1. Introduction and background
Labrador Sea Water (LSW) is a critical component of the deep, southward limb of the Atlantic meridional overturning circulation (AMOC). Numerical simulations indicate that the changes in AMOC are linked to the rate of LSW production on interannual to decadal time scales (Böning et al. 2006; Biastoch et al. 2008; Getzlaff et al. 2005). Strong cooling during winters leads to unstable surface stratification that drives convective overturning to depths of approximately 1500 m in the central Labrador Sea (Talley and McCartney 1982). The product of this overturning is a distinct water mass (LSW) with relatively low temperature, salinity (Talley and McCartney 1982; Pickart et al. 2002; Stramma et al. 2004; Yashayaev 2007), and potential vorticity (PV; Talley and McCartney 1982; Stramma et al. 2004) and high concentrations of dissolved oxygen (Pickart et al. 2002) and chlorofluorocarbons (CFCs; Rhein et al. 2002; Kieke et al. 2007).
The exchange of heat and freshwater across the air–sea interface and the horizontal advection of heat and salt into the basin via boundary currents have been proposed as the two major factors that create variability in LSW volume and properties (Stramma et al. 2004). Interannual and interdecadal variability of LSW properties, measured using up to 60 years of hydrographic data, have been linked to changes in the North Atlantic Oscillation (NAO) index (Stramma et al. 2004; Curry et al. 1998; Kieke and Yashayaev 2015). A relatively shallow, warm, and salty layer of LSW was produced from the 1950s to 1970 when the NAO index was negative, a state characterized by a small difference between the Azores high and Icelandic low, reduced heat flux to the atmosphere, and weak westerlies in the subpolar region (Sarafanov 2009; van Aken et al. 2011). The decreased heat flux and the relatively weak westward extension of the subpolar gyre together contributed to the warm and salty intermediate water formed in the Labrador Sea (Sarafanov 2009). In the early 1990s, when the NAO was positive, the strongest recorded convection occurred, resulting in a thick, cold, and fresh LSW layer. From 1994 until 2008, weak convection and steady warming were observed. Enhanced LSW formation resumed in 2008 and was attributed to strong atmospheric cooling (Yashayaev and Loder 2009). Most recently, observations in the central Labrador Sea in May of 2015 revealed a strong convection during the 2014/15 winter (I. Yashayaev 2015, personal communication).
Subsequent to its formation, LSW spreads to other parts of the North Atlantic. From the identification of waters with a PV minimum in the basin, Talley and McCartney (1982) identified three major spreading pathways for this water mass. The first pathway they identified is a southward branch along the western boundary, where low PV is detectable as far as 20°N. Another of their identified spreading pathways extends northeastward into the Irminger Sea, while the third spreads eastward across the subpolar North Atlantic. These pathways have been further confirmed by temperature, salinity, and CFC data (Sy et al. 1997). Rhein et al. (2002) quantified the three LSW spreading pathways using tracer and hydrographic data. According to their results, 21% of the CFC inventory is found south of 53°N, 20% enters the Irminger Sea, and 31% intrudes the eastern subpolar gyre. The rest of the inventory is still in the Labrador Sea.
Recent studies have focused on the spreading of LSW in a Lagrangian frame by tracking floats launched in this water mass. Lavender et al. (2005) studied the middepth circulation in the subpolar North Atlantic with neutrally buoyant profiling floats. Floats that left the Labrador Sea initially drifted southeastward along the Labrador slope, but none followed the Deep Western Boundary Current (DWBC) beyond 44°N, a current previously considered as the major conduit for subpolar water masses to reach the subtropical basin. Similarly, for the RAFOS floats launched at LSW depths in the DWBC off the Labrador coast from 2003 to 2006, only 8% were able to enter the subtropical basin via the DWBC (Bower et al. 2009). Instead, the majority of the RAFOS floats that reached the subtropical basin from the subpolar latitudes did so by interior pathways (Bower et al. 2009; Lozier et al. 2013; Lozier 2012), a finding with implications for where and how the deep limb of the AMOC is measured. The dynamics of these interior pathways, and their favorable comparison with tracer spreading patterns, have recently been explored (Gary et al. 2011, 2012).
This focus on interior pathways, however, has overshadowed another interesting feature of those observed floats, namely, that only 30% of the RAFOS floats were exported to the subtropical gyre over their 2-yr lifetime, despite the fact that all of the floats were launched in LSW in the DWBC. This observation raises the question as to the mechanism that determines how much LSW is exported from the subpolar to the subtropical gyre each year.
For decades it was assumed that convection strength in the Labrador Sea would alter deep water export from the subpolar gyre and modify downstream properties at intermediate depths in the North Atlantic. As for the latter, observational studies have shown a strong correlation between LSW thickness and property anomalies near Bermuda (Curry et al. 1998), with the former leading by 6 years. Similarly, Molinari et al. (1998) and van Sebille et al. (2011) both reported a 10-yr transit time of the property signals from the Labrador Sea to 26.5°N in the DWBC within LSW layers. A more recent study (Pena-Molino et al. 2011) observed that water properties in the classical LSW layer within the DWBC at the Line W mooring array (39°N), measured in the early 2000s, reflect the intense Labrador Sea convection during the mid-1990s, indicating a 9-yr propagation time scale from the central Labrador Sea to Line W. As for the relationship between LSW production and deep water export, some past studies indicate a linkage between LSW production and DWBC strength (Böning et al. 2006; Han et al. 2010), yet others show a contrary result (Schott et al. 2004; Dengler et al. 2006). Relatedly, a linkage between LSW production and AMOC strength, via fast boundary wave propagation, has been revealed from a modeling study (Biastoch et al. 2008) with interannual buoyancy forcing. However, when interannual wind forcing was also considered, the LSW production and AMOC relationship was masked by higher-frequency variability in the AMOC anomalies. Despite this focus on LSW production and its downstream impact, currently unanswered is whether there is a relationship between LSW production and the export of that water mass to the subtropical North Atlantic through advection. Understanding this relationship, which differs from the LSW production/DWBC and LSW production/AMOC relationships, is the goal of this study.
Specifically, we seek to understand the extent to which LSW production impacts the volume of newly formed LSW and blended LSW that are advected across the intergyre boundaries to the subtropical basin, and across 30°N, where LSW and the other components of the southward-flowing North Atlantic Deep Water are less likely to recirculate back to the subpolar gyre. By doing so, we aim to improve the understanding of how convection in the Labrador Sea impacts the lower limb of the AMOC.
The rest of the paper is organized as follows: section 2 describes the data and methods used in this paper. In section 3, we describe our major results, with a focus on the spreading pathways of LSW in a Lagrangian frame and the relationship between LSW production and its export to the subtropical gyre. A summary is given in section 4.
2. Data and methods
In this section, we summarize the observational (section 2a) and model (section 2b) data used in this analysis, and then we compare the model fields with observations in the Labrador basin (section 2c). In section 2d, we give a definition for newly formed LSW, which is used for the float launch configuration described in section 2e. The definition of the intergyre boundaries, which separate the subpolar and subtropical gyres, is described in section 2f.
a. Hydrographic data
The Labrador Sea Monitoring Program of Fisheries and Oceans Canada has been conducting oceanographic observations in the Labrador Sea since 1990 along the Atlantic Repeat Hydrography Line 7 West (AR7W), which extends from Hamilton Bank on the Labrador shelf to Cape Desolation on the Greenland shelf (Fig. 1). It has been occupied annually, typically in May, allowing for a determination of LSW vertical structure at the end of each winter. To compare with the model’s LSW properties, we use the hydrographic data collected from 1992 to 2004 (CDIAC 2015).

