1. Introduction
The Atlantic meridional overturning circulation (AMOC) carries warm and salty waters northward throughout the upper 1000 m and transports cold and fresh waters southward at a deeper layer, which is argued to be of vital importance for the Earth’s climate (Buckley and Marshall 2016; Zhao et al. 2018). Therefore, ongoing efforts have been made to monitor and understand the variability of AMOC. Climate models show that the AMOC has been weakening due to anthropogenic global warming since the Industrial Revolution (Gregory et al. 2005; Liu et al. 2017; McCarthy et al. 2020a). Long-term fingerprints also provide indirectly detected evidence for a weakening AMOC (Caesar et al. 2018; Rahmstorf et al. 2015; Thibodeau et al. 2018; Thornalley et al. 2018). Observing arrays at different latitudes have been established to seek direct evidence for the expected AMOC slowdown (Frajka-Williams et al. 2019). However, due to a lack of long-term observational records for the AMOC over the past several decades, direct evidence of the expected AMOC slowdown remains inadequate (Bryden et al. 2005; Robson et al. 2014; Smeed et al. 2014, 2018; Srokosz and Bryden 2015). Among the short-term observations in the recent decade, an AMOC decline was detected by the Rapid Climate Change (RAPID)–Meridional Overturning Circulation and Heat-Flux Array (MOCHA) project (hereafter RAPID), which led to strong interest in the AMOC variability across the subtropical Atlantic (McCarthy et al. 2015; Smeed et al. 2018). Since then, continuous efforts have been dedicated to clarify what is responsible for this decline (Roberts et al. 2013; Robson et al. 2014; Zhao and Johns 2014).
Based on wind-driven dynamics and thermal wind relationship respectively, two hypotheses have been raised up for this interannual to quasi-decadal AMOC decline trend.
Zhao and Johns (2014) simulated the interannual variation of the 26.5°N AMOC with a wind-driven model, suggesting that the AMOC decline might arise from changes in sea surface wind stress. Physically, AMOC fluctuations related to the wind stress are mostly confined to intraseasonal and interannual time scales (Gu et al. 2020; Polo et al. 2014). Therefore, the wind-driven mechanism suggests that the 26.5°N AMOC decline may be part of interannual cycle.
Robson et al. (2014) attributed this detected short-term AMOC decline to a rapid density drop from the Labrador Sea since the late 1990s, which was linked to a rapid warming of the upper North Atlantic Ocean and might be part of a substantial AMOC reduction occurring on a decadal time scale. Smeed et al. (2018) claimed that the AMOC decline related to the density decreasing from the Labrador Sea might lead to a hypothesis that it is part of the air–sea interaction cycle.
Recently, it was proposed that over 80% of the low-frequency AMOC variance can be calculated by the thermal wind relationship from density difference between both zonal boundaries (Waldman et al. 2021). Waldman et al. (2021) further proved that the southward-propagating temperature anomalies along the western boundary were the main cause of AMOC variance. According to the RAPID line, the AMOC decline is also dominated by the strengthening of upper southward geostrophic transport (GEO), which is calculated from thermal wind relationship (Smeed et al. 2018; Fig. 2; Table 1). This evidence agrees with the hypothesis in Robson et al. (2014) that the warming signal exported from the Labrador Sea induces this subtropical AMOC decline. However, it is still unclear to what extent the warming signal contributes to the subtropical AMOC decline and whether there is another signal related to this decline. Quantifying and tracing the contributions of respective density anomaly signals with different sources to the GEO variation may provide a feasible method to clarify the above question.
Mean and variation values of transport components. The variations significant at the 95% confidence level are shown in bold font. The calculation methods of variations and their significances are introduced in the section 2c. Positive values represent northward transport for means and northward strengthening for variations.
To clarify the contributions and sources of corresponding density anomaly signals, comparing the relative importance of temperature and salinity offers substantial benefits. This is because the relative contributions of temperature and salinity to the density anomalies may vary with different sources. According to the subtropical AMOC variability, however, people pay more attention to the influence of temperature changes, while the contribution of salinity is overlooked (Robson et al. 2014). A recent work with Coupled Model Intercomparison Project phase 5 (CMIP5) results proposed that salinity might play a more complex role than previously thought (Levang and Schmitt 2020). In model simulations, salinity perturbations originating from the subpolar Atlantic are indicated to trigger a corresponding change in AMOC strength (Zhang 2010). Accordingly, salinity is recognized to play a vital role in regulating the subpolar AMOC by influencing deep convection (Yang et al. 2016). Once salinity changes in the subpolar Atlantic, the signal can be exported into the subtropical Atlantic via advection, boundary waves, or both (Zhang 2010), and the imported signal finally changes the subtropical AMOC via a thermal wind relationship.
Therefore, this work is dedicated to quantifying and comparing the contributions of temperature and salinity anomalies to the AMOC decline, and tries to identify the thermohaline signals which act as prominent drivers of the AMOC decline. For that, we introduce in section 2 the data and a GEO decomposition method that quantifies the contributions of temperature and salinity. In section 3, GEO is decomposed into its temperature and salinity components to compare their contributions. Besides, the comparisons are made in three vertical layers. In section 4, we identify two significant thermohaline signals which are causes of the AMOC decline and relate them to the sources of dense water variability in subpolar North Atlantic. In section 5, we conclude and discuss our results.
2. Data and methods
Data used here are mainly from two hydrographic arrays in the Atlantic: the continuous transbasin array moored at 26.5°N from 2004 to 2020 named RAPID line and the repeated hydrographic sections across the Labrador Sea from 1990 to 2012 named the AR7W line (Hall et al. 2013) (Fig. 1). The main body of this paper is based on the mooring profiles of the RAPID line to compare the contributions of temperature and salinity to the subtropical AMOC decline. Before the following analysis, data from the RAPID project are bin-averaged into monthly time series. The temperature and salinity records of the upstream AR7W line are compared with those of the RAPID line to explore the sources of corresponding signals.
Locations of transects across 26.5°N (RAPID) and across the Labrador Sea (AR7W). The blue shading shows topography. Red pentagrams mark the midocean end points of the RAPID line (26.99°N, 16.23°W at east and 26.52°N, 76.74°W at west). The magenta dashed line shows the path of the deep western boundary current, as is assumed in this study, along which the thermohaline signals propagate southward. The magenta dashed line is drawn along the core of maximum velocity within the path of the deep western boundary current in ECCO4 (see Fig. S1 in the supplemental material).
