Recent thermohaline changes in the layer of intermediate waters (IW) advected into the eastern subpolar North Atlantic from lower latitudes are quantified using the data from the repeated transatlantic sections. Positive trends in temperature and salinity in the IW density class at ∼53°N (0.049°C/a and 0.0088/a, 1992–2002) and ∼60°N (0.044°C/a and 0.0085/a, 1997–2005) are derived. The unexpectedly high rates of the IW warming and salinification cannot be explained solely by the long-term and recent decadal changes at the intermediate levels in the midlatitude North Atlantic and appear to be a consequence of the northward advance of the source water masses caused by the North Atlantic Oscillation-induced contraction of the subpolar gyre.
 Most studies of temperature (potential temperature, θ) and salinity (S) changes at the intermediate and deep levels in the subpolar North Atlantic (NA) are focused on water masses formed at the northern periphery of the region, the Labrador Sea Water (LSW) and Nordic overflow-derived deep waters. The long-term freshening of the deep waters along with nonsteady freshening of LSW over the last 3–4 decades of the 20th century [Dickson et al., 2002] and the subsequent rapid warming and salinification of these waters in the mid-1990s to mid-2000s [Sarafanov et al., 2007] have been documented. The θ and S changes of intermediate waters (∼500–1250 m) arriving to the eastern subpolar region from the subtropical NA, are less well investigated. Quantification of these changes in the 1990s–2000s on the basis of the data from the repeated hydrographic sections (Figure 1) constitutes the subject of our study.
 Water masses formed by wintertime convection in the northern NA and subsequently modified by along-path mixing are rich in oxygen due to their recent contact with the atmosphere. In the eastern subpolar NA, relatively old oxygen-poor intermediate waters of southern origin conveyed into the region in the permanent pycnocline (σ0 < 27.7), are clearly distinguishable by the low oxygen concentrations (∼200–250 μmol/kg) from the oxygenated (>250 μmol/kg) waters of subpolar origin (Figure 2a), namely, the overlying Subpolar Mode Water (SPMW) and underlying LSW [see van Aken and de Boer, 1995; van Aken and Becker, 1996].
 The intermediate oxygen minimum layer in the NA eastern subpolar region has been tentatively designated “Intermediate Water” (IW) by van Aken and de Boer , and the following waters are suggested to be the main contributors to this layer [see van Aken and Becker, 1996]: the Mediterranean Overflow-derived Water (MOW) advected northward along the European slope from the Gulf of Cadiz into the Rockall Trough [Reid, 1979] and the Antarctic Intermediate Water (AAIW) brought to the region by the North Atlantic Current [Tsuchiya et al., 1992; Álvarez et al., 2004]. In the eastern subpolar NA, north of ∼50°N, these waters appear in a strongly diluted form in the ∼27.5–27.6 density range [see Tsuchiya et al., 1992] corresponding to the oxygen minimum layer, and only the MOW traces can be identified there in the θ–S space [see van Aken and Becker, 1996]. Following van Aken and de Boer  and van Aken and Becker , we use the conditional general term “IW” for the low-oxygen waters of southern origin in the eastern subpolar NA assuming that these waters represent a mixture of the two mentioned water types with the subpolar waters, though a contribution of each of them to the IW layer is uncertain. While it is not the main objective of this study, a brief discussion of the IW layer structure is given in section 4.
 On the basis of the application of isopycnal analysis to the high precision CTD and oxygen data from the repeated transatlantic sections in the subpolar NA (Figure 1), we provide a quantification of the recent changes in the IW θ and S. Using the CTD-O2 and bottle oxygen data collected at ∼60°N (1997–2007) and ∼53°N (1990–2002), we defined the IW density class and calculated the mean (layer-averaged) IW θ and S values for each repeat of the section. The revealed changes are discussed in the context of warming of intermediate waters in the subtropical NA and the recent circulation changes in the subpolar region.
 In the eastern midlatitude NA domain (30–40°N, east of 20°W) adjoining the Gulf of Cadiz, θ and S in the intermediate salinity maximum associated with the MOW core increased by 0.10°C/decade and 0.028/decade, respectively, from 1955 to 1993 [Potter and Lozier, 2004]. In the regions more distant from the MOW source, the magnitudes of the long-term θ and S increase in the intermediate layer are smaller. In the repeated zonal sections across the subtropical and tropical NA, the 1000–2000 m layer, influenced by MOW and AAIW, warmed at rates of about 0.05°C/decade between the late 1950s and the 1980s to early 1990s [Arbic and Owens, 2001]. In the western subtropical NA (20–35°N, 52–66°W), this warming was accompanied by the water mass salinification (0.005–0.007/decade, 1954–1997) [see Joyce et al., 1999, Figure 11]. Recent estimates based on the 1957–2004 data from the 24.5°N section across the subtropical NA [Cunningham and Alderson, 2007] showed the 5-decade θ increase in the intermediate layer (900–1750 dbar) by 0.065°C/decade and 0.014°C/decade in the western and eastern basins, respectively, and the net 1957–2004 warming was dominated by isopycnal heave; the corresponding net S change was relatively small, being most prominent in the western basin (0.004/decade on isobars and 0.002/decade along isopycnals).
