOC declines across datasets from 2004 to 2014. However, when we extend our analysis to 2004–2019, there are no significant trends at any latitude. These observations suggest that AMOC decadal variability is meridionally coherent from 26°N to 41°N and that the DWBC may reflect this variability.

Key Points

We compile an Atlantic Meridional Overturning Circulation (AMOC) time series at 35°N from deep moorings, floats, and altimeter that agrees with the Estimating the Circulation and Climate of the Ocean state estimate
The 2004 to 2014 slowing of the deep western boundary current corresponded to an AMOC decline at 35°N
We find no evidence of long-term AMOC decline, but consistent decadal variability across 26°N, 35°N, and 41°N

Plain Language Summary

The Atlantic ocean hosts an overturning circulation that is thought to be an important piece of our climate system. This circulation pattern spans the width of the basin, making it difficult and costly to measure, so direct observations of the overturning circulation are scarce. In this study we combine existing mooring, float, and satellite altimeter observations to estimate the overturning circulation at a new latitude (35°N), and compare it to existing estimates at 26°N and 41°N as well as the ECCO ocean state estimate. We find that the long term (about 10 year) AMOC variability is consistent across latitudes and data products. While we cannot rule out a decreasing AMOC trend during the 20th century, we find that natural variability is too large to detect a net AMOC decrease in direct observations since 2004.

1 Introduction

The Atlantic Meridional Overturning Circulation (AMOC) is a vast system of ocean currents that redistributes heat, freshwater, and carbon across the Atlantic. The AMOC is expected to weaken as the climate warms and freshwater melts into the North Atlantic, which would in turn have consequences for global climate (IPCC et al., 2019). Caesar et al. (2021) estimated that there has been a 3 Sv decline in the AMOC since the mid-1900s (≈0.04 Sv yr⁻¹), however, this is under significant debate in the literature (Fraser & Cunningham, 2021; Kilbourne et al., 2022; Piecuch, 2020; Rossby et al., 2020; Thornalley et al., 2018). Centennial scale studies are necessarily based on numerical models and proxies as there are very few direct AMOC observations and they span relatively short time periods.

The longest running AMOC observing system is the RAPID-MOCHA (Rapid Climate Change-Meridional Overturning Circulation and Heatflux Array) program (hereafter, RAPID), which has measured the overturning strength at 26°N since 2004 (McCarthy et al., 2015). The AMOC at 26°N weakened from 2004 to 2017, but has subsequently recovered, potentially due to increased convection in the subpolar North Atlantic (Frajka-Williams et al., 2021; Moat et al., 2020; Smeed et al., 2018). The RAPID record is regularly used to ground-truth models, but it is unclear that the AMOC at this latitude is representative of the overturning circulation throughout the Atlantic (Frajka-Williams et al., 2019).

The Line W moorings, named for Val Worthington, measured one component of the AMOC, the deep western boundary current (DWBC), off the coast of Cape Cod at about 40°N from 2004 to 2014 (Joyce et al., 2005; Pena-Molino et al., 2011; Toole et al., 2011). During this period, the Line W program documented a decrease in DWBC transport of −0.74 ± 0.20 Sv yr⁻¹ (Toole et al., 2017). This decline, as well as the evolution of Labrador Sea Water properties at Line W, was consistent with changes in Labrador Sea Water production 5–7 years prior (Le Bras et al., 2017). However, as Line W focused on the western boundary, it is unclear whether the DWBC slowdown corresponded to an overall AMOC weakening.

Here we combine Line W mooring observations with satellite altimeter and Argo float data at 35°N to construct an across-basin data set. The altimeter and float data are combined following the Willis and Fu (2008) approach, which was previously used to construct an AMOC time series and heat transport at 41°N (Hobbs & Willis, 2012; Willis, 2010). Our reconstruction shows a decline of the AMOC associated with the DWBC slowing from 2004 to 2014. We find that our AMOC reconstruction agrees well with the ECCO state estimate and we assess the sensitivity of our calculations in the context of this comparison. We compare our 35°N AMOC time series with the 26°N RAPID AMOC and an updated 41°N AMOC and suggest that they exhibit consistent decadal variability. We do not find clear evidence of a long term AMOC decline.

2 Data

2.1 Line W Moorings

The Line W moorings were deployed on the continental slope southeast of Cape Cod from May 2004 to April 2014 (Toole et al., 2017) (Figure 1b). We focus on the inshore-most five moorings as the sixth mooring was not added until 2008. Our analysis is based on daily profiles of temperature, salinity, and velocity. All data were filtered to remove near-inertial and super-inertial period signals, which is appropriate given our focus on time scales longer than 3 months. The reader is referred to Toole et al. (2017) for the details of the instrumentation, filtering, interpolation, and gap-filling techniques.

Details are in the caption following the image — **Figure 1**
Open in figure viewer PowerPoint

(a) Time mean (2004–2014) ocean speed at 1,000 m depth deduced from Argo float trajectories. The 35 N Atlantic Meridional Overturning Circulation (AMOC) data product section is shown in black, with the Line W mooring location highlighted in green. Dotted lines indicate the locations of 41°N and 26°N AMOC data products. The 3,000 m isobath is shown in gray. (b) Time mean (2004–2014) across-section velocity measured by the Line W moorings. Mooring positions are indicated by vertical lines. The colorbar is asymmetric to emphasize the southward deep western boundary current flow. Black circles indicate deep areas filled by extrapolation. (c) Volume transport across the Line W mooring section between 500 and 2,000 m depth (orange box in panel (b)). Thin green lines represent daily estimates from the Line W mooring data. Thick green lines show Line W transport filtered to match the effective 3-month resolution of the float and altimeter synthesis product transport shown in purple.

2.2 Float and Altimeter Synthesis Product

Satellite observations of sea surface height (Ducet et al., 2000) were combined with temperature and salinity profiles, as well as subsurface displacements observed by Argo floats (Argo, 2022). These observations were combined as described in Willis and Fu (2008) to produce gridded estimates of temperature, salinity, density, and absolute dynamic height at 1,000 m for each month from January 2002 to December 2021. Although centered on a sliding monthly time step, each monthly estimate was computed using altimetry and Argo observations from a 3-month window. So, each grid point represents a 3-month centered moving average. Previous validations of this product as well as updated data coverage are detailed in the Supporting Information S1 (Text S1 and Figures S1–S3).