Citation: Journal of Physical Oceanography 53, 4; 10.1175/JPO-D-22-0087.1
a. Data introduction
1) RAPID observations
The RAPID project detects the whole net transport across 26.5°N since 2004 (Frajka-Williams et al. 2021; McCarthy et al. 2015). It consists of five components [see full details of configuration and data processing in McCarthy et al. (2015)]: cable recorded Gulf Stream across Florida Strait, namely, the western boundary current (WBC), upper geostrophic shear transport (GEO) between the Bahamas and Canary Islands derived from temperature and salinity profiles at both zonal boundaries via thermal wind, transport across western boundary wedge areas (WBW) between the Bahamas and 76.74°W directly observed by current meters, Ekman transport (EKM) derived from reanalysis wind data, and barotropic external component (COM) calculated by the zero-flow assumption. Frajka-Williams et al. (2021) published the transport components from April 2004 to March 2020 (Fig. 2). The GEO, WBW, and COM are not separately supplied by the data publisher but given as their sum named upper midocean transport (UMO). The northward WBW was treated as part of WBC in Smeed et al. (2018). In this study, WBC is defined as the cable recorded transport across Florida Strait, while WBW is part of UMO. Their contributions to the AMOC decline are discussed in section 2d.
Time series of monthly transport components at 26.5°N. For all time series, positive values represent northward transport. The violet, green, orange, and blue lines show the total Atlantic meridional overturning circulation (MOC), the western boundary current restricted to the Florida Strait (WBC), the Ekman component (EKM), and the upper midocean recirculation (UMO). All the time series above are provided by Frajka-Williams et al. (2021). Means are shown for two periods: April 2004–March 2008 and April 2008–March 2020. The shadings show the 95% confidence intervals for these means. The red lines overlapped by MOC and UMO are total transport across the whole basin (AMOC1000) and across the midocean (UMO1000) above 1000 m, respectively. The values of means and variations are shown in Table 1.
Citation: Journal of Physical Oceanography 53, 4; 10.1175/JPO-D-22-0087.1
The AMOC strength at 26.5°N is defined as the maximum of the overturning streamfunction in depth space. Due to the northward transport is approximately restricted to the upper ocean above 1000 m depth, the overturning streamfunction [also provided by Frajka-Williams et al. (2021)] at 1000 m (AMOC1000) was commonly used to stand for the strength of AMOC (Bryden et al. 2005), with the same mean and variation values with the total AMOC in Frajka-Williams et al. (2021) (see Fig. 2 and Table 1). Following the above approximation, in this study, we focus on what causes the decline of AMOC1000, namely, the AMOC decline.
2) ECCO4 assimilation results
Estimating the Circulation and Climate of the Ocean version 4 release 4 (ECCO4; Forget et al. 2015; Fukumori et al. 2020) is used to exhibit the progress of the southward propagation of the freshening signal in Nordic Sea Water. ECCO4 is based on the ocean circulation model (MITgcm). Almost all the observational records are assimilated. It provides monthly data covering the period from January 1992 to December 2017.
b. Calculating and decomposing the GEO transport
The shallowest measurement of the RAPID profiles is within the 100–200-dbar depth range. Different methods of extrapolation will lead to a 0.6-Sv (1 Sv ≡ 106 m3 s−1) deviation in the mean value of AMOC (McCarthy et al. 2015). To ensure the validity of the results, different extrapolation methods have been evaluated (see details in appendix A).
In this study, it is assumed that density anomaly signals imported by the deep western boundary current generally induce this AMOC decline. A rough comparison of thermohaline signals between individual boundaries attributes this AMOC decline to the temperature and salinity anomalies along deep western boundary. The deep western boundary current consists of Labrador Sea Water (LSW) and Nordic Sea Water. Then the ocean can be divided into three vertical layers. The boundary of upper and middle layers is the 1000-m isobath, which is roughly the interface between upper northward limb and lower southward limb of AMOC. The 2000-m isobath is used as the interface of middle and lower layers, which is roughly the interface of LSW and Nordic Sea Water (Fig. 4). Accordingly, Eq. (1) is used to calculate components of GEO, SC, and TC in the upper, middle, and lower layers. When we calculate ϕ in the upper-contributed SC, only original salinity time series upper than 1000 m at both boundaries are preserved. Meanwhile, we replace salinity time series lower than 1000 m and all the temperature time series with their climatology from April 2004 to March 2020. The similar methods are used in calculating other components. Then we remove the 2004–20 climatology from each component to compare their anomalies. This decomposition evaluates the contributions of density signals from different layers. Furthermore, to evaluate the contributions of individual thermohaline signals, we use the same method to distinguish GEO, SC and TC anomalies arising from individual signals in LSW and Nordic Sea Water.
c. Variations of transport components and their significances
Besides a notable drop in 2009, the transport from April 2008 to March 2017 has been fairly stable with the mean AMOC strength roughly 2.7 Sv weaker than that from April 2004 to March 2008 (Frajka-Williams et al. 2019; Smeed et al. 2018). When the observation period is extended to March 2020, the mean value from April 2008 to March 2020 is also significantly less than that during the first four years (Fig. 2). Accordingly, variation value calculated by subtracting transport mean during April 2004 to March 2008 from that during April 2008 to March 2020 is used to quantify the AMOC decline in this study. The 95% confidence intervals for the means during corresponding periods are calculated as
d. Revalidating the dominant role of GEO strengthening
By identifying the dynamical drivers of AMOC variability in the CNRM-CM6 climate model, Waldman et al. (2021) claimed that GEO explained over 80% of the low-frequency AMOC variance. In the observations at 26.5°N, the low-frequency AMOC variance has also been mirrored by the GEO transport in previous studies (Bryden et al. 2005). In this study, we focus on the cause of the southward strengthening of GEO to explain what leads to the recent AMOC decline at 26.5°N. However, in addition to GEO, there are four other components (i.e., WBC, EKM, WBW, and COM) constituting the total volume transport of the AMOC. Here we show that the southward strengthening of GEO dominates the AMOC decline.
Figure 2 shows the time series of total AMOC and its components provided directly by the RAPID project. Here we revalidate the dominant role of GEO strengthening in the AMOC decline with the extended data. The mean of AMOC (as well as AMOC1000) from April 2008 to March 2020 was 2.5 Sv weaker than that from April 2004 to March 2008 (see values in Table 1). Eighty-four percent of this significant AMOC decline comes from a 2.1-Sv strengthening of the southward UMO (as well as UMO1000). Meanwhile, WBC weakens by 0.5 Sv and EKM strengthens by 0.1 Sv, both of which are not significant at the 95% level.