Climatological PV (calculated as described in section 2d) and salinity (black contours) from ORCA025 averaged over the LSW layers (700–1500 m). Modeled AR7W is designated with red dots. Bathymetry shallower than 700 m is shaded gray (land masses are dark gray); 1500- and 3000-m isobaths are contoured in gray.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1

Climatological PV (calculated as described in section 2d) and salinity (black contours) from ORCA025 averaged over the LSW layers (700–1500 m). Modeled AR7W is designated with red dots. Bathymetry shallower than 700 m is shaded gray (land masses are dark gray); 1500- and 3000-m isobaths are contoured in gray.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
Climatological PV (calculated as described in section 2d) and salinity (black contours) from ORCA025 averaged over the LSW layers (700–1500 m). Modeled AR7W is designated with red dots. Bathymetry shallower than 700 m is shaded gray (land masses are dark gray); 1500- and 3000-m isobaths are contoured in gray.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
b. ORCA025
To conduct our study, we use ORCA025, a global ocean–sea ice model implemented on a quasi-isotropic Oceanic Remote Chemical Analyzer (ORCA) grid (a tripolar grid) at eddy-permitting resolution (¼°) under the framework of Drakkar, a European modeling project (Barnier et al. 2006, 2007). As described by Barnier et al. (2006, 2007), the configuration of ORCA025, which has 1442 × 1021 grid points and 46 vertical layers, is based on the Nucleus for European Modeling of the Ocean (NEMO) system. Vertical grid spacing increases from 6 m near the surface to 250 m at the bottom. Horizontal resolution increases with latitude, with the coarsest resolution, 27.75 km, at the equator. The model uses 2-min gridded bathymetric data (ETOPO2) from the National Geophysical Data Center (now the National Centers for Environmental Information), and initial conditions are set with a combination of temperature and salinity data derived from Levitus et al. (1998), the Polar Science Center Hydrographic Climatology version 2.1 (PHC2.1; Steele et al. 2001), and the Medatlas climatology (Jourdan et al. 1998).
As reported by Barnier et al. (2006), the climatological daily mean wind stress vector, derived from ERS scatterometer data (CERSAT 2002) and the NCEP–NCAR reanalysis (Kalnay et al. 1996), is used to provide the surface momentum flux. An empirical bulk parameterization (Goosse 1997) is used to compute the surface heat fluxes and freshwater fluxes, with climatological daily mean air temperatures from the NCEP–NCAR reanalysis; climatological monthly mean precipitation from CMAP (Xie and Arkin 1997); and monthly mean humidity (Trenberth et al. 1989), cloud cover (Berliand and Strokina 1980) and climatological daily mean wind speed from a blended ERS and NCEP–NCAR reanalysis.
The 5-day model output used in this study is from the model run forced with the global hindcast dataset from 1961 to 2004. Lagrangian trajectories produced from this model output have been shown to reproduce realistic signatures of the deep recirculation gyres in the North Atlantic (Gary et al. 2011). Gary et al. (2012) also demonstrate that ORCA025 accurately reproduces the distribution of CFC-11 tracers at LSW depths in the North Atlantic. As will be shown in section 3b, ORCA025 can reproduce the LSW spreading pattern depicted by RAFOS floats. Additionally, the modeled LSW volume transport at 53°N of −11.9 ± 0.9 Sv (1 Sv ≡ 106 m3 s−1) compares favorably with the observed transport of −11.3 ± 1.0 Sv based on shipboard lowered ADCP measurements (Fischer et al. 2010). Therefore, we consider ORCA025 highly suitable for the purposes of our study.
c. Model–observation comparison of LSW properties
To ascertain the ability of the model to capture the variability of LSW properties in the Labrador Sea, we compare the model’s temperature, salinity, and density fields at LSW depths (between 700 and 1500 m) along AR7W to observations, using only model data contemporaneous with the observed data. As seen in Fig. 2, the model’s long-term trends and interannual variability compare favorably with those from observations: the decadal trend for observed and modeled temperatures are 0.05° and 0.03°C yr−1, respectively, and after detrending the time series, the standard deviations (SDs) are 0.04° and 0.05°C, respectively. For salinity, the trends between observations and the model are 0.0009 and 0.0003 psu yr−1, with the SD of 0.006 psu for the former and 0.006 psu for the latter after detrending. For density, the trend is −0.004 kg m−3 yr−1 for observations and −0.003 kg m−3 yr−1 for the model. After detrending, the SD for density is 0.006 kg m−3 and 0.004 kg m−3 for the observations and the model, respectively.

Time series of observed (black solid lines) and modeled (gray dashed lines) temperature, salinity, and density averaged between 700 and 1500 m along AR7W. Note that the modeled salinity is slightly higher than observed, and thus the modeled LSW is slightly denser.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1

Time series of observed (black solid lines) and modeled (gray dashed lines) temperature, salinity, and density averaged between 700 and 1500 m along AR7W. Note that the modeled salinity is slightly higher than observed, and thus the modeled LSW is slightly denser.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
Time series of observed (black solid lines) and modeled (gray dashed lines) temperature, salinity, and density averaged between 700 and 1500 m along AR7W. Note that the modeled salinity is slightly higher than observed, and thus the modeled LSW is slightly denser.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
d. Definition of LSW


e. Trajectory computation and Labrador Sea launch configuration
Synthetic floats were launched along a section in the model that replicates AR7W. All launches were dynamic: floats were launched from each grid along AR7W only when properties in a particular grid met the established water mass criteria, given above as PV < 4 × 10−12 m−1 s−1 and density within [27.75, 27.84] kg m−3 (Fig. 3). A year-to-year comparison of the number of floats launched in newly formed LSW along AR7W with the total number of floats launched in newly formed LSW in the entire Labrador Sea yields a correlation coefficient of 0.93. Therefore, the number of floats launched along AR7W for any 5-day period can be used as a proxy for the amount of LSW present for that same period.