2.3 Wind Stress Data

The Cross-Calibrated Multi-Platform wind vector analysis (CCMP) is a combination of satellite and in situ wind observations using the variational analysis method (Wentz et al., 2015). The resulting 10 m wind velocity product is 6-hourly, and provided on a 0.25° grid. To convert wind speed to wind stress we follow McCarthy et al. (2015) and Smith (1980): τ_along = q_ac_du_along|u|, where τ_along is the wind stress along the section, u_along is the wind speed at 10 m along the section |u| is the magnitude of the full wind velocity at 10 m, q_a = 1.25 kg m⁻³ is the density of air, and c_d is the drag coefficient. We use a drag coefficient of 10⁻³ for wind speeds lower than 7.5 m s⁻¹ and (0.61 + 0.063|u|) × 10⁻³ for higher wind speeds.

2.4 ECCO Ocean and Sea-Ice State Estimate

Estimating the Circulation and Climate of the Ocean (ECCO) Version 4 Release 4 (V4r4) is a free-running solution to a global 1°, 50 vertical level configuration of the Massachusetts Institute of Technology general circulation model (MITgcm) (Marshall et al., 1997) whose time-trajectory has been fit to ocean and sea-ice observations using least-squares (ECCO Consortium et al., 2021). Data constraints in ECCO V4r4 include satellite observations of sea surface height (from altimeters), ocean bottom pressure (from the Gravity Recovery and Climate Experiment), sea surface temperature and salinity, and sea-ice concentration. In situ hydrography data constraints include temperature and salinity profiles from Argo profiling floats, World Ocean Circulation Experiment and Global Ocean Ship-Based Hydrographic Investigations Program transects, instrumented pinnipeds, Arctic ice-tethered profilers, and instrumented moorings. The Line W shipboard hydrographic data are included in the ECCO state estimate, but not the Line W mooring data. Gridded ECCO state estimate fields include daily and monthly-mean potential temperature, salinity, density, pressure, velocity, sea-surface height and 3D fluxes of volume, temperature, salinity, and momentum. The 1992–2018 estimation period of ECCO V4r4 spans the entire Line W mooring program.

3 Methods

3.1 AMOC Data Set Construction

The starting point for our AMOC data product is the float and altimeter synthesis product. We extracted geostrophic velocity and σ₂ along the extended Line W mooring line from 40.3°N, 70.2°W to 35°N, 66.4°W and then eastward along 35°N (Figure 1a). The section begins at the origin used for previous analyses of the Line W data, which is situated on the 86 m isobath (Toole et al., 2011). The latitude 35°N was selected because it ensured the most direct crossing of the Gulf Stream extension and minimized eddy noise while staying close to the Line W moorings' latitude.

The Line W moorings were placed along a satellite altimeter track and the available Line W upward-looking acoustic Doppler current profiler velocity measurements show high coherence with satellite derived geostrophic velocities on 20–30 days time scales (Toole et al., 2017). The volume transport across the Line W section estimated from the mooring data corresponds well with that from the float and altimeter product (Figure 1c, r = 0.90), further supporting that it is sensible to stitch the mooring data and float and altimeter data product together.

The Line W mooring data was low-pass filtered using a second order Butterworth filter with a 3 month cutoff to match the resolution of the float and altimeter synthesis product. Below 2,000 m from 86 to 279 km along the section, our AMOC data product is comprised of the filtered mooring data (Figure 1b). Between 1,000 and 2000 m depth, the product is a linear combination of the filtered mooring data and the float and altimeter synthesis product such that it is entirely the float and altimeter synthesis product at 1,000 m, entirely the filtered Line W data at 2,000 m and their average at 1,500 m.

The Ekman transport was calculated using ∫τ_along/ρ₀f dx, where τ_along is the wind stress along the section, ρ₀ = 1,030 kg m⁻³ is a reference density, f is the Coriolis parameter, and we integrate along the section. As done in the RAPID data set, we use the CCMP wind stress, which is calibrated against in situ data, and distribute it evenly over the top 100 m (McCarthy et al., 2015). We compare CCMP with other wind stress data products in the Supporting Information S1 (Text S2 and Figure S4).

Closing the volume budget and obtaining an AMOC streamfunction requires filling in the portion of our section east of Line W below 2,000 m. We estimate the total flow across the entire section to be −1 Sv; this is the time-mean flow across this section in ECCO during this time period. We interpolate all velocities from their value at 2,000 m to zero at the seafloor and apply a uniform compensation velocity over the entire section such that the total flow across the section is −1 Sv at all times. The streamfunctions and time series presented in Figures 2 and 3 were constructed using this method for volume closure, but we investigate another method for volume closure in which we limit the compensation velocity to the unobserved regions, that is, the ocean below 2,000 m east of Line W (Figures S5 and S6 in Supporting Information S1). To assess the impact of the Line W mooring data on our analysis, we also calculate the depth-space AMOC in the float and altimeter synthesis product without incorporating the mooring data. We use the same Ekman transport and investigate the same two methods for the volume closure. The extended time series is limited by the availability of CCMP wind data; they extend through April 2019.

We construct the vertical streamfunction in depth space by integrating the across-section velocity along the section and subtracting the net flow across the section. Subtracting the net flow across the section is standard practice in calculating overturning strength at RAPID and Overturning in the Subpolar North Atlantic Program (Lozier et al., 2019; McCarthy et al., 2015). Constructing a streamfunction in density space requires knowledge of the density structure in the unmeasured portion of our section east of Line W below 2,000 m. Rather than use an estimated mean structure, we assign all unmeasured regions σ₂ = 37.5 kg m⁻³, which is greater than the density in the measured regions of the section. In this way, all compensation transport is relegated to the densest bin and can be identified in Figure 2c. Though we discuss the differences between the depth-space and density-space AMOC at 35°N, we focus primarily on the depth-space AMOC as the deep density structure is uncertain and other AMOC estimates are in depth coordinates.