Then UMO1000 is decomposed into GEO and remainder (Fig. 3). The southward strengthening of GEO explains 124% of the UMO1000 variation, while the extra 24% is compensated by a southward weakening in remainder. The strengthening of GEO is significant at the 95% level, but the weakening of remainder is not. Here the remainder, whose physical meaning is COM added by WBW (their transport components above 1000 m), is calculated by subtracting GEO from UMO1000. The remainder’s contribution to the AMOC decline will not be further discussed in this study referring to the previous hypothesis that the geostrophic shear transport can mirror the low-frequency AMOC variability (Frajka-Williams 2015; Waldman et al. 2021). Collectively, the 2.6-Sv southward GEO strengthening can generally explain the 2.5-Sv AMOC decline.
Decompositions of (left) UMO1000 and (right) upper geostrophic shear transport (GEO). Here, UMO1000 is decomposed into two components: (a) GEO computed by Eq. (1) and (b) the remainder computed by subtracting GEO from UMO1000. The remainder consists of transport component across western boundary wedge areas (WBW) and the barotropic external component (COM). GEO is decomposed into its (c) salinity component (SC) and (d) temperature component (TC). The details have the same meaning as Fig. 2. The values of means and variations are shown in Table 1. Thin and thick lines separately show monthly and 13-month low-pass-filtered results.
Citation: Journal of Physical Oceanography 53, 4; 10.1175/JPO-D-22-0087.1
3. Contributions of temperature and salinity changes
Since the AMOC decline can be mirrored by the strengthening of GEO, the GEO variation is then decomposed into its temperature (TC) and salinity (SC) components to explore their causes (Figs. 3c,d and Table 1). The southward SC and TC separately strengthen by 1.6 and 1.0 Sv. The variation of SC is significant but that of TC is not compared with its large amplitude of high-frequency variation. SC seems to play a more important role in the GEO strengthening, which is different from the previous assumption that a warming signal in LSW causes the detected AMOC decline (Robson et al. 2014). Actually, previous studies based on CMIP5 results have raised the point that salinity plays a nonnegligible role on the long-term AMOC variability (McCarthy et al. 2017; Levang and Schmitt 2020). These results call for a reassessment of halosteric influence on the AMOC variability and raises the question about what causes the recent AMOC decline.
The GEO transport is calculated by the double vertical integration of density gradients as shown in Eqs. (1) and (2). Density anomalies at any depth will be related to corresponding changes in GEO. Here we clarify that all the GEO components discussed later are geostrophic shear transport integrated from 0- to 1000-m depth [as is shown in Eq. (1)] and they are divided by contributions of density anomaly signals within different depth or neutral density ranges. Figure 4 exhibits density anomalies along both boundaries of the mid ocean at 26.5°N and their salinity and temperature contributions. To prove the assumption that the thermohaline signals imported by the deep western boundary current play an important role in the AMOC decline, first, we roughly divide the density anomalies along both boundary profiles into four cells based on depth range: the upper western boundary, the deep western boundary, the upper eastern boundary, and the deep eastern boundary. A depth of 1000 m is chosen as the interface of the upper and deep layers. Along the upper western boundary, the upper eastern boundary and the deep eastern boundary, high-frequency oscillations occur during the whole observation period and no prominent long-term trend is shown (also shown in Figs. S2 and S3 in the online supplemental material). By contrast, the density profile along the deep western boundary shows a significant decreasing trend. Along the deep western boundary, salinity changes deeper than 2000 m make an important contribution to this decreasing trend, while the thermal contribution to the low-frequency trend is restricted to a depth range from 1000 to 2000 m. The 2000-m depth is another important interface separating the deep western boundary into two parts with different relative importance of temperature and salinity changes to the density anomalies. The physical meaning of this 2000-m interface will be discussed in the next section.
Density anomalies (ρa) along the (a)–(c) western and (d)–(f) eastern boundary profiles at 26.5°N and their thermohaline components. (left) Density anomalies and their (center) salinity and (right) temperature components. Gray lines in (a)–(c) are interfaces of different water masses, whose neutral density values are 27.897, 27.983, 28.066, and 28.125 kg m−3. The three water masses divided by gray lines are Labrador Sea Water (LSW), Iceland–Scotland Overflow Water (ISOW), and Denmark–Strait Overflow Water (DSOW) from top to bottom. Black bullets show the depths of instruments deployed in October 2012 (McCarthy et al. 2015). The line plots with confidence intervals corresponding to each vertical range are shown in Figs. S2 and S3.
Citation: Journal of Physical Oceanography 53, 4; 10.1175/JPO-D-22-0087.1
We then interpret these density anomalies into transport anomalies in GEO arising from different vertical layers (Fig. 5). GEO strengthening arising from the middle and lower layers is significant. For the middle layer, temperature changes result in a significant southward strengthening in TC, which is partly compensated by the southward weakening in SC. Density changes in the middle and lower layers collectively bring about the GEO strengthening, and their contributions are 0.8 Sv (31%) and 1.0 Sv (38%), respectively (Table 2). For the lower layer, salinity seems to make more important contribution to the GEO variation by causing a significant southward strengthening in SC. Although temperature changes in the lower layer cause a 0.5-Sv southward strengthening in TC, this strengthening is not significant compared with the large amplitude of the high-frequency oscillations. Temperature and salinity changes along the western boundary are much more significant than those along the eastern boundary (Fig. 4; Figs. S2–S4). This is consistent with the previous result that density changes in the western Atlantic basin play a more important role in the low-frequency AMOC variations (Bingham and Hughes 2008; Frajka-Williams 2015; Kanzow et al. 2008). For the density changes along the eastern boundary, investigation of their causes and influence on the transport variation is left for a future study.
(a) GEO, (b) SC, and (c) TC anomalies arising from density anomalies in the upper, middle, and lower layers. The top, middle, and bottom lines separately show transport anomalies arising from density anomalies across the upper, middle, and lower layers of both boundaries. The depths of 1000 and 2000 m are two interfaces of the three vertical layers. For each transport component, the 2004–20 climatology is removed. The details have the same meanings as Fig. 2.
Citation: Journal of Physical Oceanography 53, 4; 10.1175/JPO-D-22-0087.1
Variation values of GEO, SC, and TC anomalies arising from thermohaline anomalies in the middle and lower layers. The variations significant at the 95% confidence level are shown in bold font. Positive values represent northward strengthening.
Variation values of GEO, SC, and TC anomalies arising from individual signals in LSW and Nordic Sea Water (NSW). The variations significant at the 95% confidence level are shown in bold font. Positive values represent northward strengthening.