PV along the model’s AR7W on (left) 26 Mar 1968 and (right) 1 Mar 1990, when the maximum volume of LSW was formed for each year; 75 floats were launched on the former date and 521 on the latter. Black dots indicate float launch positions. White contours denote where PV is smaller than 4 × 10−12 m−1 s−1, and black dashed lines indicate where density is between 27.75 and 27.84 kg m−3. To avoid surface intensification of floats induced by decreasing model vertical resolution with depth, floats were released at a fixed vertical interval, 200 m, which is the maximum vertical resolution in the upper 2600 m of the model.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1

PV along the model’s AR7W on (left) 26 Mar 1968 and (right) 1 Mar 1990, when the maximum volume of LSW was formed for each year; 75 floats were launched on the former date and 521 on the latter. Black dots indicate float launch positions. White contours denote where PV is smaller than 4 × 10−12 m−1 s−1, and black dashed lines indicate where density is between 27.75 and 27.84 kg m−3. To avoid surface intensification of floats induced by decreasing model vertical resolution with depth, floats were released at a fixed vertical interval, 200 m, which is the maximum vertical resolution in the upper 2600 m of the model.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
PV along the model’s AR7W on (left) 26 Mar 1968 and (right) 1 Mar 1990, when the maximum volume of LSW was formed for each year; 75 floats were launched on the former date and 521 on the latter. Black dots indicate float launch positions. White contours denote where PV is smaller than 4 × 10−12 m−1 s−1, and black dashed lines indicate where density is between 27.75 and 27.84 kg m−3. To avoid surface intensification of floats induced by decreasing model vertical resolution with depth, floats were released at a fixed vertical interval, 200 m, which is the maximum vertical resolution in the upper 2600 m of the model.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
Though floats are launched every 5 days from 1961 to 2004, only one launch is selected each year for the calculation of trajectories. Since we are interested in the export of newly formed LSW following its wintertime production, we choose the launch for which the float number is maximized. As such, float integration for each year begins from the 5-day period when the water mass volume has reached its maximum for that winter. We note that convection during one winter may reach the depth of fossil LSW, that is, LSW formed the previous year or years. In this case, the fossil LSW would be considered part of the newly formed LSW as long as it shares the low PV signature.
For each launch site, float trajectories were calculated from the model’s three-dimensional velocity field using Ariane, a Fortran code for trajectory computation (Blanke and Grima 2010). All float trajectories were integrated for 40 years, and for those launched after 1964, velocity fields were recycled with a single discontinuity between 31 December 2004 and 1 January 1961. This method has been successfully used in previous studies (Gary et al. 2011, 2012). We have also tested the validity of this method by comparing the trajectories computed from sequential years of data and those computed from data with this discontinuity (not shown). The difference in trajectories between the two is inconsequential to our results.
f. Intergyre boundaries






The 44-yr (1961–2004) mean surface dynamic height anomaly
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1

The 44-yr (1961–2004) mean surface dynamic height anomaly
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
The 44-yr (1961–2004) mean surface dynamic height anomaly
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
3. Results
Section 3a describes the interannual variability of LSW formation and its relation to NAO. Detailed pathways and the ages of newly formed LSW in the North Atlantic are summarized in sections 3b and 3c, respectively. In section 3d, the relationship between LSW formation and blended LSW export is studied.
a. Interannual variability of LSW formation and its link to NAO
As mentioned above, we have chosen to quantify the amount of LSW formed each year in the model with the number of floats launched in newly formed LSW. To ascertain the representativeness of the float number as a proxy for LSW formation, we compare the time series of float number with the LSW thickness in the model (Fig. 5). The model LSW thickness is computed as the average depth along AR7W of the water that has PV < 4 × 10−12 m−1 s−1 and density between 27.75 and 27.84 kg m−3 on the same day of the float release. The two variables yield a strong correlation (r = 0.91), leading us to conclude that the number of floats launched when the maximum amount of newly formed LSW is present is a good indicator of the interannual variability in LSW thickness, which has been the traditional measure. Actually, we believe that the float number provides a more accurate estimate of LSW formation since the number of floats released depends not just on the vertical extent of the newly formed water, but on its lateral extent as well.

Interannual variability of float number anomaly (solid red) and LSW thickness anomaly (dashed blue) derived from ORCA025 when the strongest convection takes place each year. The winter NAO index (gray bars) is also shown from 1961 to 2004. The winter NAO index is computed by averaging the monthly NAO index from December to March each year. Data are from NOAA/NWS/CPC (2015).
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1

Interannual variability of float number anomaly (solid red) and LSW thickness anomaly (dashed blue) derived from ORCA025 when the strongest convection takes place each year. The winter NAO index (gray bars) is also shown from 1961 to 2004. The winter NAO index is computed by averaging the monthly NAO index from December to March each year. Data are from NOAA/NWS/CPC (2015).
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
Interannual variability of float number anomaly (solid red) and LSW thickness anomaly (dashed blue) derived from ORCA025 when the strongest convection takes place each year. The winter NAO index (gray bars) is also shown from 1961 to 2004. The winter NAO index is computed by averaging the monthly NAO index from December to March each year. Data are from NOAA/NWS/CPC (2015).
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
As can be seen from Fig. 5, Labrador Sea convection has strong interannual to interdecadal variability, which has been previously linked to the winter NAO index (Kieke and Yashayaev 2015; Sarafanov 2009; van Aken et al. 2011; Kieke et al. 2007; Rhein et al. 2011). A negative NAO winter index has been linked to weaker LSW production from the 1960s to 1970s, while a persistent positive NAO since the 1980s has resulted in strong convective activity and greater LSW formation. The float number anomaly is correlated with NAO (r = 0.62), as is the model’s LSW thickness anomaly (r = 0.66). Both correlations are significant at the 95% confidence level based on a t test.
b. Pathways of newly formed LSW
Bower et al. (2009) have shown the 2-yr spreading pathways of RAFOS floats launched in the DWBC at 50°N every 3 months from 2003 to 2005. To verify that ORCA025 can accurately reproduce this pattern, test floats were launched within the model with a similar observational design, namely, model floats were launched at 50°N at 700 and 1500 m every 3 months from 1961 to 2004 and then integrated forward for 2 years. An example of the float trajectories within 2 years is shown in Fig. 6. Of the floats launched over many releases, 27% ± 6% were able to reach the southern tip of the Grand Banks (43°N), with 7% ± 6% following the DWBC continuously. The majority of floats (73% ± 6%) drifted eastward and northeastward within the subpolar gyre, which compares favorably to the observed percentage (70%; 28/40) of RAFOS floats that took this route (Bower et al. 2009). Thus, based on this comparison, we consider the model capable of approximating the spreading pathways of LSW.

Test float trajectories launched at LSW depths in years from 2000 to 2002 with 2-yr lifetimes. Initial launch locations for all 108 floats (72 at 700 m and 36 at 1500 m) at 50°N are shown in red, and the final positions are indicated with black dots. Colors represent the normalized temperature anomaly (°C) along the path of each float, computed following Bower et al. (2009): (T − Ti)/δTmax, where Ti is each float’s initial temperature, and δTmax is the maximum temperature difference: 3°C for floats launched at 700 m and 0.72°C for those launched at 1500 m. The 700-, 1500-, 3000-m isobaths are contoured in gray.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1

