Transatlantic temperature and salinity changes at 24.5°N from 1957 to 2004

Authors


Abstract

[1] In 2004 a fifth occupation of a transatlantic section at 24.5°N allows us to examine decadal temperature and salinity changes using high quality full-depth hydrographic data since 1957. Waters shallower than 1750 dbar have been warming and salinifying at least since 1981 and in 2004 are significantly warmer and saltier than at any time since 1957, while deeper than 3000 dbar there has been continuous cooling and freshening since 1957. Temperature and salinity changes at constant pressures are partitioned into changes on isopycnal surfaces and changes due to the vertical movement of the isopycnals. Warming in the western Atlantic thermocline since 1957 at a rate of 0.0111 °C/yr dominates the transatlantic average, while deep water has cooled and freshened at rates of −0.0021 °C/yr and −0.0003 psu/yr respectively. We argue that the shallower and deeper changes are consistent with a recently reported increased southward thermocline circulation and reduced southward flux of deep water.

1. Introduction

[2] Between 1957 and 1998 there has been four hydrographic occupations of the transatlantic 24.5°N section. A new occupation of the section in 2004 enables us to extend the timeseries of decadal changes of potential temperature (θ) and salinity (S) in the subtropical gyre (Figure 1, Table 1).

Figure 1.

Between 69° 9′W and 23° 30′W the 24.5°N transatlantic section has been occupied five times. In 1957 (black circles) and in 1981 (blue crosses) the western boundary was approached at 24.5°N while in 1992 (pink pluses) the boundary was closed perpendicular to the continental slope by a small adjustment to the zonal section. For sections in 1998 (red circles) and 2004 (green pluses) the western boundary is closed at 26.5°N. Vertical dashed lines show the western and eastern regions that are used to diagnose temperature and salinity changes between sections.

Table 1. Hydrographic Sections Along 24.5°Na
YearDate of OccupationNumber of StationsReference
  • a

    The 1957 data differ most from the other occupations: the number of stations is much lower and temperature and salinity data were obtained from discrete samples at approximately 25 depths (the vertical resolution is 50 dbar spacing between 0 and 250 dbar increasing to 400 dbar spacing below 3500 dbar). Using the constancy and linearity of the Eastern basin deep Θ/S relationship between 2 and 2.5°C and assuming constant deep water characteristics, the 1957 salinities are between 0.004 to 0.006 saltier than in subsequent years [Bryden et al., 1996; Arbic and Owens, 2001].

1957October38Fuglister [1960]
1981August90Roemmich and Wunsch [1985]
199220th July–8th August101Parrilla et al. [1994a]
199824th Jan–23rd Feb130McTaggart et al. [1999]
20045th April–9th May125Cunningham [2005]

[3] Since 1957 the transatlantic average temperature and salinity changes have a pattern like the baroclinic first mode, with warming and salinification of thermocline and Intermediate water (IW) above 1750 dbar and cooling and freshening of North Atlantic Deep Water (NADW) from 1750 dbar to 5500 dbar. Antarctic Bottom Water (AABW), confined to the western basin has also cooled and freshened.

[4] According to Bryden et al. [1996] the thermocline steadily salinified from 1957 to 1992 at a rate of 0.001 psu/yr at constant potential temperatures and on constant density surfaces warmed and freshened at rates of 0.0065 °C/yr and 0.0018 psu/yr. In the IW [Parrilla et al., 1994b; Bryden et al., 1996] there has been a maximum warming and salinification at 1100 dbar of 0.01 °C/yr and 0.0023 psu/yr respectively between 1981 and 1992. Using Argo floats in the eastern basin Vargas-Yáñez et al. [2004] report that the eastern thermocline (from 17.5 to 42.5°W) at 400 dbar has been warming at a higher rate of 0.042 °C/yr from 1992 to 2002. An accompanying salinity increase conserves the temperature and salinity structure so the changes at fixed depth are explained by a downward displacement (heave) of water masses.