3.2 Timeseries Comparison and Analysis

Our composite data product and the ECCO state estimate are based on complimentary approaches for combining available ocean observations. Both products are based largely on float and altimeter data as these provide the most coverage. Because the ECCO model is dynamically consistent, we use it to select our volume closure method (Figure S6 in Supporting Information S1). Our composite data product has the advantage of connecting directly with detailed western boundary current observations that are necessarily represented differently in ECCO due to its resolution. We extract data from ECCO along the same section described above using the ecco_v4_py Python library. As in the composite data product, we remove the net flow across the section from ECCO to calculate the overturning strength. The net flow across the section is time-variable in ECCO.

In order to investigate AMOC connectivity across the North Atlantic, we also compare to the time series constructed by Willis (2010) at 41°N using the same methodology we use for the interior of our data product (Section 2.2). Finally, we compare with the RAPID AMOC data set at 26°N (Frajka-Williams et al., 2021), which is constructed from cable measurements in the Florida Straits, direct velocity moorings and dynamic height moorings across the basin (McCarthy et al., 2015). To match the effective resolution of our data product, we take the 3 month rolling average of the RAPID time series. All trends are reported with 95% confidence intervals and only correlations that are statistically significant at the 95% confidence level are reported in the main text. Detailed statistical methods as well as additional trends and correlations are presented in the Supporting Information S1 (Text S3 and Tables S1–S5).

4 Results

The vertical streamfunction in depth space at 35°N reflects net northward flow above about 1,000 m and net southward flow below (Figure 2b). The streamfunction in density space, however, has two maxima (Figure 2c). This is because the shallow southward return flow primarily carries subtropical mode water, while the deep-reaching Gulf Stream extension carries waters northward that are both lighter and denser (Figure 2a). The AMOC is stronger in density space than in depth space because of the isopcynal tilt associated with the Gulf Stream which separates southward flow to the west and the Gulf Stream extension to the east (Figure 2a). AMOC variability is very similar in depth and density space and both decrease over the course of the record (Figure 2d). The decreasing AMOC trend is barely significantly different from zero at the 95% confidence level in depth space (−0.28 ± 0.25 Sv yr⁻¹) and not significantly different from zero in density space (−0.25 ± 0.27 Sv yr⁻¹).

Comparing with the AMOC at different latitudes we find that the 35°N AMOC has a correlation of 0.52 (0.39) with the 26°N AMOC with a 2 (1) month lag for 2004–2014 (2004–2019), with RAPID leading (Figure 3). This is explained by their Ekman components, which have a correlation of 0.60 (0.52) with a 3 month lag. The time series are not correlated for either time period when the Ekman transport is subtracted from each. The 35°N and 41°N AMOC time series are similarly correlated because of their Ekman components: 0.50 for the full AMOC and 0.45 for the Ekman component, with no lags. However they are not significantly correlated for 2004–2019. All correlations and their p-values can be found in Table S3 in Supporting Information S1. All datasets have a dip around January 2010, which has been associated with anomalous wintertime winds (McCarthy et al., 2012). The AMOC is decreasing from 2004 to 2014 in all three datasets, though it is not quite statistically significant at 41°N (Figure 3). These trends are not consistently driven by the Ekman transport (Table S2 in Supporting Information S1). There is not a significant AMOC trend at any latitude from 2004 to 2019, contrary to the expectation from the paleo-oceanographic and modeling literature of a long-term decline.

The depth-space AMOC in the ECCO state estimate corresponds well with the depth-space AMOC in our data product (Figure S6a in Supporting Information S1); the correlation coefficient between the time series is 0.52 and their trends are similar (the trend in ECCO is −0.40 ± 0.32 Sv yr⁻¹). Interestingly, if the mooring data are not included in our calculation, the AMOC trend from 2004 to 2014 decreases to −0.15 ± 0.23 Sv yr⁻¹, which does not agree with ECCO as well (Figure S6a and Table S2 in Supporting Information S1). The mean AMOC also decreases from 14.7 to 13.1 Sv when the mooring data are not included. These discrepancies highlight how the mooring data impact our composite data set through the barotropic velocity compensation we apply for volume closure; the strong southward flow at Line W requires compensation by a northward flow, which gets evenly distributed in the water column and increases the AMOC strength. The decreasing transport trend present in the mooring data gets projected onto the shallow northward flow in a similar manner.

To further test the dependence of our results on the volume closure method, we tried assuming that all transport required for volume closure was located in the unobserved portion of the section, that is, below 2,000 m and east of Line W (Figure S6b in Supporting Information S1). Using this closure method the trend decreases to −0.21 ± 0.29 Sv yr⁻¹, which is weaker than but broadly consistent with both ECCO and our time series using a barotropic velocity compensation. The correlation with ECCO decreases to r = 0.45 and the root mean squared difference with the AMOC ECCO increases from about 800 to 1,000 Sv, that is, there is generally worse correspondence with ECCO using this method. As the deep flow does not influence the shallow flow in this calculation and there is good agreement between Line W and the float and altimeter data set above 2,000 m (Figure 1c), it makes little difference whether the mooring data are included in the calculation using this method (Figure S6b in Supporting Information S1). The correlations between all volume closure methods and the ECCO state estimate are listed in Table S4 in Supporting Information S1.

Though the AMOC in ECCO corresponds well with that from our composite data product, the structure of their deep flows differs. We find a time-mean southward transport of about 14 Sv below 1,000 m in the Line W region (Figure S6c in Supporting Information S1). The transport below 1,000 m in ECCO is not as western-intensified and its time-mean does not reach 14 Sv until east of the Mid-Atlantic Ridge (Figure S6c in Supporting Information S1). The Line W and ECCO time-mean velocity fields are shown in Figure 1b and Figure S5 in Supporting Information S1 respectively for a more detailed comparison of the flow structure. Furthermore, we find that the trend below 1,000 m in our composite data product, 0.3 ± 0.6 Sv yr⁻¹, is as large as the overall AMOC trend, while in ECCO the trend is split relatively evenly across the basin, with a trend of 0.14 ± 0.49 Sv yr⁻¹ in the Line W region (positive trends reflect a weakening of southward flow). We find a weaker trend than the Toole et al. (2017) Line W analysis (0.74 ± 0.20 Sv yr⁻¹) because of differences in how we define the DWBC. Toole et al. (2017) integrate the transport of each water mass eastward from the continental slope until the maximum transport is reached. There is likely some offshore compensation by the deep Gulf Stream that we include by integrating over a fixed depth range and eastward extent in this study. These results do not agree with Koelling et al. (2020), who use gravimetry data to suggest that the DWBC decrease was compensated for by an increase in interior flows over a simiar time period. In contrast, both ECCO and our composite data product suggest an overall slowing of the southward deep flow from 2004 to 2014.