The GEO variation arising from density anomalies in the upper layer is not significant and accounts for around 31% of the GEO strengthening (Fig. 5). For the upper layer, salinity changes cause a significant southward strengthening in SC, which is partly compensated by the slight southward weakening in TC. TC brings about prominent high-frequency oscillations in GEO and makes the GEO variation insignificant by comparison. Previous study attributed the upper subtropical Atlantic salinity and temperature changes across the western and eastern subtropical subbasins to meridional convergence or divergence (Bryden et al. 2020; Volkov et al. 2019). A wind-driven variability in Atlantic water mass distribution may explain the mechanism of the upper-contributed GEO strengthening (Evans et al. 2017). Less attention will be paid to the sources and contributions of temperature and salinity changes upper than 1000 m in this study. We follow the previous hypothesis that southward-propagating density signals along deep western boundary current cause the subtropical AMOC decline (Robson et al. 2014; Levang and Schmitt 2020; Waldman et al. 2021). Therefore, we focus on what leads to the temperature and salinity changes along the deep western boundary to explore the causes of the AMOC decline in this paper. According to the results of model simulation, this density signal may be exported from the Labrador Sea (Bailey et al. 2005; Danabasoglu et al. 2012; Eden and Willebrand 2001; Getzlaff et al. 2005). Robson et al. (2014) speculated that density decrease across the Labrador Sea since the middle 1990s might be the cause of the AMOC decline detected by the RAPID project. They related this density decrease to a rapid warming of the upper North Atlantic Ocean but ignored the contribution of salinity. Our results argue that the overlooked salinity makes an important contribution and different signals from multiple sources, rather than a single warming signal exported from the Labrador Sea, along the deep western boundary collectively cause the AMOC decline.
4. Contributions of signals from different water masses
The vertically coherent density decreasing signal along the deep western boundary plays an important role in the AMOC decline. However, the relative importance of temperature and salinity is incoherent between the depth ranges from 1000 to 2000 m and from 2000 m to the bottom (Figs. 4a–c). This incoherence suggests that density decreasing signals within the two depth ranges may come from different sources. These signals locate on the path of the deep western boundary current and so are assumed to be exported from Subpolar North Atlantic. Waldman et al. (2021) showed that the Labrador Sea and the Nordic seas were two major source regions of dense water variability, which drove the centennial AMOC variability in CNRM-CM6 climate model. To identify individual signals from different sources, here we use the standard introduced by Toole et al. (2011) to divide waters along the deep western boundary into LSW and Nordic Sea Water (see details in the description of Fig. 4). Nordic Sea Water consists of Iceland–Scotland Overflow Water (ISOW) and Denmark–Strait Overflow Water (DSOW).
Figure 6 shows time series of the LSW-contributed and Nordic Sea Water–contributed GEO, SC, and TC. Density decreasing signals in the two water masses both significantly strengthen the southward GEO by 0.8 Sv. In LSW, a rapid warming signal causes a 1.8-Sv strengthening in GEO, which is much higher than the contributions of other signals. This result agrees with the hypothesis in Robson et al. (2014). Here the warming signal significantly contributes to the AMOC decline. But the opposite transport variation arising from thermal changes in the upper layer partly cancels the contribution of temperature (Fig. 5). As a result, the contribution of temperature seems to be less important than that of salinity in general (Fig. 3; Table 1). In LSW, this transport strengthening related to the warming signal is partly canceled by a weakening arising from the salty signal. Finally, contribution of density decreasing in LSW to the GEO strengthening is equal to that in Nordic Sea Water. Meanwhile, a significant freshening signal in Nordic Sea Water seems to play a more important role in the strengthening of GEO than temperature, while the variation of TC is not significant.
GEO, SC, and TC anomalies arising from density anomalies in Labrador Sea Water (LSW) and Nordic Sea Water (NSW). Components in (a) GEO, (b) SC, and (c) TC. The top and bottom rows separately show transport anomalies arising from density anomalies in LSW and Nordic Sea Water. The depth range of LSW is from 1250 to 2000 m as shown in Figs. 4a–c. Nordic Sea Water, whose depth range is from 2000 to 3500 m, consists of Iceland–Scotland Overflow Water (ISOW) and Denmark–Strait Overflow Water (DSOW). For each transport component, the 2004–2020 climatology is removed. The details have the same meanings as Fig. 2. The values of variations in each component are shown in Table 3.
Citation: Journal of Physical Oceanography 53, 4; 10.1175/JPO-D-22-0087.1
The warming signal in LSW and the freshening signal in Nordic Sea Water are proved to be two major causes of the AMOC decline. To trace the origins of these two signals, a comparison of the noncontemporaneous records between AR7W, a transect across the Labrador Sea, and the western profile of the RAPID array is conducted. The upstream AR7W section is about 6000 km away from the 26.5°N profile and is exhibited by taking a 14-yr lead, an approximate time scale of advection with a 0.01 m s−1 velocity, of the 26.5°N record (Fig. 7). For convenience, here we grid the temperature and salinity in neutral density coordinate because depth ranges of individual water masses may vary with latitudes. According to LSW, the warming signal can also be found in the upstream record. This signal is proved to be transported southward by the deep western boundary current (Robson et al. 2014). The warming signal in LSW was linked to a rapid warming of the upper North Atlantic Ocean (Robson et al. 2012). This mechanism suggests a decadal cycle between the temperature across the upper North Atlantic and the AMOC strength.
Salinity and temperature anomalies from two transects along the path of deep western boundary current. Salinity anomalies from (a) AR7W and (c) RAPID. Temperature anomalies from (b) AR7W and (d) RAPID. For the profiles from the same transect, we interpolate salinity and temperature profiles to density spaces and, for all profiles in AR7W, derive mean values at each density grid. Gray dashed lines are interfaces of different water masses, with the same meanings as gray lines in Figs. 4a–c but in density space. The upstream record from AR7W is exhibited by taking a 14-yr lead, an approximate time scale of advection, of the 26.5°N record. Downward arrows above (a) and (b) denote the years when the AR7W was occupied.