[5] Diagnosing the thermocline and intermediate water mass warming Bryden et al. [1996] report that from 1957 to 1981 this was principally due to downward heave of isopycnals and isotherms by about 50 dbar, with little change in the θ/S structure. From 1981 to 1992 the warming and salinification was dominated by changes in the water mass characteristics with higher salinities and temperatures on isopycnal surfaces.

[6] At depth Bryden et al. [1996] show that cooling and freshening on isopycnals since 1981 is principally due to changes in water mass characteristics.

[7] In this paper we report temperature and salinity changes across 24.5°N from five high quality full-depth transatlantic hydrographic sections in 1957, 1981, 1992, 1998 and 2004.

2. Data and Methods

[8] Changes in temperature or salinity at constant pressure can occur as changes along a neutral density surface [Jackett and McDougal, 1997] or as the vertical movement of a water mass with constant properties. Bindoff and McDougall [1994] show that these changes can be written as,

equation image

Where equation image is the rate of change of property ϕ on an isobar p, equation image is the rate of change of ϕ on the neutral density surface γ, equation image is the rate of vertical displacement of the neutral surface (also known as heave) and equation image is the vertical gradient of ϕ.

[9] We find that property changes on either side of the Mid-Atlantic Ridge (MAR) often result from different causes. To emphasise these differences we compute the terms in (1) for the western basin between 65° and 75°W and the eastern basin between 25° and 35°W.

3. Results

[10] Firstly, we describe temperature and salinity changes on constant pressures between 1957 and 2004 before examining the changes between each of our five sections in terms of equation (1) above.

3.1. Potential Temperature and Salinity Changes on Constant Pressure Surfaces From 1957 to 2004

[11] The dominant pattern of temperature and salinity change is a pattern like the baroclinic first mode (Figure 2). From 1957 to 2004 there has been warming of the upper ocean between 200 dbar and 2500 dbar with salinity increases from 200 dbar to 1750 dbar and cooling and freshening of deeper water masses below 2500 dbar and 1750 dbar respectively. In the thermocline and IW the basin wide average temperature and salinity changes are dominated by changes in the western basin, with changes in the eastern basin matching in sign but of much smaller amplitude. The freshening and cooling of the deep waters is much larger in the western basin than in the east.

Figure 2.

(a) Potential temperature and (b) salinity change from 1957 to 2004 averaged zonally on pressure surfaces (2004–1957) plotted against pressure. Shading shows ±1 standard error of the mean zonal difference. Full transatlantic section (solid); 65 to 75°W (dashed); 25 to 35°W (dotted). The water mass layers are: thermocline, 300 to 800 dbar; Intermediate Water, 900 to 1750 dbar; upper North Atlantic Deep Water, 1750 to 2500 dbar; western basin lower North Atlantic Deep Water, 3000 to 4000 dbar; eastern basin lower North Atlantic Deep Water, 3000 dbar to sea-bed; and western basin Antarctic Bottom Water, 5000 dbar to sea-bed.

[12] In Figure 2 the basin wide top-to-bottom average warming is 0.046 ± 0.116 °C (0.098 °C/century) so upper ocean warming dominates deep cooling while the salinity change is not significant at 0.001 ± 0.018 psu.

[13] In the following sections we diagnose the temperature and salinity changes observed on the five transatlantic sections at 24.5°N using equation (1) to partition changes at a fixed pressure into isopycnal changes, and changes due to heave of the isopycnals.