 A larger θ and S increase in the MOW layer occurred at higher latitudes in the eastern NA during the 1990s to early 2000s. In the Bay of Biscay, the rates of the density-compensated warming and salinification at the MOW levels (27.5 ≤ σ0 ≤ 27.6) were 0.023°C/a and 0.005/a in 1995–2003 [González-Pola et al., 2005]. Isopycnal analysis applied by Johnson and Gruber  to the 20°W section data has shown a large θ and S increase throughout the permanent pycnocline (up to ∼1°C and ∼0.15 in the MOW-induced S maximum, see their Figure 13) at 45–55°N between 1993 and 2003. This θ–S change has been attributed to an increased influence of MOW during the post-1995 period of below average North Atlantic Oscillation (NAO) index. Note that the latter estimate is based on the three section repeats, and the analysis of the data from more frequent observations performed in this study suggests smaller uncertainties resulting from aliasing.
3. Data and Method
 We used the CTD and oxygen data from the seven repeats of the zonal transatlantic section along 59.5–60°N (hereinafter, the 60°N section) carried out by the Russian RVs in 1997–2007 (Figure 1, Table 1). The accuracy of the temperature/salinity data was 0.003°C/0.003 in 1997–2003 and 0.001°C/0.002 in 2004–2007, see details in the work of Sarafanov et al. . Additionally, in order to examine the IW θ and S changes at the upstream location, at 52–53°N (hereinafter, ∼53°N), we used the CTD and oxygen data from the five repeats of the WOCE A1E line (1990–1996) and two repeats of the ∼53°N transatlantic section (2001 and 2002) (Figure 1, Table 1). The temperature/salinity accuracy is 0.002°C/0.002 for the A1E sections and 0.003°C/0.003 for the ∼53°N ones.
Table 1. Sections Used in the Study and the IW Layer Characteristicsa
Research Vessel, Cruise
Number of Stations
Mean θ, °C
Standard Deviation of θ, °C
Standard Deviation of S
Number of stations is the number of stations from each cruise used for computation of the distance-weighted means of potential temperature (mean θ, °C) and salinity (mean S) in the IW layer. Standard deviations of θ and S are standard deviations within the layer, a measure of the layer heterogeneity.
Professor Shtokman, 36
Akademik Mstislav Keldysh, 48
Akademik Ioffe, 13
Akademik Ioffe, 15
Akademik Ioffe, 18
Akademik Ioffe, 21
Akademik Ioffe, 23
A1E Line, 52–53°N Sections
Akademik Ioffe, 9
Akademik Mstislav Keldysh, 48
 During most of the cruises, oxygen concentrations were derived from the water samples, and in 2007 at 60°N, the oxygen profiles were obtained at each station using the Sea-Bird oxygen sensor module. The oxygen data uncertainties are about 2 μmol/kg for the bottle data and better then 5 μmol/kg for the CTD-O2 data calibrated with the oxygen concentrations in samples.
 In the first step, we defined the density limits of the IW layer and its conditional longitudinal margin (see section 4). Since the main characteristic of IW is the intermediate oxygen minimum, the IW density class (27.45 ≤ σ0 ≤ 27.65) was determined from inspection of vertical oxygen distributions and the θ–S–O2 plots (section 4). Then, we calculated the average values of θ and S in the IW density class for each section repeat as the distance-weighted means of the CTD data within the defined limits. This water mass-focused technique previously applied for quantification of the θ–S changes in the LSW and deep water layers at 60°N [Sarafanov et al., 2007], allows one to detect the water mass θ–S changes excluding changes due to isopycnal heave. The calculation results are summarized in Table 1 and presented in the form of time series in Figure 3.