Citation: Journal of Physical Oceanography 53, 4; 10.1175/JPO-D-22-0087.1
For Nordic Sea Water, a rapid freshening signal is also detected by the AR7W array from 1990 to 2002. The 14-yr-lag correlation coefficient between the salinity time series of Nordic Sea Water detected by the two arrays is 0.75, passing the 95% confidence level. Here we assume that the freshening signal in Nordic Sea Water at 26.5°N is imported by the deep western boundary current from the Labrador Sea as is suggested in Figs. 7a and 7c. Figure 8 shows the progress of the southward propagation of the Nordic Sea Water’s freshening signal in ECCO4. This freshening signal across the Labrador Sea is part of long-term freshening during 1965–2002, imported partly from the Nordic seas (Dickson et al. 2002), which is associated with an increase in the direct export of sea ice from the Arctic Ocean (Dickson et al. 2000; Vinje et al. 1998; Vinje 2000). This freshening signal weakened the RAPID AMOC by about 0.03 ± 0.02 Sv yr−1 (slope ± its 95% interval). This is close to the −0.06 Sv yr−1 weakening trend in climate models (Cheng et al. 2013; McCarthy et al. 2020b; Reintges et al. 2017). This evidence suggests a linkage between the detected AMOC decline and modern anthropogenic warming. Temperature oscillation in Nordic Sea Water at 26.5°N is inconsistent with the upstream cooling trend. There may be other temperature sources for the 26.5°N Nordic Sea Water.
Salinity anomaly along the magenta dashed path in Fig. 1. The vertical mean salinity anomaly of Nordic Sea Water from ECCO4 is shown here. Distance (km) is used to identify Nordic Sea Water in ECCO4 (see a detailed description in appendix B). The y axis denotes distance from the northernmost point in the magenta dashed line in Fig. 1.
Citation: Journal of Physical Oceanography 53, 4; 10.1175/JPO-D-22-0087.1
5. Conclusions and discussion
In this study, we compare the contributions of temperature and salinity changes to the RAPID AMOC decline. In a departure from the previous view that the warming signal in LSW leads to this decline, salinity seems to play a more important role than temperature. The vertically coherent density decreasing signal along the deep western boundary is proved to be the major cause of this AMOC decline. Here, to evaluate the contributions of salinity and temperature to the AMOC decline, an artificial definition of transport variation is adapted. This artificial definition ignores some other low-frequency trends of AMOC. For example, AMOC seems to slightly strengthen since 2009 (Fig. 2), although much weaker than the AMOC decline during 2004–09. This strengthening trend may be caused by more complicated mechanisms and is left for a future work. Frajka-Williams (2015) produced a GEO proxy at 26.5°N using sea level anomalies near the western boundary based on the fact that dynamic height anomalies along the western boundary dominate the GEO interannual variation during 2004–14. Her estimate of AMOC variability at 26.5°N shows a reduction of roughly 1 Sv between the periods from 1993 to 2013 and from 2004 to 2014. However, when the time series was extended to 2020, the previous western domination relationship seemed to be no longer significant (Fig. S4). The thermohaline variations along the western boundary strengthened the southward GEO by 3.8 Sv, while the variations along the eastern boundary weakened GEO by 1.2 Sv. Although the western-contributed component made much more contribution to the 2.6-Sv GEO strengthening, we cannot overlook the 1.2-Sv GEO weakening arising from the thermohaline variations along the eastern boundary. Recently, after capturing changes to the deep circulation and signal along the eastern boundary, a 30-yr reconstruction of subtropical AMOC shows no decline (Worthington et al. 2021). A recent AMOC evaluation based on repeated hydrographic data also suggested that the AMOC remains stable since the 1990s (Fu et al. 2020). These questions are left for the future studies.
The nonnegligible contribution of density changes in LSW to the subtropical AMOC decline underscores the need for continued investigation of the role of the Labrador Sea in the AMOC variability. Results from the Overturning in the Subpolar North Atlantic Program (OSNAP) observation system suggest that transport across the Labrador Sea makes a minor contribution to subpolar AMOC variations (Lozier et al. 2019). Further downstream, with the RAPID results, the variation of transport in the LSW layer also has little influence on the total transport variability (Smeed et al. 2018). These results challenge the previous model results in which shutdown of deep convection in the Labrador Sea is a key determinant of the long-term AMOC slowdown (Brodeau and Koenigk 2016; Heuzé 2017; Yang et al. 2016). Simulated and observed results, however, have shown that the detected AMOC decline at 26.5°N is strongly associated with density anomalies (1000–2500 m in depth) in the Labrador Sea (Robson et al. 2014). Lozier et al. (2019) offered a reconciliation with these results is possible if the density anomalies in the Labrador Sea are signatures of upstream density anomalies. The reconciliation is supported by our conclusion that density anomalies in the Nordic Sea Water layer, together with those in the LSW layer, cause the detected AMOC decline. However, contribution of density change in LSW still occupies 31% of the GEO strengthening at 26.5°N, which is beyond the OSNAP result that the Labrador Sea contributes minimally to the transport variation.
At last, we discuss the time scale of this AMOC decline. The decline is attributed to a warming signal in LSW and a freshening signal in Nordic Sea Water. The warming signal suggests a decadal cycle in AMOC, while the freshening signal suggests a decline on a larger time scale. This freshening signal is speculated to be traced back to the subpolar Atlantic, which may result partly from melting sea ice across the Arctic (Dickson et al. 2002) and spread southward via the deep western boundary current. During 2004–20, freshening in Nordic Sea Water causes a −0.03 ± 0.02 Sv yr−1 weakening trend in AMOC. This is consistent with the previous proposed relationship between the detected AMOC decline and anthropologic global warming (IPCC 2013; McCarthy et al. 2020b). Hence, it is suggested that a slight long-term weakening trend may be overlaid by this AMOC record.
Acknowledgments.
This work was supported by the National Natural Science Foundation of China (42076027, 91858203, and 41606014) and the Hainan Provincial Joint Project of Sanya Yazhou Bay Science and Technology City (120LH059). We acknowledge the use of data from RAPID and AR7W arrays. The authors thank Professor Xiaopei Lin and Professor Xianyao Chen for their advice to improve this manuscript. The authors wish to express their thanks to all the scientific staff, officers, and crews engaged in the progress of observation, and to two anonymous reviewers for their constructive comments on the earlier version of this manuscript.
Data availability statement.
Data from the RAPID AMOC monitoring project are funded by the Natural Environment Research Council and are freely available from www.rapid.ac.uk/rapidmoc (the data can be downloaded from a more direct link: https://rapid.ac.uk/rapidmoc/rapid_data/datadl.php). Labrador Sea data (AR7W) are accessed through https://www.pangaea.de. ECCO4 data can be downloaded from https://ecco.jpl.nasa.gov/drive/files/Version4/Release4 or http://apdrc.soest.hawaii.edu/las/v6/dataset?catitem=4856. If anyone needs the intermediate data used in this study, please contact the corresponding author. MATLAB codes for intermediate data in the main body and appendix A of the paper can be downloaded from https://github.com/Wenbo-Lu/Codes-for-readers.