3.2. Thermocline

[14] At a constant pressure changes in temperature and salinity of the thermocline are dominated by isopycnal heave carrying the local θ/S up or down, with temperature and salinity changes on isopycnals being relatively small (Figure 3). Although the thermocline alternately shallows and deepens there is a trend in heave since 1957. However, the western thermocline and eastern thermocline are out of phase: deepening in the west corresponding to shallowing in the east, though the eastern basin vertical motions are about half of those in the west. Thus the thermocline appears to increase in slope by pivoting around the MAR, and we would expect the times of steeper thermocline to correspond to increasing transport. On isopycnals from 1957 to 1998, the basin wide thermocline has warmed and salinified but cooled and freshened from 1998 to 2004. Thus, there are long-term trends in properties on isopycnals compared to the changes caused by heave. This is reflected in the net changes from 1957 to 2004 where the isopycnal warming and salinification is a much larger proportion of the net changes. In the western basin the total temperature increase at fixed pressure is 0.0111 ± 0.003 °C/yr with contributions from isopycnal increases of 0.0028 ± 0.0013 °C/yr and heave −0.0076 ± 0.0037 °C/yr. For salinity, the increase at fixed pressure is 0.002 ± 0.0003 psu/yr from isopycnal increases 0.0008 ± 0.0005 psu/yr and heave −0.0011 ± 0.0005 psu/yr. In the eastern basin the net temperature increase at constant pressure is 0.0026 ± 0.0013 °C/yr, with cooling on isopycnals −0.0002 ± 0.0009 °C/yr and warming due to downward heave −0.0028 ± 0.001 °C/yr. The salinity increase at constant pressure is 0.0003 ± 0.0002 psu/yr, with freshening on isopycnals of −0.0001 ± 0.0002 psu/yr and salinification due to downward heave of −0.0003 ± 0.0002 psu/yr.

Figure 3.

(a) equation image = equation imageequation image, where equation image (black) is the time rate of change of potential temperature (θ) evaluated at fixed pressure (p), equation image (red) is the time rate of change of θ evaluated on neutral density surfaces (γ) and equation image (green) is the time rate of change of θ evaluated at p due to heave equation image of the local potential temperature gradient equation image. dt is the time interval between sections, and each term is multiplied by dt to give the net change between each time and then plotted as a running sum. The units are °C for θ, practical salinity units for salinity, and dbar for heave. Terms are a zonal average in the western basin from 65 to 75°W, and the error bars represent one standard deviation of the vertical average of each water mass layer. The left column is θ, middle is salinity (S), and the right column is the isopycnal heave equation image × dt. (b) Same as for Figure 3a but for a zonal average in the eastern basin from 25 to 35°W. Layers defined in Figure 2.

[15] In summary from 1957 to 2004 warming and salinification in the western basin dominates the basin wide thermocline changes. The western basin warms and salinifies on isopycnals that deepen at a rate of 0.40 ± 0.13 dbar/yr. The eastern thermocline has also warmed and salinified, though this has been counteracted by cooling and freshening on isopycnals. Deepening of the thermocline dominates the isopycnal changes, though the deepening rate of 0.18 ± 0.05 dbar/yr is less than in the western basin, resulting in a steeper thermocline overall.

3.3. Intermediate Water

[16] Intermediate water circulation at 24.5°N is weak, sandwiched between the southward flowing thermocline and NADW below. At shallower IW levels the meridional transport is northward but below 1000 dbar is southward, whilst the net zonally averaged meridional flow is a few Sverdrups northward [Bryden et al., 2005b]. This causes a variable θ/S relationship that in the west combines Antarctic Intermediate Water and Western North Atlantic Central Water and in the east is strongly influenced by the Mediterranean Overflow Water (MOW).

[17] In the west temperature changes at fixed pressure (Figure 3) are dominated by isopycnal heave. From 1957 to 2004 the temperature change at fixed pressure is 0.0065 ± 0.0018 °C/yr, on isopycnals is 0.0013 ± 0.0015 °C/yr and due to heave is −0.0052 ± 0.0014 °C/yr. In contrast, occupation-to-occupation salinity changes at fixed pressure are explained by variations in salinity along isopycnals with only minor contributions from heave. However, over five decades, the salinity change at fixed pressure is 0.0004 ± 0.0002 psu/yr, partitioned equally between changes along isopycnals of 0.0002 ± 0.0003 psu/yr and due to heave −0.0002 ± 0.0001 psu/yr.