4. IW Layer and Its Zonal Structure
 In the 60°N section, IW is clearly identified as the oxygen minimum layer (∼215–250 μmol/kg) that extends between the Scottish slope and the Reykjanes Ridge at depths of ∼500–1250 m (Figure 2a). The density distribution along the section superimposed on the oxygen pattern together with the θ–S–O2 plot constructed for the 2007 CTD-O2 data (Figures 2a and 2d) show that east of 25–28°W, IW occupies the 27.45 ≤ σ0 ≤ 27.65 density range. The θ–S–O2 plot based on the 1997–2006 bottle oxygen data (Figure 2e) confirms that the chosen density range is relevant to the entire 1997–2007 time period.
 We chose the 25°W longitude as the conditional western margin of IW. West of ∼25°W in all the 60°N section repeats, the σ0 = 27.45 isopycnal rises to depths of less than 500 m indicating a rapid westward transition to lower temperatures at the IW depths, and the upper part of the defined density layer includes more oxygenated (>250 μmol/kg) subpolar waters (Figures 2a and 2b). Inspection of the data collected at ∼53°N showed that the chosen IW limits are applicable for this location as well. Thus, IW is defined in this study as the 27.45 ≤ σ0 ≤ 27.65 layer east of 25°W.
 The most prominent zonal feature of IW is an eastward increase of salinity (Figures 2c). The highest vertically averaged salinities (∼35.15–35.25) in the IW layer at 60°N are found east of the Rockall-Hatton Plateau. This indicates an increased contribution of strongly diluted MOW arriving from the Rockall Trough. A similar but more prominent feature is observed upstream at ∼53°N. West of ∼13°W at 60°N, the IW salinity decreases westward to 35.05–35.15 (Figures 2c), and this decrease is accompanied by the drop of the mean oxygen concentrations from 230 to 250 to 220–230 μmol/kg (Figure 2b). The westward decrease in the IW salinity could be considered as a result of the MOW mixing with fresher subpolar waters, while the lower oxygen concentrations cannot be attributed to such mixing and require a contribution of less-oxygenated waters, older than MOW. Most likely, these old oxygen-poor waters are the strongly modified intermediate waters of Antarctic origin, a penetration of which into the eastern subpolar region has been previously documented [see Tsuchiya et al., 1992; Álvarez et al., 2004].
5. IW Temperature and Salinity Changes and Apparent Trends
 At ∼53°N, the IW temperature and salinity did not change much in 1990–1992 and increased by 0.67°C and 0.121, respectively, over the following decade (1992–2002). It should be noted that east of 25°W, the stations of the 2001 and 2002 ∼53°N sections are located, on average, 70 km south of the A1E line, and this influences the 1996–2002 comparison results for the location to some extent. However, the meridional gradients of the mean θ and S in the IW layer estimated from differences of the IW mean θ and S between 60°N and ∼53°N in 2002, are 0.014°C/70 km and 0.0045/70 km, respectively, and thus are relatively small, being an order of magnitude (26 and 16 times) less than the 1996–2002 increments of the IW θ and S (0.375°C and 0.070).
 Downstream, in the 60°N section, the IW θ/S steadily increased by 0.35°C/0.064 for 8 years from 1997 to 2005 and decreased during the following 2 years by 0.156°C/0.018 (Table 1, Figure 3).
Figure 2c based on the 1997, 2005, and 2007 60°N salinity data, shows that the revealed changes affected the entire IW layer at this latitude: the 1997–2005 salinification and the subsequent 2005–2007 freshening in the layer are seen at almost all longitudes east of 25°W; the same is true for the IW θ changes at 60°N and for the 1992–2002 θ and S increase in the IW layer at ∼53°N (no figure shown).
 The linear regression fits constructed for the overlapping time periods of sustained warming and salinification of IW at ∼53°N (1992–2002) and 60°N (1997–2005) show the significant positive trends (coefficients of determination are 0.73 ≤ r2 ≤ 0.96, see Figure 3) with similar magnitudes at the two locations. At ∼53°N, the IW layer has been warming and salinifying at rates of 0.049 ± 0.014°C/a and 0.0088 ± 0.0026/a, respectively, during the decade after 1992, and in the 60°N section, the IW θ and S have been increasing by 0.044 ± 0.005°C/a and 0.0085 ± 0.0007/a until 2005.
6. Attribution of the Observed Changes
 Magnitudes of the derived positive trends substantially exceed the rates of the long-term and resent decadal warming and salinification of the intermediate waters in the subtropical and tropical NA (≤0.023°C/a, ≤ 0.005/a, see section 2). Therefore, there should be an additional cause of the intense IW warming and salinification besides the θ and S increase in the source water masses.