5 Discussion and Conclusions

Subtropical AMOC changes are thought to be preceded by water mass transformations at high-latitude. One suggested mechanism is the advection of Labrador Sea Water anomalies. Recent modeling work suggests that this occurs on advective time scales, with Jackson et al. (2016) finding that Labrador Sea density anomalies precede AMOC changes at 26°N by about 10 years, and Desbruyères et al. (2019) finding a 5 year lag between water mass transformations and the AMOC at 45°N. From observations, Van Sebille et al. (2011) find a 10 year transit time to 26°N and Le Bras et al. (2017) found a 5–7 year transit time for Labrador Sea water changes to reach Line W at 39°N via the DWBC. These time scales were recently confirmed by Chomiak et al. (2022).

As our focus is from 2004 to 2014, the most relevant period of strong water mass transformations is that of the early to mid 1990s, when there was intense production of Labrador Sea Water (Yashayaev, 2007). We would expect that there would be a corresponding peak in the AMOC at 35°N and 41°N in the mid 1990s to early 2000s and a later peak at 26°N in the mid-2000s. This is broadly consistent with our observation that the AMOC trends from 2004 to 2014 at 26°N are larger than those at 35°N and 41°N. Examining the full 2004–2019 time period, we find that the AMOC trends decrease are no longer statistically significant at any latitude. This is generally consistent with the arrival of new Labrador Sea Water produced starting in 2014 (Yashayaev & Loder, 2016a, 2016b) countering the previous downward trend, but more work is needed to elucidate this further. In particular, it will be difficult to confirm Labrador Sea Water changes along the western boundary without direct deep observations such as Line W (Le Bras et al., 2017).

Caesar et al. (2021) claim that there has been a 3 Sv decline in the AMOC since the mid-1900s (≈0.04 Sv yr⁻¹), which is almost 10 times weaker than the decadal trend we identify from 2004 to 2014. Kilbourne et al. (2022) argue that the AMOC is too poorly constrained from proxy data and it is unclear whether the expected decline has started. Lobelle et al. (2020) suggest that about 40 years of RAPID AMOC measurements may be needed before a secular decline will be detectable. Our analysis confirms that there are not enough data to identify whether a longer term AMOC decline has begun; if present it is currently in the noise of our direct observations.

There are many sources of error in our AMOC calculation, all of which are ultimately linked to data sparsity. At 35°N the AMOC is the difference between large numbers: the Gulf Stream extension transport and its recirculations. As an illustration of this, though their variability agrees well, the transport at the western boundary measured by the Line W moorings and the float and altimeter synthesis product often differ by several Sverdrups (Figure 1c). Hence the true uncertainty in the long-term trends we report are larger than the statistical error (Text S1 in Supporting Information S1). Additionally, the volume transport compensation has been shown to be a significant source of error in all AMOC datasets, including RAPID (Sinha et al., 2018). We find that our AMOC calculations are sensitive to the volume closure method we use. Interestingly, we find the best agreement with ECCO when the Line W observations impact the overall AMOC through a barotropic velocity compensation (Figure S6 and Table S2 in Supporting Information S1). It is unclear whether our AMOC data set at 35°N can be used for AMOC monitoring post 2014 without the Line W moorings. Along the same lines, its possible the AMOC time series at 41°N requires a barotropic velocity compensation that is missed by the altimeter and Argo float product. At the same time, the flow is much less noisy at 41°N and this has been validated extensively (Text S1 in Supporting Information S1). We do not find that any one of the observation-based datasets we consider (26°N, 35°N, 41°N) agrees better with the AMOC in the ECCO state estimate than any other (Table S5 in Supporting Information S1).

Despite the uncertainties, we are reassured of the utility of this exercise by the correspondence of our AMOC timeseries at 35°N with the ECCO state estimate, which is a dynamically consistent synthesis of observations. Additionally, the AMOC trends we find from 2004 to 2014 and 2004–2019 are similar to those at 26°N and 41°N, consistent with our expectations for volume conservation across the subtropical North Atlantic. The AMOC time series are also coherent at seasonal time scales, but this coherence is due to the Ekman transport component and can be explained by large-scale wind patterns. Interestingly, the Ekman components of the 35°N and 41°N time series are well correlated on seasonal time scales from 2004 to 2014, but not when the time series is extended to 2019, potentially indicating shifts in the wind patterns in this region. Our observations generally support the idea that AMOC decadal variability has some meridional coherence, a topic which continues to be debated in the literature (Bingham et al., 2007; Lozier, 2010; Zou et al., 2019). Additionally, the broad consistency of our findings with hypotheses of Labrador Sea Water advection generated by previous modeling and observational studies is intriguing (Chomiak et al., 2022; Desbruyères et al., 2019; Jackson et al., 2016; Le Bras et al., 2017). Our study connects a decline in the subtropical DWBC with an AMOC decline between 2004 and 2014, suggesting that some decadal variability may be communicated southward along the deep western boundary.

Recent studies have emphasized the role of deep southward interior pathways in the North Atlantic relative to the DWBC, largely inspired by Lagrangian float observations (Bower et al., 2009). However beyond these observations, this work has relied on ocean models and reanalysis products (Gary et al., 2011; Lozier et al., 2013; Zou et al., 2019). Most recently, Zhai et al. (2021) found that about 40% of the southward North Atlantic Deep Water flow in ECCO is east of the Mid-Atlantic Ridge between 35°N and 50°N. Though their analysis is in density space and ours is in depth space, we find that a similar significant portion of the deep flow (≈30%) is east of the Mid-Atlantic Ridge in ECCO. The Line W moorings measured a strong flow focused narrowly at the western boundary, while in ECCO the flow is weaker and more diffuse across the western basin. This analysis suggests caution in interpreting modeled deep flows and highlights the continuing need for observations of deep western boundary flows in order to better understand inter-decadal AMOC changes.