Test float trajectories launched at LSW depths in years from 2000 to 2002 with 2-yr lifetimes. Initial launch locations for all 108 floats (72 at 700 m and 36 at 1500 m) at 50°N are shown in red, and the final positions are indicated with black dots. Colors represent the normalized temperature anomaly (°C) along the path of each float, computed following Bower et al. (2009): (T − Ti)/δTmax, where Ti is each float’s initial temperature, and δTmax is the maximum temperature difference: 3°C for floats launched at 700 m and 0.72°C for those launched at 1500 m. The 700-, 1500-, 3000-m isobaths are contoured in gray.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
Test float trajectories launched at LSW depths in years from 2000 to 2002 with 2-yr lifetimes. Initial launch locations for all 108 floats (72 at 700 m and 36 at 1500 m) at 50°N are shown in red, and the final positions are indicated with black dots. Colors represent the normalized temperature anomaly (°C) along the path of each float, computed following Bower et al. (2009): (T − Ti)/δTmax, where Ti is each float’s initial temperature, and δTmax is the maximum temperature difference: 3°C for floats launched at 700 m and 0.72°C for those launched at 1500 m. The 700-, 1500-, 3000-m isobaths are contoured in gray.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
To describe the pathways of the simulated float trajectories after the AR7W launch, floats are placed in four categories:
never exported—these floats circulated solely within the subpolar basin during their entire lifetime and did not cross either intergyre boundary;
recirculated—these floats crossed the subtropical boundary once or more, but they recirculated back and ended their mission in the subpolar gyre;
exported—these floats crossed the subtropical boundary and at the end of their mission were in the subtropical gyre (this category includes those that crossed only once during their lifetime, i.e., they never returned to the subpolar gyre, and those that were repeatedly exported, i.e., they crossed the intergyre boundaries several times before they ended their mission in the subtropical basin); and
in between—these floats ended their mission located between the subpolar and subtropical boundaries.
An example of the spatial distribution of each trajectory category is revealed by a probability map (Fig. 7) constructed from pathway positions for 40 years after the AR7W launch in 1990. Probability maps from other launch years yield qualitatively the same maps. In general, floats are confined north of 30°N during the 40 years of integration. At the end of those 40 years, 31% of them reside in the subpolar gyre: 17% circulated in the subpolar basin the entire time, while 14% returned to the subpolar basin after one or more exports to the subtropical gyre. Only 46% of the floats ended up in the subtropical gyre 40 years after launch, of which 7% crossed the intergyre boundaries only once and 17% ended up south of 30°N. The floats that reached the subtropical gyre did so by a number of pathways, including those in the central and eastern North Atlantic basin. The central basin pathways were described previously by recent studies focused on interior pathways (Bower et al. 2009; Lozier et al. 2013) of the subpolar to subtropical export. The eastern basin pathway is along the eastern flank of the Reykjanes Ridge, in agreement with prior observational studies (Kieke et al. 2009; Rhein et al. 2015). This pathway is similar to the southward pathway of Iceland–Scotland Overflow Water (ISOW) revealed from both observational and modeling studies (Lankhorst and Zenk 2006; Xu et al. 2010).

Probability maps of trajectories 40 years after launch on 1 Mar 1990: (a) never exported floats (17% of total), (b) recirculated floats (15%), (c) exported floats (42%), and (d) in-between floats (26%). The probability map is created by first dividing the North Atlantic into 0.25° × 0.25° grids, counting the number of times floats pass through each grid (including repetitions), and then dividing the number of passes in each grid by the total float passes over all grids (Gary et al. 2012). Black solid lines represent the intergyre boundaries. Black dots indicate final float positions; 700- and 1500-m isobaths are contoured in gray.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1

Probability maps of trajectories 40 years after launch on 1 Mar 1990: (a) never exported floats (17% of total), (b) recirculated floats (15%), (c) exported floats (42%), and (d) in-between floats (26%). The probability map is created by first dividing the North Atlantic into 0.25° × 0.25° grids, counting the number of times floats pass through each grid (including repetitions), and then dividing the number of passes in each grid by the total float passes over all grids (Gary et al. 2012). Black solid lines represent the intergyre boundaries. Black dots indicate final float positions; 700- and 1500-m isobaths are contoured in gray.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
Probability maps of trajectories 40 years after launch on 1 Mar 1990: (a) never exported floats (17% of total), (b) recirculated floats (15%), (c) exported floats (42%), and (d) in-between floats (26%). The probability map is created by first dividing the North Atlantic into 0.25° × 0.25° grids, counting the number of times floats pass through each grid (including repetitions), and then dividing the number of passes in each grid by the total float passes over all grids (Gary et al. 2012). Black solid lines represent the intergyre boundaries. Black dots indicate final float positions; 700- and 1500-m isobaths are contoured in gray.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
Repeating the above analysis for all 44 launches (1961–2004) reveals that there is relatively little variability over time (Fig. 8): on average, 34% of the total floats stayed within the subpolar basin 40 years after launch, 46% were able to reach the subtropics, and 20% stayed between the two boundaries. Despite the significant change in the number of floats launched each year, the fractions remain relatively constant (±7%) except for the years 1961, 1969, and 1971, when fewer than 10 floats were launched along AR7W, a paucity that indicates weak convective activity in the Labrador Sea during those years. We consider these floats too few to draw conclusions on their preferred pathways, but given the pathways from the other years, we feel confident in our assessment that there is little interannual variability in LSW pathways.

The percentage of floats in the subpolar gyre (blue), in the subtropical gyre (red), and in the area bounded by the intergyre boundaries (black) as a function of time after the launch across AR7W. The solid curves designate the percentage averaged over all 44 releases while the colored dots represent the percentages for each release.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1

The percentage of floats in the subpolar gyre (blue), in the subtropical gyre (red), and in the area bounded by the intergyre boundaries (black) as a function of time after the launch across AR7W. The solid curves designate the percentage averaged over all 44 releases while the colored dots represent the percentages for each release.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
The percentage of floats in the subpolar gyre (blue), in the subtropical gyre (red), and in the area bounded by the intergyre boundaries (black) as a function of time after the launch across AR7W. The solid curves designate the percentage averaged over all 44 releases while the colored dots represent the percentages for each release.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
The decrease (increase) in the percentage of floats in the subpolar (subtropical) gyre persists over the full 40 years, an indication of the long residence time for LSW in the subpolar basin. The percentage of floats between boundaries, however, remains relatively constant after 25 years, indicating a smaller residence time scale for this region.
c. The age of the exported LSW
So far, we have seen that only 46% of the floats launched along AR7W end up in the subtropical gyre after 40 years, which indicates a long advective time scale for LSW to reach the subtropical gyre. A calculation and mapping of LSW age confirms this expectation (Fig. 9, left). The youngest LSW is found in the Labrador and Irminger Seas, as well as along the DWBC, through which floats reach the subtropical gyre within 15 years (the fastest float took less than 3 years). The average age of floats at the intergyre boundary region is 22 ± 10 years, which is also evident in the cross-sectional plot of age along AR7W in Fig. 9 (right). As discussed above, the majority of the floats launched in the central basin along AR7W took 20 years on average to first cross the subtropical boundary. However, floats launched in the DWBC, which account for less than 5% of the total launched, took less than 15 years to be exported to the subtropical gyre.