 For instance, the 1997–2005 IW salinity increase at 60°N east of 13°W, where the MOW influence is most prominent, is 0.066 (0.0086/a). If we chose the rates of the MOW long-term salinification (∼0.003/a) documented by Potter and Lozier  and the 1995–2003 MOW salinification (0.005/a) reported by González-Pola et al.  (see section 2) as the reference values, then only ∼35–60% (roughly half) of the IW salinity increase at 60°N east of 13°W can be attributed to the MOW salinification, but a different explanation is required to account for the remaining ∼40–65%.
 As has been recently shown for the 1950s–1990s time period, northward penetration of MOW into the northeastern NA is governed by the NAO on a decadal time scale [Lozier and Stewart, 2008]. During persistent high NAO periods (strong atmospheric forcing), the subpolar gyre cyclonic circulation intensifies and the subpolar front (SF) moves eastward limiting northward advection of MOW and thus producing negative salinity anomalies in the Rockall Trough at the MOW levels. Conversely, during low NAO periods (weak forcing), the SF moves westward resulting in positive salinity anomalies associated with northward advance of MOW. Therefore we suggest that the most likely cause of the intense warming and salinification of IW is an increase of contribution of the source water masses (MOW and AAIW) to the IW layer due to a slowing and contraction of the subpolar gyre and corresponding northwestward shift of the SF in response to the NAO decline after 1995 [see Häkkinen and Rhines, 2004; Hátún et al., 2005; Bersch et al., 2007].
 To examine this suggestion, we have tracked the SF position in the 60°N section in 1997–2007 (see Figures 4a and 4c) and found a close positive correlation of 0.80 (significant at 0.95 confidence level) between the IW salinity and SF longitude (Figure 4b). This result suggests that the SF longitude changes may have accounted for up to 64% (r2 = 0.64) of the IW salinity changes at 60°N during the considered decade, being generally consistent with the above estimate of the unexplained part (40–65%) of the 1997–2005 IW salinity trend east of 13°W.
 In 1997–2005, the SF steadily moved westward by ∼280 km from its easternmost position at 32.3°W in 1997 to the westernmost one at 37.5°W in 2005 (Figure 4a) indicating a retreat of cold fresh subpolar waters and an advance of warm saline subtropical waters that likely reinforced the IW temperature and salinity increase. This increase was accompanied by the significant decrease of minimum oxygen concentrations in the IW layer (from ∼230 to 215–220 μmol/kg west of 13°W and from ∼250 to 235–240 μmol/kg east of this longitude), and this also makes the case for an increase of subtropical water fraction at the section latitude.
 The 2005–2007 IW cooling and freshening (Figure 3) can be explained in terms of the SF displacement as well. After 2005, the SF shifted back eastward to 34.1–34.6°W (Figure 4a) indicating the subpolar gyre expansion that likely resulted in retraction of subtropical intermediate waters and thus in the observed decrease of IW temperature and salinity.
 Both the SF longitude and IW salinity at 60°N in 1997–2007 appeared to be closely negatively correlated with the winter NAO index (Figure 5); the closest correlations (−0.77 for SF longitude, −0.91 for IW S, significant at 0.95 and 0.99 levels, respectively) are achieved for the index averaged for the five winters preceding the section repeats, i.e., the 5-year running mean NAO lagged by 2 years (hereinafter, rmNAO).
 The 1997–2005 steady westward displacement of the SF (Figure 4a) and steady increase of IW temperature and salinity at 60°N (Figure 3) are consistent with the 1997–2005 persistent decrease of rmNAO (Figure 5); the subsequent eastward shift of the SF and corresponding IW θ and S decrease are likewise consistent with the rmNAO increase. Overall, the IW salinity, SF longitude and rmNAO changes at 60°N were persistent in 1997–2005; all these characteristics reached their extreme values in 2005, and the tendency reversed in 2005–2006. This coherence goes in line with the suggestion that the SF location and IW properties both depend on the intensity and zonal extension of the subpolar gyre managed by the NAO-related atmospheric forcing.
 The 1990–2002 IW salinity at 53°N is closely correlated with the NAO index as well (Figure 6); the correlation is slightly lower than at 60°N, but is still high, being closest (−0.83, significant at 0.95 level) for the index averaged for the three preceding winters.
 Our estimates are generally consistent with the recent results on the hydrographic changes in the northeastern NA. Thus, the IW salinity changes at 60°N are (qualitatively) close to the upper-ocean salinity changes in the Faroe-Shetland Channel (∼61°N), where salinity began to increase in 1996, reached a peak in 2004 (i.e., ∼1 year earlier than IW S) and showed a decrease in 2005–2006 [Holliday et al., 2008]. Negative correlation between the SF longitude at 60°N and rmNAO (−0.77) during the considered decade corresponds to the close negative correlation (−0.81) between the SF longitude (at 55°N) and NAO on a decadal time scale [see Lozier and Stewart, 2008]. The 2-year “running mean NAO – SF longitude” lag, at which the best correlation is achieved, agrees with the notion of the 1–2-year response of the SF position to the NAO change [see Bersch et al., 2007].