[18] In contrast, eastern basin MOW temperature and salinity changes at fixed pressure are determined by changes along isopycnals. Although there are large vertical motions of the isopycnals the vertical gradients of temperature and salinity are small and have negligible impact on temperature and salinity changes at fixed pressure. Over the period from 1957 to 2004, the net temperature and salinity changes are small, but suggest a slightly larger role of heave for the net temperature change. The temperature change at fixed pressure is 0.0014 ± 0.0017 °C/yr, on isopycnals is −0.0006 ± 0.0016 °C/yr and due to heave is −0.0023 ± 0.0006 °C/yr. For salinity, the change at fixed pressure is −0.0002 ± 0.0004 psu/yr, on isopycnals is −0.0002 ± 0.0003 psu/yr and due to heave is −0.0001 ± 0.0001 psu/yr.

3.4. Upper North Atlantic Deep Water

[19] Upper NADW (Labrador Sea Water) is formed by deep winter convection in the Labrador Basin [Lazier, 1973; Lazier, 1980] and the western basin is directly ventilated by the deep western boundary current that advects uNADW to 26.5°N in about 10 years [Molinari et al., 1998]. In contrast the eastern basin is more indirectly ventilated in the region of the Charlie Gibbs Fracture Zone at 54°N, then southward possibly on the east flank of the MAR [Talley and McCartney, 1982; Cunningham and Haine, 1995], and consequently the temperature and salinity variability at fixed pressure differs west to east across the MAR. In the west from 1957 to 1981, isopycnals in the core of LSW deepen by almost 100 dbar, thereafter alternately shallowing and deepening only slightly. This pattern of heave has led to contrasting behaviours for potential temperature and salinity. Temperature changes at fixed pressure are created by heave of the isopycnals, but for salinity there are large and compensating changes between isopycnal changes and heave. Post 1992 the variability in salinity at fixed pressure is caused by salinity changes along isopycnals. From 1957 to 2004, the net changes in salinity are smaller than the error estimate for the 1957 salinities. Temperature shows a small warming at fixed pressure of 0.0009 ± 0.0004 °C/yr, caused by a cooling on isopycnals of −0.0014 ± 0.0003 °C/yr and by heave induced changes of −0.0023 ± 0.001 °C/yr (net deepening of isopycnals of almost 100 dbar).

[20] In the east heave plays little role in explaining temperature and salinity variations at fixed pressure with a strong compensation from temperature and salinity changes along isopycnals, with the 2004 temperature and salinity properties being indistinguishable from those in 1957.

3.5. lNADW

[21] The western basin lNADW layer is dominated by the circulation of the DWBC [Bryden et al., 2005a]. In the DWBC at constant pressure and on isopycnals, there has been continuous cooling and freshening since 1957 (except from 1992 to 1998 when there was warming at fixed pressure). From 1957 to 2004, isopycnals have shallowed by 49 ± 10.5 dbar. At fixed pressure there has been a cooling of −0.0021 ± 0.0002 °C/yr, caused by a cooling on isopycnals of −0.0014 ± 0.0002 °C/yr and by heave induced changes of 0.0006 ± 0.0002 °C/yr. For salinity the change at fixed pressure is −0.0003 ± 0.0001 psu/yr, on isopycnals is −0.0003 ± 0.0000 psu/yr and due to heave is 0.0000 ± 0.0000 psu/yr.

[22] In the east temperature and salinity changes at fixed pressure are small and the sign of heave dictates the sign of the changes though there are larger changes of temperature and salinity on isopycnals than in the west. For the period 1957 to 2004 at fixed pressure there has been a cooling of −0.0008 ± 0.0001 °C/yr, caused by a cooling on isopycnals of −0.001 ± 0.0002 °C/yr and by heave induced changes of −0.0001 ± 0.0001 °C/yr. For salinity the change at fixed pressure is −0.0002 ± 0.0000 psu/yr, on isopycnals is −0.0002 ± 0.0000 psu/yr and due to heave is 0.0000 ± 0.0000 psu/yr.