(left) Average age of floats for 44 releases of 40 years: the whole domain is divided into 0.25° × 0.25° grids, and the age for each grid is computed by averaging the time elapsed since launch for each particle. Repeated visits by the same float to the same grid are included in the age calculation. To avoid biasing the average, only when the box has more than 100 float occurrences is the mean age computed. The black solid lines indicate the subpolar boundary (north) and the subtropical boundary (south); the black dashed line indicates 30°N. Initial launch locations are shown in red; 700-, 1500-, and 3000-m isobaths are shown in gray. (right) The average age of all floats from 44 releases that cross the subtropical boundary in 40 years as a function of initial position along AR7W. Black dashed contours show where climatological density is between [27.75, 27.84] kg m−3; white contours show the area with PV smaller than 4 × 10−12 m−1 s−1.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1

(left) Average age of floats for 44 releases of 40 years: the whole domain is divided into 0.25° × 0.25° grids, and the age for each grid is computed by averaging the time elapsed since launch for each particle. Repeated visits by the same float to the same grid are included in the age calculation. To avoid biasing the average, only when the box has more than 100 float occurrences is the mean age computed. The black solid lines indicate the subpolar boundary (north) and the subtropical boundary (south); the black dashed line indicates 30°N. Initial launch locations are shown in red; 700-, 1500-, and 3000-m isobaths are shown in gray. (right) The average age of all floats from 44 releases that cross the subtropical boundary in 40 years as a function of initial position along AR7W. Black dashed contours show where climatological density is between [27.75, 27.84] kg m−3; white contours show the area with PV smaller than 4 × 10−12 m−1 s−1.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
(left) Average age of floats for 44 releases of 40 years: the whole domain is divided into 0.25° × 0.25° grids, and the age for each grid is computed by averaging the time elapsed since launch for each particle. Repeated visits by the same float to the same grid are included in the age calculation. To avoid biasing the average, only when the box has more than 100 float occurrences is the mean age computed. The black solid lines indicate the subpolar boundary (north) and the subtropical boundary (south); the black dashed line indicates 30°N. Initial launch locations are shown in red; 700-, 1500-, and 3000-m isobaths are shown in gray. (right) The average age of all floats from 44 releases that cross the subtropical boundary in 40 years as a function of initial position along AR7W. Black dashed contours show where climatological density is between [27.75, 27.84] kg m−3; white contours show the area with PV smaller than 4 × 10−12 m−1 s−1.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
At 30°N, where floats are less likely to recirculate back to the subpolar gyre, the age is 30 ± 8 years. The age distribution along AR7W of the floats that reach 30°N is similar to that shown in Fig. 9 (right), but with greater ages. The spatial distribution of LSW age shown in Fig. 9 (left) is consistent with the age of simulated LSW from Gary et al. (2012). Using a CFC observational dataset, Rhein et al. (2015) show a similar LSW age distribution: their youngest waters are in the Labrador and Irminger Seas, age increases eastward from south of the Grand Banks to the Rockall Trough near the intergyre boundary region, and age in the western subtropical gyre increases steadily to the south. In Rhein et al. (2015), the age of young LSW (age < 40 years) is 16 years when it crosses the intergyre boundary and 22–24 years at 30°N. Both of these ages fall in the range derived from Lagrangian floats in this study.
Considering the long residence time for LSW in the subpolar basin, the volume export at the subtropical boundary or at 30°N for any particular year is expected to be composed of waters with many different ages. The results in Fig. 10 confirm this expectation. In 2003, floats exported across the subtropical boundary, as well as floats arriving at 30°N, are of various ages and, importantly, there is no distinguishable difference in the number of floats from one age to the next, even though the number of the floats launched varies significantly with time. Also, the number of floats from any given launch year that contributes to the 2003 export is quite small (~2%) compared to the number of floats launched each year, which, as discussed above, is a measure of the convective strength in the Labrador Sea. Similar results are observed in other years’ export. In the Eulerian frame, the vertical diapycnal mass flux in the Labrador Sea has been estimated to be ~2 Sv (Pickart and Spall 2007); thus, the contribution of the transformed water mass to the amount of LSW exported each year is 0.04 Sv, a negligible quantity compared to the 11.3 ± 1.0 Sv of LSW within DWBC at 53°N (Fischer et al. 2010). In other words, from both an Eulerian and Lagrangian perspective, convection strength in the Labrador Sea and newly formed LSW export are not tightly coupled, as discussed further below.

Histogram of the number of floats from each launch year exported south of the subtropical boundary (light blue bars) and 30°N (dark blue bars) in 2003. Initial launch number is plotted with a black solid line; its average is denoted by the red dashed line. On average, only 2% of floats released each year along AR7W contribute to the 2003 LSW export across the subtropical boundary.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1

Histogram of the number of floats from each launch year exported south of the subtropical boundary (light blue bars) and 30°N (dark blue bars) in 2003. Initial launch number is plotted with a black solid line; its average is denoted by the red dashed line. On average, only 2% of floats released each year along AR7W contribute to the 2003 LSW export across the subtropical boundary.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
Histogram of the number of floats from each launch year exported south of the subtropical boundary (light blue bars) and 30°N (dark blue bars) in 2003. Initial launch number is plotted with a black solid line; its average is denoted by the red dashed line. On average, only 2% of floats released each year along AR7W contribute to the 2003 LSW export across the subtropical boundary.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
d. Convection in the Labrador Sea and the export of blended LSW to the subtropics
The above analyses make clear that there is no coherent arrival of LSW with the same age. Thus, to further investigate the relationship between LSW production and its export, we no longer identify floats by their launch year in the Labrador Sea. Instead, we seek to understand the relationship between LSW production and the blended LSW export to the subtropical gyre, regardless of age. To identify the locations of blended LSW near the subtropical boundary, we locate all AR7W launched floats when they are crossing the boundary (Fig. 11). In Fig. 11, floats are concentrated within the area where PV is <12 × 10−12 m−1 s−1 and density is between [27.75, 27.84] kg m−3 west of the Mid-Atlantic Ridge. These property ranges are chosen as the criteria for the new float launches into blended LSW.

Positions of simulated floats launched in the Labrador Sea when they crossed the subtropical boundary in 2003 (black dots), with annually averaged PV (m−1 s−1) shown in color. White contours outline the area where PV is smaller than 12 × 10−12 m−1 s−1, and the black dashed line indicates density between [27.75, 27.84] kg m−3. When they reached the subtropical boundary, over 70% of floats (launched from 1961 to 2002) were located within the area bounded by the two density contours. The PV map and float distributions shown here for 2003 are representative of other years.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1

Positions of simulated floats launched in the Labrador Sea when they crossed the subtropical boundary in 2003 (black dots), with annually averaged PV (m−1 s−1) shown in color. White contours outline the area where PV is smaller than 12 × 10−12 m−1 s−1, and the black dashed line indicates density between [27.75, 27.84] kg m−3. When they reached the subtropical boundary, over 70% of floats (launched from 1961 to 2002) were located within the area bounded by the two density contours. The PV map and float distributions shown here for 2003 are representative of other years.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
Positions of simulated floats launched in the Labrador Sea when they crossed the subtropical boundary in 2003 (black dots), with annually averaged PV (m−1 s−1) shown in color. White contours outline the area where PV is smaller than 12 × 10−12 m−1 s−1, and the black dashed line indicates density between [27.75, 27.84] kg m−3. When they reached the subtropical boundary, over 70% of floats (launched from 1961 to 2002) were located within the area bounded by the two density contours. The PV map and float distributions shown here for 2003 are representative of other years.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
Floats were then released on each 5 January north of the subtropical boundary (Fig. 12) from 1961 to 2004, and the blended LSW export for each year is the total number of floats that crossed the subtropical boundary by the end of that year. We consider this time series an indication of export variability for blended LSW, rather than LSW linked to water mass formation in a given year. Similar dynamic launches were conducted at 23°N, where southward-moving floats are concentrated within the DWBC and between isopycnals of [27.75, 27.84] kg m−3 (Fig. 13). Again, the time series of blended LSW export to the south of 23°N is derived from the number of floats that crossed 23°N each year.