 It should be noted that the NAO index used herein for the examination of the link between the observed hydrographic changes and changes in the atmospheric forcing represents an effective but quite a simplified measure of the surface forcing variability in the North Atlantic subpolar latitudes. In fact, the main factors affecting the circulation, and thus the temperature and salinity patterns, are wind stress curl and air–sea heat fluxes. In a number of model studies [e.g., Eden and Willebrand, 2001; Gulev et al., 2003] it has been shown that the leading modes of the wind stress curl and surface heat flux are closely correlated with each other and with the NAO index. Thus, Gulev et al.  reported correlation of 0.89 between the leading mode of the wind stress curl and the NAO index for the period from the late 1950s to the late 1990s. For the total density flux, correlation with the NAO index is somewhat smaller (0.73) being in agreement with the correlation of surface heat flux with the NAO [Gulev et al., 2007]. Therefore, a contribution of the actual surface forcing to the observed thermohaline changes could differ to some extent from that inferred herein from the IW salinity–NAO index correlation. Furthermore, the persistence of the NAO link to the circulation changes in the 2000s may not necessarily hold to the same extent as in earlier decades.
7. Contribution to the Recent Hydrographic Changes at 60°N
 The IW θ–S changes reported herein complete an overall picture of the recent warming and salinification of intermediate and deep waters in the 60°N section. Water masses of subpolar origin, LSW and the Iceland-Scotland Overflow Water, steadily warmed and salinified at 60°N in 1997–2006, and this resulted in the net warming and salinification of the LSW–deep-water layer (σ0 ≥ 27.72) by 0.20°C and 0.029 [Sarafanov et al., 2007]. Taking into account the steady θ and S increase in the IW layer in 1997–2005 and the fact that the subsequent 2005–2006 θ and S decrease was relatively small (−0.032°C, −0.003), we have estimated the total 1997–2006 θ and S change in the IW–LSW–deep-water stratum (σ0 ≥ 27.45) at 60°N and found that the entire intermediate–deep water column became 0.31°C warmer and 0.036 saltier.
8. Summary and Concluding Remarks
 Using the CTD data from the 14 hydrographic sections, we quantified the recent θ and S changes in the oxygen-minimum layer of intermediate waters of southern origin (IW) at ∼53°N and ∼60°N in the eastern subpolar North Atlantic. The warming and salinification of IW at rates of 0.044–0.049°C/a and 0.0085–0.0088/a during the 1990s to the first half of the 2000s is reported. Sustained warming and salinification of the IW, Labrador Sea Water and deep waters during the recent decade (1997–2006) resulted in increase of temperature/salinity at the intermediate–deep levels (σ0 ≥ 27.45) at 60°N by ∼0.3°C/0.036, and this represents the most prominent event of warming and salinification ever observed at these levels in the subpolar North Atlantic.
 Substantially (roughly twice) higher magnitude of the observed IW salinity increase relative to what was expected given the upstream trends, along with close correlations between the IW salinity, subpolar front location at 60°N and lagged NAO index suggest that temperature and salinity at the intermediate levels in the northeastern North Atlantic are strongly affected by the NAO-dependent extent of northward advection of warm saline intermediate waters from the subtropical gyre.
 Though the two seven-point salinity time series analyzed herein allowed us to deduce a likely strong dependence of the IW properties on the NAO state, the time series are currently too short to reliably assess an extent of this dependence and an exact lag of the property response to the NAO signal. Very tentatively, the NAO-related changes in the subpolar gyre zonal extension may account for more than half of the observed θ–S changes at the IW levels, and the IW temperature and salinity appear to respond to the NAO changes within 2 years. Data from the ongoing annual observations at 60°N will be used to prolong the time series and verify these estimates. The extent to that the regional thermohaline changes at the intermediate levels depend on the wind stress curl and heat flux variations still has to be investigated.
 We thank all who contributed to the data acquiring and processing, especially Sergey Gladyshev. Our special thanks to Sergey Gulev for useful suggestions. The two anonymous reviewers are gratefully acknowledged for their constructive comments that helped to improve the manuscript. This study was supported by the Russian Ministry of Education and Science (contract 02.515.11.5032), the Russian Foundation for Basic Research (grants 08–05–00858, 07–05–00657 and 08–05–00943), and the Russian President grants MK–1998.2008.5 and MK–1656.2007.5.