Acknowledgments

ILB and JW gratefully acknowledge the National Aeronautics and Space Administration Grant 80NSSC20K0421. This work was done in part at the Jet Propulsion Laboratory, California Institute of Technology under a contract from NASA. The Argo float data were collected and made freely available by the International Argo Program and the national programs that contribute to it (https://argo.ucsd.edu,https://www.ocean-ops.org). The Argo Program is part of the Global Ocean Observing System. ECCO is supported by NASA's Physical Oceanography, Modeling Analysis and Prediction, and Cryosphere programs. We thank John Toole, Magdalena Andres, and the many other scientists and mariners who went to sea to collect the in situ observational data, particularly through the Line W program.

Open Research

Data Availability Statement

Our 35°N composite AMOC data set can be accessed at https://zenodo.org/record/7262142 and cited using https://doi.org/10.5281/zenodo.7262142. The code for our analysis is available at https://github.com/ilebras/35N-AMOC-code/. The Line W mooring data are accessible on the project website: https://scienceweb.whoi.edu/linew/download_data.php. Data from the RAPID AMOC monitoring project is funded by the Natural Environment Research Council and are freely available from www.rapid.ac.uk/rapidmoc. Gridded sea surface height products were provided by the Copernicus Marine Environment Monitoring Service https://doi.org/10.48670/moi-00148 (Ducet et al., 2000). Argo data are available through the Argo Global Data Assembly Centre (GDAC) https://www.seanoe.org/data/00311/42182 (Argo, 2022). Cross-Calibrated Multi-Platform wind vector analysis (CCMP) wind data (Wentz et al., 2015) were downloaded from https://data.remss.com/ccmp/ on 14 April 2021. CCMP Version-2.0 analyses are produced by Remote Sensing Systems and sponsored by NASA Earth Science funding. The ECCO Version 4 Release 4 ocean and sea-ice state estimate (ECCO Consortium et al., 2021) was downloaded from NASA's Physical Oceanography Distributed Active Archive Center (PO.DAAC) https://podaac.jpl.nasa.gov/ECCO.