Initial (blue dots) and final (red dots) locations of floats launched on 5 Jan 2003 that were able to reach the subtropical basin within 1 year. This float distribution map differs only slightly from year to year. One-tenth of the total data points were randomly selected for plotting. Black solid curve indicates the subtropical boundary. The black dashed box (north of the subtropical boundary) indicates the dynamic launch area: longitude [75°W, 26°W], latitude [35°N, 50°N], PV [0, 12 × 10−12] m−1 s−1, and density [27.75, 27.84] kg m−3. Only a few floats launched outside the box made it to the subtropics within a year. The 700-, 1500-, and 3000-m isobaths are contoured in gray.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1

Initial (blue dots) and final (red dots) locations of floats launched on 5 Jan 2003 that were able to reach the subtropical basin within 1 year. This float distribution map differs only slightly from year to year. One-tenth of the total data points were randomly selected for plotting. Black solid curve indicates the subtropical boundary. The black dashed box (north of the subtropical boundary) indicates the dynamic launch area: longitude [75°W, 26°W], latitude [35°N, 50°N], PV [0, 12 × 10−12] m−1 s−1, and density [27.75, 27.84] kg m−3. Only a few floats launched outside the box made it to the subtropics within a year. The 700-, 1500-, and 3000-m isobaths are contoured in gray.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
Initial (blue dots) and final (red dots) locations of floats launched on 5 Jan 2003 that were able to reach the subtropical basin within 1 year. This float distribution map differs only slightly from year to year. One-tenth of the total data points were randomly selected for plotting. Black solid curve indicates the subtropical boundary. The black dashed box (north of the subtropical boundary) indicates the dynamic launch area: longitude [75°W, 26°W], latitude [35°N, 50°N], PV [0, 12 × 10−12] m−1 s−1, and density [27.75, 27.84] kg m−3. Only a few floats launched outside the box made it to the subtropics within a year. The 700-, 1500-, and 3000-m isobaths are contoured in gray.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1

As in Fig. 12, but for blended LSW across 23°N (black dashed line). The inset shows the positions of those AR7W launched floats when crossing 23°N, superposed on the 44-yr mean meridional velocity, with black contours outlining the density between [27.75, 27.84] kg m−3. This distribution sets the criteria for the blended LSW launch north of 23°N (PV at this location is not distinguishable and thus is not used to identify LSW). Again, the float distribution map and velocity field are representative of all other years.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1

As in Fig. 12, but for blended LSW across 23°N (black dashed line). The inset shows the positions of those AR7W launched floats when crossing 23°N, superposed on the 44-yr mean meridional velocity, with black contours outlining the density between [27.75, 27.84] kg m−3. This distribution sets the criteria for the blended LSW launch north of 23°N (PV at this location is not distinguishable and thus is not used to identify LSW). Again, the float distribution map and velocity field are representative of all other years.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
As in Fig. 12, but for blended LSW across 23°N (black dashed line). The inset shows the positions of those AR7W launched floats when crossing 23°N, superposed on the 44-yr mean meridional velocity, with black contours outlining the density between [27.75, 27.84] kg m−3. This distribution sets the criteria for the blended LSW launch north of 23°N (PV at this location is not distinguishable and thus is not used to identify LSW). Again, the float distribution map and velocity field are representative of all other years.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
Figure 14 (top) shows the lead–lag correlation coefficients between the convection strength in the Labrador Sea, given by the float number launched each year along AR7W, and the blended LSW export to the subtropical gyre. The coefficient reaches a maximum of 0.63 with the blended LSW export lagging by 3 years, yet after detrending, no significant correlation is observed. A maximum correlation coefficient of 0.55 is found between convection and the export volume across 23°N at 0 lag–lead; a 0.38 coefficient is computed after detrending the two time series, with convection leading the export by 3 years.

(top) Cross-correlation coefficients as a function of lead–lag for the launched float number anomalies along AR7W and exported float number anomalies across the subtropical boundary. Black solid line indicates the coefficient before detrending and the dashed line indicates the value after detrending. The values at 95% confidence level are shaded with light gray. Negative values along the x axis indicate that convection is leading export. (bottom) As in the top panel, but for the float number anomalies launched and exported number anomalies across 23°N.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1