APPENDIX A
Extrapolation of Temperature and Salinity Profiles
The terms α and β are derived from reference profiles near the locations of the end points (74°–78°W, 26°–27°N for the west and 18°–12°W, 26°–27°N for the east), including CTD, ARGO, and glider profiles from the World Ocean Database (WOD). To maintain the variation on an interannual or quasi-decadal time scale, the reference profiles are divided year by year from 2004 to 2020 (numbers of profiles in each year are shown in Table A1). Then the reference profiles in each year are calculated by the inverse distance weighted method. At last, regression based on the monthly RAPID temperature and salinity profiles combined with the reference profiles is used to calculate the monthly αi and βi. Some years there are only a few WOD profiles, which may bring about deviations in the reference profiles. To check whether these deviations will influence the results in this study, the GEO transport referring to the 2005–17 climatology profiles from World Ocean Atlas (WOA) is compared with that referring to the WOD profiles (Fig. A1). The minor deviation from different reference profiles has nearly no influence on the mean and variation values of GEO.
GEO time series referring to WOD profiles (red) and WOA profiles (blue). The method of shallowest extrapolation used here is the cubic fit. Dotted lines show the mean values for the whole time series. Dashed lines show means from April 2004 to March 2008 and from April 2008 to March 2020.
Citation: Journal of Physical Oceanography 53, 4; 10.1175/JPO-D-22-0087.1
Numbers of profiles from the WOD database in each year.
Figure A2 shows the comparison of shallowest temperature and salinity profiles derived from three extrapolative methods. Compared with the linear extrapolation, the quadratic and cubic methods significantly improve the results, whose vertical changing rates are closer to the reference profiles. Therefore, the GEO transport after quadratic and cubic fit is a little different from that after linear extrapolation (Fig. A3). The mean and variation values of the cubic (as well as quadratic) extrapolative GEO are 0.2 and 0.3 Sv less than those of the linear extrapolative GEO (Table A2). In comparison with the −2.6-Sv variation in GEO, the 0.3-Sv deviation cannot be ignored, especially when discussing the contributions of different GEO components. In this paper, we use the cubic result for a more valid result. Because the GEO component is not supplied by the data provider, we cannot directly judge the accuracy of our GEO by comparing it with that in the RAPID project. As a second-best solution, our GEO is intuitively consistent with the geostrophic shear transport from a depth range of 0–1100 m in a previous study (Frajka-Williams et al. 2018, their Fig. 12b) and is believed valid enough to support the analysis in this study.
Comparison of the shallowest temperature and salinity profiles derived from three extrapolative methods. Reference profiles from WOD are also given for contrast. Mean values of WOD profiles are replaced by those of the cubic fitted profiles.
Citation: Journal of Physical Oceanography 53, 4; 10.1175/JPO-D-22-0087.1
GEO time series with the shallowest temperature and salinity extrapolated by linear (red), quadratic (green), and cubic (blue) fits. Meanings of horizontal dotted and dashed lines are the same as Fig. A1.
Citation: Journal of Physical Oceanography 53, 4; 10.1175/JPO-D-22-0087.1
Mean and variation values of GEO from three extrapolation methods for the shallowest profiles.
APPENDIX B
Trace the Nordic Sea Water’s Freshening Signal in ECCO4
Salinity anomaly profile at 26°N, 76°W in ECCO4. The 2004–16 climatology has been removed for each depth.
Citation: Journal of Physical Oceanography 53, 4; 10.1175/JPO-D-22-0087.1
The distance of the freshening water mass in Fig. B1 is less than 0.3 kg m−3 (Fig. B2), whose downwelling region is restricted to the Nordic seas (Fig. B3). Hence, the freshening signal is proved to originate from the Nordic seas in ECCO4. This is consistent with the observational result in Dickson et al. (2002) that the Nordic Sea Water’s freshening signal from 1960s to 2000s is transported to the Labrador Sea. However, the role of the Labrador Sea in ECCO4 seems to be underestimated. The Labrador Sea seems to influence the 26°N western boundary at a depth range of about 800–1000 m in ECCO4, which is much narrower than the observational results (Figs. B2 and 7). This calls for a more accurate simulation of assimilation models.
Climatology distance across the 26°N Atlantic in ECCO4. Black contours show a few typical values to compare with Fig. B3.
Citation: Journal of Physical Oceanography 53, 4; 10.1175/JPO-D-22-0087.1
Climatology distance across the 50-m North Atlantic in ECCO4. Black contours show a few typical values.
Citation: Journal of Physical Oceanography 53, 4; 10.1175/JPO-D-22-0087.1
REFERENCES
Bailey, D. A., P. B. Rhines, and S. Häkkinen, 2005: Formation and pathways of North Atlantic deep water in a coupled ice–ocean model of the Arctic–North Atlantic Oceans. Climate Dyn., 25, 497–516, https://doi.org/10.1007/s00382-005-0050-3.
Bingham, R. J., and C. W. Hughes, 2008: Determining North Atlantic meridional transport variability from pressure on the western boundary: A model investigation. J. Geophys. Res., 113, C09008, https://doi.org/10.1029/2007JC004679.
Brodeau, L., and T. Koenigk, 2016: Extinction of the northern oceanic deep convection in an ensemble of climate model simulations of the 20th and 21st centuries. Climate Dyn., 46, 2863–2882, https://doi.org/10.1007/s00382-015-2736-5.
Bryden, H. L., H. R. Longworth, and S. A. Cunningham, 2005: Slowing of the Atlantic meridional overturning circulation at 25°N. Nature, 438, 655–657, https://doi.org/10.1038/nature04385.
Bryden, H. L., W. E. Johns, B. A. King, G. McCarthy, E. L. McDonagh, B. I. Moat, and D. A. Smeed, 2020: Reduction in ocean heat transport at 26°N since 2008 cools the Eastern Subpolar Gyre of the North Atlantic Ocean. J. Climate, 33, 1677–1689, https://doi.org/10.1175/JCLI-D-19-0323.1.
Buckley, M. W., and J. Marshall, 2016: Observations, inferences, and mechanisms of the Atlantic meridional overturning circulation: A review. Rev. Geophys., 54, 5–63, https://doi.org/10.1002/2015RG000493.
Caesar, L., S. Rahmstorf, A. Robinson, G. Feulner, and V. Saba, 2018: Observed fingerprint of a weakening Atlantic Ocean overturning circulation. Nature, 556, 191–196, https://doi.org/10.1038/s41586-018-0006-5.
Cheng, W., J. Chiang, and D. Zhang, 2013: Atlantic meridional overturning circulation (AMOC) in CMIP5 models: RCP and historical simulations. J. Climate, 26, 7187–7197, https://doi.org/10.1175/JCLI-D-12-00496.1.