Supporting Information

References

Argo (2022). Argo float data and metadata from Global Data Assembly Centre (Argo GDAC). SEANOE. https://doi.org/10.17882/42182
Google Scholar
Bingham, R. J., Hughes, C. W., Roussenov, V., & Williams, R. G. (2007). Meridional coherence of the North Atlantic meridional overturning circulation. Geophysical Research Letters, 34(23). https://doi.org/10.1029/2007GL031731
10.1029/2007GL031731
Web of Science®Google Scholar
Bower, A. S., Lozier, M. S., Gary, S. F., & Böning, C. W. (2009). Interior pathways of the North Atlantic meridional overturning circulation. Nature, 459(7244), 243–247. https://doi.org/10.1038/nature07979
10.1038/nature07979
CASADSPubMedWeb of Science®Google Scholar
Caesar, L., McCarthy, G. D., Thornalley, D. J. R., Cahill, N., & Rahmstorf, S. (2021). Current Atlantic Meridional Overturning Circulation weakest in last millennium. Nature Geoscience, 14(3), 118–120. https://doi.org/10.1038/s41561-021-00699-z
10.1038/s41561-021-00699-z
CASADSWeb of Science®Google Scholar
Chomiak, L. N., Yashayaev, I., Volkov, D. L., Schmid, C., & Hooper, J. A. (2022). Inferring advective timescales and overturning pathways of the deep western boundary current in the North Atlantic through Labrador Sea Water advection. Journal of Geophysical Research: Oceans, 127(12), e2022JC018892. https://doi.org/10.1029/2022JC018892
10.1029/2022JC018892
ADSWeb of Science®Google Scholar
Desbruyères, D. G., Mercier, H., Maze, G., & Daniault, N. (2019). Surface predictor of overturning circulation and heat content change in the subpolar North Atlantic. Ocean Science, 15(3), 809–817. https://doi.org/10.5194/os-15-809-2019
10.5194/os-15-809-2019
ADSWeb of Science®Google Scholar
Ducet, N., Le Traon, P. Y., & Reverdin, G. (2000). Global high-resolution mapping of ocean circulation from TOPEX/Poseidon and ERS-1 and -2. Journal of Geophysical Research, 105(C8), 19477–19498. https://doi.org/10.1029/2000JC900063
10.1029/2000JC900063
ADSWeb of Science®Google Scholar
ECCO Consortium, Fukumori, I., Wang, O., Fenty, I., Forget, G., Heimbach, P., & Ponte, R. (2021). Synopsis of the ECCO central production global ocean and sea-ice state estimate. (Version 4 Release 4).
Google Scholar
Frajka-Williams, E., Ansorge, I. J., Baehr, J., Bryden, H. L., Chidichimo, M. P., Cunningham, S. A., et al. (2019). Atlantic meridional overturning circulation: Observed transport and variability. Frontiers in Marine Science, 6. https://doi.org/10.3389/fmars.2019.00260
10.3389/fmars.2019.00260
Web of Science®Google Scholar
Frajka-Williams, E., Moat, B. I., Smeed, D. A., Rayner, D., Johns, W. E., Baringer, M. O., et al. (2021). Atlantic meridional overturning circulation observed by the RAPID-MOCHA-WBTS (RAPID-Meridional Overturning Circulation and Heatflux Array-Western Boundary Time Series) array at 26N from 2004 to 2020 (v2020.1). British Oceanographic Data Centre - Natural Environment Research Council. https://doi.org/10.5285/cc1e34b3-3385-662b-e053-6c86abc03444
Google Scholar
Fraser, N. J., & Cunningham, S. A. (2021). 120 years of AMOC variability reconstructed from observations using the Bernoulli inverse. Geophysical Research Letters, 48(18), e2021GL093893. https://doi.org/10.1029/2021gl093893
10.1029/2021GL093893
ADSWeb of Science®Google Scholar
Gary, S. F., Lozier, M. S., Claus, W. B., & Biastoch, A. (2011). Deciphering the pathways for the deep limb of the Meridional Overturning Circulation. Deep Sea Research II, 58(17–18), 1781–1797. https://doi.org/10.1016/j.dsr2.2010.10.059
10.1016/j.dsr2.2010.10.059
ADSGoogle Scholar
Hobbs, W. R., & Willis, J. K. (2012). Midlatitude North Atlantic heat transport: A time series based on satellite and drifter data. Journal of Geophysical Research, 117(1), 1008. https://doi.org/10.1029/2011JC007039
10.1029/2011JC007039
Web of Science®Google Scholar
IPCC, Pörtner, H.-O., Roberts, D., Masson-Delmotte, V., Zhai, P., Tignor, M., et al. (2019). IPCC special report on the ocean and cryosphere in a changing climate.
Google Scholar
Jackson, L. C., Peterson, K. A., Roberts, C. D., & Wood, R. A. (2016). Recent slowing of Atlantic overturning circulation as a recovery from earlier strengthening. Nature Geoscience, 9(7), 1–6. https://doi.org/10.1038/ngeo2715
10.1038/ngeo2715
Web of Science®Google Scholar
Joyce, T. M., Dunworth-Baker, J., Pickart, R. S., Torres, D., & Waterman, S. (2005). On the deep western boundary current south of Cape Cod. Deep Sea Research II, 52(3–4), 615–625. https://doi.org/10.1016/j.dsr2.2004.12.013
10.1016/j.dsr2.2004.12.013
ADSGoogle Scholar
Kilbourne, K. H., Wanamaker, A. D., Moffa-Sanchez, P., Reynolds, D. J., Amrhein, D. E., Butler, P. G., et al. (2022). Atlantic circulation change still uncertain. Nature Geoscience, 15(3), 165–167. https://doi.org/10.1038/s41561-022-00896-4
10.1038/s41561-022-00896-4
CASADSWeb of Science®Google Scholar
Koelling, J., Send, U., & Lankhorst, M. (2020). Decadal strengthening of interior flow of North Atlantic Deep Water observed by GRACE satellites. Journal of Geophysical Research: Oceans, 125(11), e2020JC016217. https://doi.org/10.1029/2020JC016217
10.1029/2020JC016217
ADSWeb of Science®Google Scholar
Le Bras, I. A., Yashayaev, I., & Toole, J. M. (2017). Tracking Labrador Sea Water property signals along the Deep Western Boundary Current. Journal of Geophysical Research: Oceans, 122(7), 5348–5366. https://doi.org/10.1002//2017JC012921
10.1002/2017JC012921
ADSWeb of Science®Google Scholar
Lobelle, D., Beaulieu, C., Livina, V., Sévellec, F., & Frajka-Williams, E. (2020). Detectability of an AMOC decline in current and projected climate changes. Geophysical Research Letters, 47(20), e2020GL089974. https://doi.org/10.1029/2020GL089974
10.1029/2020GL089974
ADSWeb of Science®Google Scholar
Lozier, M. S. (2010). Deconstructing the conveyor belt. Science, 328(5985), 1507–1511. https://doi.org/10.1126/science.1189250
10.1126/science.1189250
CASADSPubMedWeb of Science®Google Scholar
Lozier, M. S., Gary, S. F., & Bower, A. S. (2013). Simulated pathways of the overflow waters in the North Atlantic: Subpolar to subtropical export. Deep Sea Research II, 85, 147–153. https://doi.org/10.1016/j.dsr2.2012.07.037
10.1016/j.dsr2.2012.07.037
ADSGoogle Scholar
Lozier, M. S., Li, F., Bacon, S., Bahr, F., Bower, A. S., Cunningham, S. A., et al. (2019). A sea change in our view of overturning in the subpolar North Atlantic. Science (New York, N.Y.), 363(6426), 516–521. https://doi.org/10.1126/science.aau6592
10.1126/science.aau6592
CASADSPubMedWeb of Science®Google Scholar