(top) Cross-correlation coefficients as a function of lead–lag for the launched float number anomalies along AR7W and exported float number anomalies across the subtropical boundary. Black solid line indicates the coefficient before detrending and the dashed line indicates the value after detrending. The values at 95% confidence level are shaded with light gray. Negative values along the x axis indicate that convection is leading export. (bottom) As in the top panel, but for the float number anomalies launched and exported number anomalies across 23°N.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
(top) Cross-correlation coefficients as a function of lead–lag for the launched float number anomalies along AR7W and exported float number anomalies across the subtropical boundary. Black solid line indicates the coefficient before detrending and the dashed line indicates the value after detrending. The values at 95% confidence level are shaded with light gray. Negative values along the x axis indicate that convection is leading export. (bottom) As in the top panel, but for the float number anomalies launched and exported number anomalies across 23°N.
Citation: Journal of Physical Oceanography 46, 7; 10.1175/JPO-D-15-0210.1
Since the advective time scale for floats from AR7W to the subtropical gyre (23°N) is on average 22 (30) years, the high correlations (before detrending) are not considered to result from a causal relationship between convection strength and LSW advective export. Rather, we suggest that the fast response of blended LSW export to LSW formation results from a boundary density anomaly induced by convection in the Labrador Sea, the signal of which can be propagated southward quickly though boundary waves. This supposition is supported by a modeling study that showed an overall strengthening of the AMOC within 1–2 years after LSW convection (Biastoch et al. 2008). Also demonstrated in this study is that the AMOC response to convection is primarily on decadal time scales. The fact that there are negligible correlations after detrending (which dampens decadal variability) also validates our assessment that convective variability in the Labrador Sea cannot explain the downstream variability of LSW export through advection.
4. Summary
In this study, we use trajectories of synthetic floats launched in the Labrador Sea in an ocean general circulation model to simulate the spreading pathways of newly formed LSW. We show that only 46% ± 7% of the LSW formed during winter is able to reach the subtropical boundary after 40 years. The rest of the water mass largely recirculates within the subpolar gyre or is resident in the area between the intergyre boundaries. The exported floats primarily enter the subtropics via interior pathways that extend from the western boundary to the eastern basin of the North Atlantic, though not all of those pathways indicate a direct route for export. Some floats that cross into the subtropical gyre recirculate back to the subpolar basin more than once before ending up in the subtropical gyre, which lengthens the average time scale for the floats to be exported. The mean age of floats when they first reach the subtropical boundary is 22 ± 10 years and it takes 30 ± 8 years for them to reach 30°N. The youngest floats, with ages less than 15 years, are those that originate or travel near the western boundary, yet these floats account for less than 5% of the total.
An analysis of the age of LSW when it crosses into the subtropical gyre shows that it is a combination of waters formed years or even decades prior to the year of the crossing. We show that floats launched in a particular winter contribute only marginally to future LSW volume exports. Thus, we conclude that the contribution of a particular winter’s convection to the total LSW export in any given subsequent year is too small to appreciably impact the volume of that export.
We extend our analysis to include blended LSW export, by which the water mass is defined by its hydrographic properties only, not by its age. We find no linkage between LSW formation and the export of blended LSW across the subtropical boundary or 30°N through advection. Rather, the water mass export in this layer appears to respond to Labrador Sea convection via fast western boundary waves with a time lag of no more than 1–2 years.
The relatively long time for LSW to reach 30°N (30 years on average) stands in contrast to the arrival time of LSW from the Labrador basin to the subtropical gyre based on property correlations: Curry et al. (1998) show a high correlation between LSW thickness and temperature anomalies at 32°N near Bermuda with the former leading by only 6 years, while van Sebille et al. (2011) estimate that LSW reaches 26°N (at Abaco) in 9 years based on classical LSW salinity anomalies. Thus, left unanswered in our study is the question as to how property signals observed in the Labrador Sea are transmitted to the subtropical gyre in such a relatively short time scale, if not through advection. This question forms the basis for a future study.
Acknowledgments
The authors gratefully acknowledge the U.S. National Science Foundation (Award OCE-1259103) for the support of this work. S. Zou thanks the ARIANE group for trajectory computation code, A. Biastoch and C. Böning for ORCA model data, S. Gary for code and suggestions, and N. Foukal and F. Li for helpful discussions.
REFERENCES
Barnier, B., and Coauthors, 2006: Impact of partial steps and momentum advection schemes in a global ocean circulation model at eddy permitting resolution. Ocean Dyn., 56, 543–567, doi:10.1007/s10236-006-0082-1.
Barnier, B., and Coauthors, 2007: Eddy-permitting ocean circulation hindcasts of past decades. CLIVAR Exchanges, Vol. 12, No. 3, International CLIVAR Project Office, Southampton, United Kingdom, 8–10.
Berliand, M. E., and T. G. Strokina, 1980: Global Distribution of the Total Amount of Clouds. Hydrometeorological Publishing House, 71 pp.
Biastoch, A., C. W. Böning, J. Getzlaff, J. M. Molines, and G. Madec, 2008: Causes of interannual-decadal variability in the meridional overturning circulation of the midlatitude North Atlantic Ocean. J. Climate, 21, 6599–6615, doi:10.1175/2008JCLI2404.1.
Blanke, B., and N. Grima, 2010: ARIANE v2.2.6. Laboratoire de Physique des Oceans. [Available online at http://www.univ-brest.fr/lpo/ariane/.]
Böning, C. W., M. Scheinert, J. Dengg, A. Biastoch, and A. Funk, 2006: Decadal variability of subpolar gyre transport and its reverberation in the North Atlantic overturning. Geophys. Res. Lett., 33, L21S01, doi:10.1029/2006GL026906.
Bower, A. S., M. S. Lozier, S. F. Gary, and C. W. Böning, 2009: Interior pathways of the North Atlantic meridional overturning circulation. Nature, 459, 243–247, doi:10.1038/nature07979.
CDIAC, 2015: CLIVAR repeat section AR07W. Carbon Dioxide International Analysis Center. Subset used: 1992–2004, accessed 29 January 2015. [Available online at http://cdiac.ornl.gov/oceans/RepeatSections/clivar_ar07w.html.]
CERSAT, 2002: Mean Wind Fields (MWF product) volume 1—ERS-1, ERS-2 & NSCAT user manual. Rep. C2-MUT-W-05-IF, Department of Oceanography from Space, Ifremer, 54 pp.
Curry, R. G., M. S. McCartney, and T. M. Joyce, 1998: Oceanic transport of subpolar climate signals to mid-depth subtropical waters. Nature, 391, 575–577, doi:10.1038/35356.
Dengler, M., J. Fischer, F. A. Schott, and R. Zantopp, 2006: Deep Labrador Current and its variability in 1996–2005. Geophys. Res. Lett., 33, L21S06, doi:10.1029/2006GL026702.
Fischer, J., M. Visbeck, R. Zantopp, and N. Nunes, 2010: Interannual to decadal variability of outflow from the Labrador Sea. Geophys. Res. Lett., 37, L24610, doi:10.1029/2010GL045321.