Danabasoglu, G., S. G. Yeager, Y.-O. Kwon, J. J. Tribbia, A. S. Phillips, and J. W. Hurrell, 2012: Variability of the Atlantic meridional overturning circulation in CCSM4. J. Climate, 25, 5153–5172, https://doi.org/10.1175/JCLI-D-11-00463.1.
Dickson, B., I. Yashayaev, J. Meincke, B. Turrell, and J. Holfort, 2002: Rapid freshening of the deep North Atlantic Ocean over the past four decades. Nature, 416, 832–837, https://doi.org/10.1038/416832a.
Dickson, R. R., and Coauthors, 2000: The Arctic Ocean response to the North Atlantic oscillation. J. Climate, 13, 2671–2696, https://doi.org/10.1175/1520-0442(2000)013<2671:TAORTT>2.0.CO;2.
Eden, C., and J. Willebrand, 2001: Mechanism of interannual to decadal variability of the North Atlantic circulation. J. Climate, 14, 2266–2280, https://doi.org/10.1175/1520-0442(2001)014<2266:MOITDV>2.0.CO;2.
Evans, D. G., J. Toole, G. Forget, J. D. Zika, A. C. Naveira Garabato, A. J. G. Nurser, and L. Yu, 2017: Recent wind-driven variability in Atlantic water mass distribution and meridional overturning circulation. J. Phys. Oceanogr., 47, 633–647, https://doi.org/10.1175/JPO-D-16-0089.1.
Forget, G., J.-M. Campin, P. Heimbach, C. N. Hill, R. M. Ponte, and C. Wunsch, 2015: ECCO version 4: An integrated framework for non-linear inverse modeling and global ocean state estimation. Geosci. Model Dev., 8, 3071–3104, https://doi.org/10.5194/gmd-8-3071-2015.
Frajka-Williams, E., 2015: Estimating the Atlantic overturning at 26°N using satellite altimetry and cable measurements. Geophys. Res. Lett., 42, 3458–3464, https://doi.org/10.1002/2015GL063220.
Frajka-Williams, E., M. Lankhorst, J. Koelling, and U. Send, 2018: Coherent circulation changes in the deep North Atlantic from 16°N and 26°N transport arrays. J. Geophys. Res. Oceans, 123, 3427–3443, https://doi.org/10.1029/2018JC013949.
Frajka-Williams, E., and Coauthors, 2019: Atlantic meridional overturning circulation: Observed transport and variability. Front. Mar. Sci., 6, 260, https://doi.org/10.3389/fmars.2019.00260.
Frajka-Williams, E., and Coauthors, 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, accessed 2 December 2021, https://doi.org/10.5285/cc1e34b3-3385-662b-e053-6c86abc03444.
Fu, Y., F. Li, J. Karstensen, and C. Wang, 2020: A stable Atlantic meridional overturning circulation in a changing North Atlantic Ocean since the 1990s. Sci. Adv., 6, eabc7836, https://doi.org/10.1126/sciadv.abc7836.
Fukumori, I., O. Wang, I. Fenty, G. Forget, P. Heimbach, and R. M. Ponte, 2020: ECCO version 4 release 4. PO.DAAC, accessed 21 May 2021, https://ecco.jpl.nasa.gov/drive/files/Version4/Release4/doc/v4r4_synopsis.pdf.
Getzlaff, J., C. W. Böning, C. Eden, and A. Biastoch, 2005: Signal propagation in the North Atlantic overturning. Geophys. Res. Lett., 32, L09602, https://doi.org/10.1029/2004GL021002.
Gregory, J. M., and Coauthors, 2005: A model intercomparison of changes in the Atlantic thermohaline circulation in response to increasing atmospheric CO2 concentration. Geophys. Res. Lett., 32, L12703, https://doi.org/10.1029/2005GL023209.
Gu, S., Z. Liu, and L. Wu, 2020: Time scale dependence of the meridional coherence of the Atlantic meridional overturning circulation. J. Geophys. Res. Oceans, 125, e2019JC015838, https://doi.org/10.1029/2019JC015838.
Hall, M. M., D. J. Torres, and I. Yashayaev, 2013: Absolute velocity along the AR7W section in the Labrador Sea. Deep-Sea Res. I, 72, 72–87, https://doi.org/10.1016/j.dsr.2012.11.005.
Heuzé, C., 2017: North Atlantic deep water formation and AMOC in CMIP5 models. Ocean Sci., 13, 609–622, https://doi.org/10.5194/os-13-609-2017.
Huang, R. X., L. S. Yu, and S. Q. Zhou, 2018: New definition of potential spicity by the least square method. J. Geophys. Res. Oceans, 123, 7351–7365, https://doi.org/10.1029/2018JC014306.
Huang, R. X., L.-S. Yu, and S.-Q. Zhou, 2021: Quantifying climate signals: Spicity, orthogonality, and distance. J. Geophys. Res. Oceans, 126, e2020JC016646, https://doi.org/10.1029/2020JC016646.
IPCC, 2013: Climate Change 2013: The Physical Science Basis. Cambridge University Press, 1535 pp., https://doi.org/10.1017/CBO9781107415324.
Kanzow, T., U. Send, and M. McCartney, 2008: On the variability of the deep meridional transports in the tropical North Atlantic. Deep-Sea Res. I, 55, 1601–1623, https://doi.org/10.1016/j.dsr.2008.07.011.
Levang, S. J., and R. W. Schmitt, 2020: What causes the AMOC to weaken in CMIP5? J. Climate, 33, 1535–1545, https://doi.org/10.1175/JCLI-D-19-0547.1.
Liu, W., S. P. Xie, Z. Liu, and J. Zhu, 2017: Overlooked possibility of a collapsed Atlantic meridional overturning circulation in warming climate. Sci. Adv., 3, e1601666, https://doi.org/10.1126/sciadv.1601666.
Lozier, M. S., F. Li, S. Bacon, F. Bahr, and J. Zhao, 2019: A sea change in our view of overturning in the subpolar North Atlantic. Science, 363, 516–521, https://doi.org/10.1126/science.aau6592.
McCarthy, G. D., and Coauthors, 2015: Measuring the Atlantic meridional overturning circulation at 26°N. Prog. Oceanogr., 130, 91–111, https://doi.org/10.1016/j.pocean.2014.10.006.