Marshall, J., Adcroft, A., Hill, C., Perelman, L., & Heisey, C. (1997). A finite-volume, incompressible Navier Stokes model for studies of the ocean on parallel computers. Journal of Geophysical Research, 102(C3), 5753–5766. https://doi.org/10.1029/96JC02775
10.1029/96JC02775
ADSWeb of Science®Google Scholar
McCarthy, G., Frajka-Williams, E., Johns, W. E., Baringer, M. O., Meinen, C. S., Bryden, H. L., et al. (2012). Observed interannual variability of the Atlantic meridional overturning circulation at 26.5°N. Geophysical Research Letters, 39(19). https://doi.org/10.1029/2012GL052933
10.1029/2012GL052933
PubMedWeb of Science®Google Scholar
McCarthy, G., Smeed, D., Johns, W., Frajka-Williams, E., Moat, B., Rayner, D., et al. (2015). Measuring the Atlantic meridional overturning circulation at 26°N. Progress in Oceanography, 130, 91–111. https://doi.org/10.1016/J.POCEAN.2014.10.006
10.1016/j.pocean.2014.10.006
ADSWeb of Science®Google Scholar
Moat, B. I., Smeed, D. A., Frajka-Williams, E., Desbruyères, D. G., Beaulieu, C., Johns, W. E., et al. (2020). Pending recovery in the strength of the meridional overturning circulation at 26°N. Ocean Science, 16(4), 863–874. https://doi.org/10.5194/os-16-863-2020
10.5194/os-16-863-2020
ADSWeb of Science®Google Scholar
Pena-Molino, B., Joyce, T. M., & Toole, J. M. (2011). Recent changes in the Labrador Sea Water within the deep western boundary current southeast of Cape Cod. Deep-Sea Research Part I: Oceanographic Research Papers, 58(10), 1019–1030. https://doi.org/10.1016/j.dsr.2011.07.006
10.1016/j.dsr.2011.07.006
ADSWeb of Science®Google Scholar
Piecuch, C. G. (2020). Likely weakening of the Florida Current during the past century revealed by sea-level observations. Nature Communications, 11(1), 3973. https://doi.org/10.1038/s41467-020-17761-w
10.1038/s41467-020-17761-w
CASPubMedWeb of Science®Google Scholar
Rossby, T., Chafik, L., & Houpert, L. (2020). What can hydrography tell us about the strength of the Nordic Seas MOC over the last 70 to 100 years? Geophysical Research Letters, 47(12), e2020GL087456. https://doi.org/10.1029/2020GL087456
10.1029/2020GL087456
ADSWeb of Science®Google Scholar
Sinha, B., Smeed, D. A., McCarthy, G., Moat, B. I., Josey, S. A., Hirschi, J. J. M., et al. (2018). The accuracy of estimates of the overturning circulation from basin-wide mooring arrays. Progress in Oceanography, 160, 101–123. https://doi.org/10.1016/j.pocean.2017.12.001
10.1016/j.pocean.2017.12.001
ADSWeb of Science®Google Scholar
Smeed, D. A., Josey, S. A., Beaulieu, C., Johns, W. E., Moat, B. I., Frajka-Williams, E., et al. (2018). The North Atlantic Ocean is in a state of reduced overturning. Geophysical Research Letters, 45(3), 1527–1533. https://doi.org/10.1002/2017GL076350
10.1002/2017GL076350
ADSWeb of Science®Google Scholar
Smith, S. D. (1980). Wind stress and heat flux over the ocean in gale force winds. Journal of Physical Oceanography, 10(5), 709–725. https://doi.org/10.1175/1520-0485(1980)010<0709:wsahfo>2.0.co;2
10.1175/1520-0485(1980)010<0709:WSAHFO>2.0.CO;2
ADSWeb of Science®Google Scholar
Thornalley, D. J. R., Oppo, D. W., Ortega, P., Robson, J. I., Brierley, C. M., Davis, R., et al. (2018). Anomalously weak Labrador Sea convection and Atlantic overturning during the past 150 years. Nature, 556(7700), 227–230. https://doi.org/10.1038/s41586-018-0007-4
10.1038/s41586-018-0007-4
CASADSPubMedWeb of Science®Google Scholar
Toole, J. M., Andres, M., Le Bras, I. A., Joyce, T. M., & McCartney, M. S. (2017). Moored observations of the Deep Western Boundary Current in the NW Atlantic: 2004–2014. Journal of Geophysical Research: Oceans, 122(9), 7488–7505. https://doi.org/10.1002/2017JC012984
10.1002/2017JC012984
ADSWeb of Science®Google Scholar
Toole, J. M., Curry, R. G., Joyce, T. M., McCartney, M., & Peña-Molino, B. (2011). Transport of the North Atlantic deep western boundary current about 39°N, 70°W: 2004–2008. Deep Sea Research Part II: Topical Studies in Oceanography, 58(17–18), 1768–1780. https://doi.org/10.1016/j.dsr2.2010.10.058
10.1016/j.dsr2.2010.10.058
ADSWeb of Science®Google Scholar
Van Sebille, E., Baringer, M. O., Johns, W. E., Meinen, C. S., Beal, L. M., De Jong, M. F., & Van Aken, H. M. (2011). Propagation pathways of classical Labrador Sea water from its source region to 26°N. Journal of Geophysical Research, 116(12), 1–18. https://doi.org/10.1029/2011JC007171
10.1029/2011JC007171
Web of Science®Google Scholar
Wentz, F. J., Scott, J., Hoffman, R., Leidner, M., Atlas, R., & Ardizzone, J. (2015). Remote sensing systems cross-calibrated multi-platform (CCMP) 6-hourly ocean vector wind analysis product on 0.25 deg grid, Version 2.0. Remote Sensing Systems.
Google Scholar
Willis, J. K. (2010). Can in situ floats and satellite altimeters detect long-term changes in Atlantic Ocean overturning? Geophysical Research Letters, 37(6), L06602. https://doi.org/10.1029/2010GL042372
10.1029/2010GL042372
ADSWeb of Science®Google Scholar
Willis, J. K., & Fu, L.-L. (2008). Combining altimeter and subsurface float data to estimate the time-averaged circulation in the upper ocean. Journal of Geophysical Research, 113(C12), C12017. https://doi.org/10.1029/2007JC004690
10.1029/2007JC004690
Web of Science®Google Scholar
Yashayaev, I. (2007). Hydrographic changes in the Labrador Sea, 1960-2005. Progress in Oceanography, 73(3–4), 242–276. https://doi.org/10.1016/j.pocean.2007.04.015
10.1016/j.pocean.2007.04.015
ADSWeb of Science®Google Scholar
Yashayaev, I., & Loder, J. W. (2016a). Further intensification of deep convection in the Labrador Sea in 2016. Geophysical Research Letters, 44(3), 1429–1438. https://doi.org/10.1002/2013GL058740.Received
10.1002/2016GL071668
ADSWeb of Science®Google Scholar
Yashayaev, I., & Loder, J. W. (2016b). Recurrent replenishment of Labrador Sea Water and associated decadal-scale variability. Journal of Geophysical Research: Oceans, 121(11), 8095–8114. https://doi.org/10.1002/2016JC012046
10.1002/2016JC012046
ADSWeb of Science®Google Scholar
Zhai, Y., Yang, J., Wan, X., & Zou, S. (2021). The eastern Atlantic basin pathway for the export of the North Atlantic deep waters. Geophysical Research Letters, 48(24), e2021GL095615. https://doi.org/10.1029/2021GL095615
10.1029/2021GL095615
ADSWeb of Science®Google Scholar
Zou, S., Lozier, M. S., & Buckley, M. (2019). How is meridional coherence maintained in the lower limb of the Atlantic meridional overturning circulation? Geophysical Research Letters, 46(1), 244–252. https://doi.org/10.1029/2018GL080958
10.1029/2018GL080958
ADSWeb of Science®Google Scholar