Gary, S. F., M. S. Lozier, C. W. Böning, and A. Biastoch, 2011: Deciphering the pathways for the deep limb of the meridional overturning circulation. Deep-Sea Res. II, 58, 1781–1797, doi:10.1016/j.dsr2.2010.10.059.
Gary, S. F., M. S. Lozier, A. Biastoch, and C. W. Böning, 2012: Reconciling tracer and float observations of the export pathways of Labrador Sea Water. Geophys. Res. Lett., 39, L24606, doi:10.1029/2012GL053978.
Getzlaff, J., C. W. Böning, C. Eden, and A. Biastoch, 2005: Signal propagation related to the North Atlantic overturning. Geophys. Res. Lett., 32, L09602, doi:10.1029/2004GL021002.
Goosse, H., 1997: Modeling the large scale behaviour of the coupled ocean-sea ice system. Ph.D. thesis, Université Catholique de Louvain, 231 pp.
Han, G., K. Ohashi, N. Chen, P. G. Myers, N. Nunes, and J. Fischer, 2010: Decline and partial rebound of the Labrador Current 1993–2004: Monitoring ocean currents from altimetric and conductivity‐temperature‐depth data. J. Geophys. Res., 115, C12012, doi:10.1029/2009JC006091.
Jourdan, D., E. Balopoulos, M.-J. Garcia-Fernandez, and C. Maillard, 1998: Objective analysis of temperature and salinity historical data set over the Mediterranean basin. OCEANS ‘98 Conference Proceedings, Vol. 1, IEEE, 82–87, doi: 10.1109/OCEANS.1998.725649.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437–471, doi:10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2.
Kieke, D., and I. Yashayaev, 2015: Studies of Labrador Sea Water formation and variability in the subpolar North Atlantic in the light of international partnership and collaboration. Prog. Oceanogr., 132, 220–232, doi:10.1016/j.pocean.2014.12.010.
Kieke, D., M. Rhein, L. Stramma, W. M. Smethie, J. L. Bullister, and D. A. LeBel, 2007: Changes in the pool of Labrador Sea Water in the subpolar North Atlantic. Geophys. Res. Lett., 34, L06605, doi:10.1029/2006GL028959.
Kieke, D., B. Klein, L. Stramma, M. Rhein, and K. P. Koltermann, 2009: Variability and propagation of Labrador Sea Water in the southern subpolar North Atlantic. Deep-Sea Res. I, 56, 1656–1674, doi:10.1016/j.dsr.2009.05.010.
Knauss, J. A., 1997: Introduction to Physical Oceanography. Waveland Press, Inc., 338 pp.
Lankhorst, M., and W. Zenk, 2006: Lagrangian observations of the middepth and deep velocity fields of the northeastern Atlantic Ocean. J. Phys. Oceanogr., 36, 43–63, doi:10.1175/JPO2869.1.
Lavender, K. L., W. B. Owens, and R. E. Davis, 2005: The mid-depth circulation of the subpolar North Atlantic Ocean as measured by subsurface floats. Deep-Sea Res. I, 52, 767–785, doi:10.1016/j.dsr.2004.12.007.
Levitus, S., and Coauthors, 1998: Introduction. Vol. 1, World Ocean Database 1998, NOAA Atlas NESDIS 18, 346 pp.
Lozier, M. S., 2012: Overturning in the North Atlantic. Annu. Rev. Mar. Sci., 4, 291–315, doi:10.1146/annurev-marine-120710-100740.
Lozier, M. S., S. F. Gary, and A. S. Bower, 2013: Simulated pathways of the overflow waters in the North Atlantic: Subpolar to subtropical export. Deep-Sea Res. II, 85, 147–153, doi:10.1016/j.dsr2.2012.07.037.
Molinari, R. L., R. A. Fine, W. D. Wilson, R. G. Curry, J. Abell, and M. S. McCartney, 1998: The arrival of recently formed Labrador sea water in the Deep Western Boundary Current at 26.5°N. Geophys. Res. Lett., 25, 2249–2252, doi:10.1029/98GL01853.
NOAA/NWS/CPC, 2015: North Atlantic Oscillation (NAO). NOAA/National Weather Service/Climate Prediction Center, subset used: December 1960–March 2004. [Available online at http://www.cpc.ncep.noaa.gov/products/precip/CWlink/pna/nao.shtml#publication.]
Pena-Molino, B., T. M. Joyce, and J. M. Toole, 2011: Recent changes in the Labrador Sea Water within the deep western boundary current southeast of Cape Cod. Deep-Sea Res. I, 58, 1019–1030, doi:10.1016/j.dsr.2011.07.006.
Pickart, R. S., and M. A. Spall, 2007: Impact of Labrador Sea convection on the North Atlantic meridional overturning circulation. J. Phys. Oceanogr., 37, 2207–2227, doi:10.1175/JPO3178.1.
Pickart, R. S., D. J. Torres, and R. A. Clarke, 2002: Hydrography of the Labrador Sea during active convection. J. Phys. Oceanogr., 32, 428–457, doi:10.1175/1520-0485(2002)032<0428:HOTLSD>2.0.CO;2.
Rhein, M., and Coauthors, 2002: Labrador Sea Water: Pathways, CFC inventory, and formation rates. J. Phys. Oceanogr., 32, 648–665, doi:10.1175/1520-0485(2002)032<0648:LSWPCI>2.0.CO;2.
Rhein, M., and Coauthors, 2011: Deep water formation, the subpolar gyre, and the meridional overturning circulation in the subpolar North Atlantic. Deep-Sea Res. II, 58, 1819–1832, doi:10.1016/j.dsr2.2010.10.061.
Rhein, M., D. Kieke, and R. Steinfeldt, 2015: Advection of North Atlantic Deep Water from the Labrador Sea to the southern hemisphere. J. Geophys. Res. Oceans, 120, 2471–2487, doi:10.1002/2014JC010605.
Sarafanov, A., 2009: On the effect of the North Atlantic Oscillation on temperature and salinity of the subpolar North Atlantic intermediate and deep waters. ICES J. Mar. Sci., 66, 1448–1454, doi:10.1093/icesjms/fsp094.
Schott, F. A., R. Zantopp, L. Stramma, M. Dengler, J. Fischer, and M. Wibaux, 2004: Circulation and deep-water export at the western exit of the subpolar North Atlantic. J. Phys. Oceanogr., 34, 817–843, doi:10.1175/1520-0485(2004)034<0817:CADEAT>2.0.CO;2.
Steele, M., R. Morley, and W. Ermold, 2001: PHC: A global ocean hydrography with a high-quality Arctic Ocean. J. Climate, 14, 2079–2087, doi:10.1175/1520-0442(2001)014<2079:PAGOHW>2.0.CO;2.
Stramma, L., D. Kieke, M. Rhein, F. Schott, I. Yashayaev, and K. P. Koltermann, 2004: Deep water changes at the western boundary of the subpolar North Atlantic during 1996 to 2001. Deep-Sea Res. I, 51, 1033–1056, doi:10.1016/j.dsr.2004.04.001.
Sy, A., M. Rhein, J. Lazier, K. P. Koltermann, J. Meincke, A. Putzka, and M. Bersch, 1997: Surprisingly rapid spreading of newly formed intermediate waters across the North Atlantic Ocean. Nature, 386, 675–679, doi:10.1038/386675a0.
Talley, L. D., and M. S. McCartney, 1982: Distribution and circulation of Labrador Sea Water. J. Phys. Oceanogr., 12, 1189–1205, doi:10.1175/1520-0485(1982)012<1189:DACOLS>2.0.CO;2.
Trenberth, K. E., J. G. Olson, and W. G. Large, 1989: A global ocean wind stress climatology based on ECMWF analyses. NCAR Tech. Note NCAR/TN-338+STR, 93 pp., doi:10.5065/D6ST7MR9.
van Aken, H. M., M. F. De Jong, and I. Yashayaev, 2011: Decadal and multi-decadal variability of Labrador Sea Water in the north-western North Atlantic Ocean derived from tracer distributions: Heat budget, ventilation, and advection. Deep-Sea Res. I, 58, 505–523, doi:10.1016/j.dsr.2011.02.008.
van Sebille, E., M. O. Baringer, W. E. Johns, C. S. Meinen, L. M. Beal, M. F. de Jong, and H. M. van Aken, 2011: Propagation pathways of classical Labrador Sea water from its source region to 26°N. J. Geophys. Res., 116, C12027, doi:10.1029/2011JC007171.
Xie, P., and P. A. Arkin, 1997: Global precipitation: A 17-year monthly analysis based on gauge observations, satellite estimates, and numerical model outputs. Bull. Amer. Meteor. Soc., 78, 2539–2558, doi:10.1175/1520-0477(1997)078<2539:GPAYMA>2.0.CO;2.
Xu, X., W. J. Schmitz, H. E. Hurlburt, P. J. Hogan, and E. P. Chassignet, 2010: Transport of Nordic Seas overflow water into and within the Irminger Sea: An eddy-resolving simulation and observations. J. Geophys. Res., 115, C12048, doi:10.1029/2010JC006351.
Yashayaev, I., 2007: Hydrographic changes in the Labrador Sea, 1960–2005. Prog. Oceanogr., 73, 242–276, doi:10.1016/j.pocean.2007.04.015.
Yashayaev, I., and J. W. Loder, 2009: Enhanced production of Labrador Sea Water in 2008. Geophys. Res. Lett., 36, L01606, doi:10.1029/2008GL036162.