McCarthy, G. D., M. B. Menary, J. V. Mecking, B. I. Moat, W. E. Johns, M. B. Andrews, D. Rayner, and D. A. Smeed, 2017: The importance of deep, basinwide measurements in optimized Atlantic meridional overturning circulation observing arrays. J. Geophys. Res. Oceans, 122, 1808–1826, https://doi.org/10.1002/2016JC012200.
McCarthy, G. D., and Coauthors, 2020a: Sustainable observations of the AMOC: Methodology and technology. Rev. Geophys., 58, e2019RG000654, https://doi.org/10.1029/2019RG000654.
McCarthy, G. D., L. C. Jackson, S. A. Cunningham, N. P. Holliday, D. A. Smeed, and D. P. Stevens, 2020b: Effects of climate change on the Atlantic Heat Conveyor relevant to the UK. MCCIP Science Review 2020, Marine Climate Change Impacts Partnership, 190–207, https://doi.org/10.14465/2020.arc09.ahc.
Polo, I., J. Robson, R. Sutton, and M. A. Balmaseda, 2014: The importance of wind and buoyancy forcing for the boundary density variations and the geostrophic component of the AMOC at 26°N. J. Phys. Oceanogr., 44, 2387–2408, https://doi.org/10.1175/JPO-D-13-0264.1.
Rahmstorf, S., J. Box, G. Feulner, M. E. Mann, A. Robinson, S. Rutherford, and E. Schaffernicht, 2015: Corrigendum: Evidence for an exceptional twentieth-century slowdown in Atlantic Ocean overturning. Nat. Climate Change, 5, 956–956, https://doi.org/10.1038/nclimate2781.
Reintges, A., T. Martin, M. Latif, and N. S. Keenlyside, 2017: Uncertainty in twenty-first century projections of the Atlantic meridional overturning circulation in CMIP3 and CMIP5 models. Climate Dyn., 49, 1495–1511, https://doi.org/10.1007/s00382-016-3180-x.
Roberts, C. D., and Coauthors, 2013: Atmosphere drives recent interannual variability of the Atlantic meridional overturning circulation at 26.5°N. Geophys. Res. Lett., 40, 5164–5170, https://doi.org/10.1002/grl.50930.
Robson, J., R. Sutton, K. Lohmann, D. Smith, and M. D. Palmer, 2012: Causes of the rapid warming of the North Atlantic Ocean in the mid-1990s. J. Climate, 25, 4116–4134, https://doi.org/10.1175/JCLI-D-11-00443.1.
Robson, J., D. Hodson, E. Hawkins, and R. Sutton, 2014: Atlantic overturning in decline? Nat. Geosci., 7, 2–3, https://doi.org/10.1038/ngeo2050.
Smeed, D. A., and Coauthors, 2014: Observed decline of the Atlantic meridional overturning circulation 2004–2012. Ocean Sci., 10, 29–38, https://doi.org/10.5194/os-10-29-2014.
Smeed, D. A., and Coauthors, 2018: The North Atlantic Ocean is in a state of reduced overturning. Geophys. Res. Lett., 45, 1527–1533, https://doi.org/10.1002/2017GL076350.
Srokosz, M. A., and H. L. Bryden, 2015: Observing the Atlantic meridional overturning circulation yields a decade of inevi surprises. Science, 348, 1255575, https://doi.org/10.1126/science.1255575.
Thibodeau, B., C. Not, J. Zhu, A. Schmittner, D. Noone, C. Tabor, J. Zhang, and Z. Liu, 2018: Last century warming over the Canadian Atlantic shelves linked to weak Atlantic meridional overturning circulation. Geophys. Res. Lett., 45, 12 376–12 385, https://doi.org/10.1029/2018GL080083.
Thornalley, D. J. R., and Coauthors, 2018: Anomalously weak Labrador Sea convection and Atlantic overturning during the past 150 years. Nature, 556, 227–230, https://doi.org/10.1038/s41586-018-0007-4.
Toole, J. M., R. G. Curry, T. M. Joyce, M. Mccartney, and B. Pena-Molino, 2011: Transport of the North Atlantic deep western boundary current about 39°N, 70°W: 2004–2008. Deep-Sea Res. II, 58, 1768–1780, https://doi.org/10.1016/j.dsr2.2010.10.058.
Vinje, T., 2000: Fram Strait ice fluxes and atmospheric circulation: 1950–2000. J. Climate, 14, 3508–3517, https://doi.org/10.1175/1520-0442(2001)014<3508:FSIFAA>2.0.CO;2.
Vinje, T., N. Nordlund, and A. Kvambekk, 1998: Monitoring ice thickness in Fram Strait. J. Geophys. Res., 103, 10 437–10 449, https://doi.org/10.1029/97JC03360.
Volkov, D. L., M. Baringer, D. Smeed, W. Johns, and F. W. Landerer, 2019: Teleconnection between the Atlantic meridional overturning circulation and sea level in the Mediterranean Sea. J. Climate, 32, 935–955, https://doi.org/10.1175/JCLI-D-18-0474.1.
Waldman, R., J. Hirschi, A. Voldoire, C. Cassou, and R. Msadek, 2021: Clarifying the relation between AMOC and thermal wind: Application to the centennial variability in a coupled climate model. J. Phys. Oceanogr., 51, 343–364, https://doi.org/10.1175/JPO-D-19-0284.1.
Worthington, E. L., B. I. Moat, D. A. Smeed, J. V. Mecking, R. Marsh, and G. D. McCarthy, 2021: A 30-year reconstruction of the Atlantic meridional overturning circulation shows no decline. Ocean Sci., 17, 285–299, https://doi.org/10.5194/os-17-285-2021.
Yang, Q., T. H. Dixon, P. G. Myers, J. Bonin, D. Chambers, M. R. Van den Broeke, M. H. Ribergaard, and J. Mortensen, 2016: Recent increases in Arctic freshwater flux affects Labrador Sea convection and Atlantic overturning circulation. Nat. Commun., 7, 10525, https://doi.org/10.1038/ncomms10525.
Zhang, R., 2010: Latitudinal dependence of Atlantic meridional overturning circulation (AMOC) variations. Geophys. Res. Lett., 37, L16703, https://doi.org/10.1029/2010GL044474.
Zhao, J., and W. Johns, 2014: Wind-forced interannual variability of the Atlantic meridional overturning circulation at 26.5°N. J. Geophys. Res. Oceans, 119, 2403–2419, https://doi.org/10.1002/2013JC009407.
Zhao, J., A. Bower, J. Yang, X. Lin, and N. Penny Holliday, 2018: Meridional heat transport variability induced by mesoscale processes in the subpolar North Atlantic. Nat. Commun., 9, 1124, https://doi.org/10.1038/s41467-018-03134-x.