References From the Supporting Information

Bayazit, M., & Önöz, B. (2007). To prewhiten or not to prewhiten in trend analysis? Hydrological Sciences Journal, 52(4), 611–624. https://doi.org/10.1623/hysj.52.4.611
10.1623/hysj.52.4.611
Web of Science®Google Scholar
Bayley, G. V., & Hammersley, J. M. (1946). The “Effective” number of independent observations in an autocorrelated time series. Supplement to the Journal of the Royal Statistical Society, 8(2), 184–197. https://doi.org/10.2307/2983560
10.2307/2983560
Google Scholar
Bessa, I., Makaoui, A., Hilmi, K., Idrissi, M., Agouzouk, A., Ettahiri, O., & Afifi, M. (2021). Marine circulation along the Moroccan Atlantic coast. Frontiers in Science and Engineering, 11(1). https://doi.org/10.34874/IMIST.PRSM/fsejournal-v11i1.28864
10.34874/IMIST.PRSM/fsejournal-v11i1.28864
Google Scholar
Casanova-Masjoan, M., Pérez-Hernández, M. D., Vélez-Belchí, P., Cana, L., & Hernández-Guerra, A. (2020). Variability of the Canary Current diagnosed by inverse box models. Journal of Geophysical Research: Oceans, 125(8), e2020JC016199. https://doi.org/10.1029/2020JC016199
10.1029/2020JC016199
ADSWeb of Science®Google Scholar
Collaud Coen, M., Andrews, E., Bigi, A., Martucci, G., Romanens, G., Vogt, F. P. A., & Vuilleumier, L. (2020). Effects of the prewhitening method, the time granularity, and the time segmentation on the Mann–Kendall trend detection and the associated Sen’s slope. Atmospheric Measurement Techniques, 13(12), 6945–6964. https://doi.org/10.5194/amt-13-6945-2020
10.5194/amt-13-6945-2020
ADSWeb of Science®Google Scholar
Dee, D. P., Uppala, S. M., Simmons, A. J., Berrisford, P., Poli, P., Kobayashi, S., et al. (2011). The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quarterly Journal of the Royal Meteorological Society, 137(656), 553–597. https://doi.org/10.1002/qj.828
10.1002/qj.828
ADSWeb of Science®Google Scholar
Efron, B., & Gong, G. (1983). A leisurely look at the bootstrap, the jackknife, and cross-validation. The American Statistician, 37(1), 36–48. https://doi.org/10.2307/2685844
10.1080/00031305.1983.10483087
Web of Science®Google Scholar
Elipot, S., Frajka-Williams, E., Hughes, C. W., & Willis, J. K. (2014). The observed North Atlantic meridional overturning circulation: Its meridional coherence and ocean bottom pressure. Journal of Physical Oceanography, 44(2), 517–537. https://doi.org/10.1175/JPO-D-13-026.1
10.1175/JPO-D-13-026.1
ADSWeb of Science®Google Scholar
Firing, E. (1989). Mean zonal currents below 1500 m near the equator, 159°W. Journal of Geophical Research, 94(88), 2023–2028. https://doi.org/10.1029/jc094ic02p02023
10.1029/JC094iC02p02023
ADSWeb of Science®Google Scholar
Flagg, C. N., Dunn, M., Wang, D. P., Rossby, H. T., & Benway, R. L. (2006). A study of the currents of the outer shelf and upper slope from a decade of shipboard ADCP observations in the Middle Atlantic Bight. Journal of Geophysical Research, 111(6), 1–20. https://doi.org/10.1029/2005JC003116
10.1029/2005JC003116
PubMedWeb of Science®Google Scholar
Forsyth, J., Andres, M., & Gawarkiewicz, G. (2020). Shelfbreak jet structure and variability off New Jersey using ship of opportunity data from the CMV Oleander. Journal of Geophysical Research: Oceans, 125(9). https://doi.org/10.1029/2020JC016455
10.1029/2020JC016455
Google Scholar
Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., & Gandin, L. (1996). The NCEP/NCAR 40-year reanalysis project. Bulletin of the American Meteorological Society.
10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2
Google Scholar
NCEP, N. W. S. N. O. A. A. (1996). NCEP/NCAR reanalysis monthly mean subsets (from DS090.0), 1948-continuing. Research Data Archive at the National Center for Atmospheric Research, Computational and Information Systems Laboratory. National Centers for Environmental Prediction/National Weather Service/NOAA/U.S. Department of Commerce. https://doi.org/10.5065/4Z6T-J350
Google Scholar
Palter, J. B., Caron, C. A., Law, K. L., Willis, J. K., Trossman, D. S., Yashayaev, I. M., & Gilbert, D. (2016). Variability of the directly observed, middepth subpolar North Atlantic circulation. Geophysical Research Letters, 43(6), 2700–2708. https://doi.org/10.1002/2015GL067235
10.1002/2015GL067235
ADSWeb of Science®Google Scholar
Pearson, E. S. (1931). The test of significance for the correlation coefficient. Journal of the American Statistical Association, 26(174), 128–134. https://doi.org/10.1080/01621459.1931.10503208
10.1080/01621459.1931.10503208
Google Scholar
Pérez-Hernández, M. D., McCarthy, G. D., Vélez-Belchí, P., Smeed, D. A., Fraile-Nuez, E., & Hernández-Guerra, A. (2015). The Canary Basin contribution to the seasonal cycle of the Atlantic Meridional Overturning Circulation at 26°N. Journal of Geophysical Research: Oceans, 120(11), 7237–7252. https://doi.org/10.1002/2015JC010969
10.1002/2015JC010969
ADSWeb of Science®Google Scholar
Yue, S., & Wang, C. Y. (2002). Applicability of prewhitening to eliminate the influence of serial correlation on the Mann-Kendall test. Water Resources Research, 38(6), 4-1–4-7. https://doi.org/10.1029/2001WR000861
10.1029/2001WR000861
ADSWeb of Science®Google Scholar

Citing Literature

Volume50, Issue10

28 May 2023

e2022GL101931

The Atlantic Meridional Overturning Circulation at 35°N From Deep Moorings, Floats, and Satellite Altimeter

Abstract

Key Points

Plain Language Summary

1 Introduction