3.6. AABW

[23] AABW is confined to the western basin, flowing northward against the western flank of the MAR. For the period 1957 to 2004 at fixed pressure there has been a cooling of −0.0015 ± 0.0001 °C/yr, caused by a cooling on isopycnals of −0.0011 ± 0.0002 °C/yr and by heave induced changes of 0.0002 ± 0.0001 °C/yr. For salinity, the change at fixed pressure is −0.0003 ± 0.0001 psu/yr, on isopycnals is −0.0003 ± 0.0001 psu/yr and due to heave is 0.0000 ± 0.0000 psu/yr.

4. Discussion and Conclusion

[24] Following Bindoff and McDougall [1994], we diagnose the temperature and salinity changes at 24.5°N in the Atlantic for five transatlantic sections between 1957 and 2004 by decomposing the changes at fixed pressure into changes on isopycnals and heave of the vertical temperature or salinity gradients. For this simple model two mechanisms can contribute to heave. Firstly, the gyre circulation strength may change in response to changing wind forcing and secondly isopycnal layers may change thickness by changing renewal rates. Isopycnal changes, in the absence of mixing, reflect changing buoyancy forcing at the water mass sources. As noted by Bryden et al. [1996] there will be differing time responses for these three processes and we would also emphasise differing spatial responses due to the circulation and bathymetry. We now discuss possible causes for the changes.

[25] The upper ocean warming and salinification at 24.5°N are observed over all of the sub-tropical gyre extending southward across the equator at least as far as 32°S. From 32°S to 36°N the average temperature change between 1000 dbar and 2000 dbar is 0.00375 ± 0.00089 °C/yr [Arbic and Owens, 2001; Joyce et al., 1999]. Parrilla et al. [1994b] suggests that the upper ocean temperature increases at 24.5°N are consistent with the results of climate models forced with a CO2 doubling. Curry et al. [2003] show that the salinification of the subtropical thermocline results from enhanced evaporation co-incident with a protracted high state of the North Atlantic Oscillation: the NAO enhancing zonal winds leading to increasing evaporation and increasing subduction of saltier water into the thermocline.

[26] While one cause of increasing temperature and salinity is global warming and associated hydrological changes, circulation changes may also play a role. Bryden et al. [1996] estimate that horizontal shifts of the thermocline water mass structure are unlikely to explain upper ocean variability because the spatial gradients of temperature and salinity are small and the required displacement of water masses is several thousand kilometres. However, analysing the five repeat hydrography sections Bryden et al. [2005b] calculate that the southward thermocline flux increased by 50% from −12.7 to −22.8 Sv between 1957 and 2004, with the majority of the increase occurring between 1992 and 1998. At 24.5°N thermocline temperature and salinity changes are driven mainly by deepening of the isopycnals. The heave forcing is out of phase west to east, and is much larger in the west. This results in a thermocline slope that in 2004 is steeper than at any time since 1957, consistent with the calculations of larger southward thermocline transport.

[27] In the deep ocean, the net effect of heave on temperature and salinity changes at fixed pressure varies subtly as the vertical gradients of potential temperature and salinity vary. Of particular note in this regard is the contrasting response of temperature and salinity in the uNADW. For salinity prior to 1992 isopycnal and heave dominate changes at fixed pressure but from 1992 to 2004 rapid freshening at fixed pressures is due to freshening along isopycnals. The arrival of LSW at 26°N takes around 10 years [Molinari et al., 1998] and since the 1980s LSW has been getting colder and since 1985 fresher as well, and we see this water crossing 26°N after 1992. For NADW in the eastern basin, far from the source regions isopycnal changes are more likely to be caused by mixing rather than air-sea interaction in their source region.

[28] Bryden et al. [2005b] report a 50% reduction in the transport of the lNADW from −14.8 Sv in 1957 to −6.9 Sv in 2004 compensating the increased southward thermocline flow. This has been accompanied by a continuous cooling and freshening of lNADW and a net shallowing of the isopycnals. Between 65 and 75°W the isopycnals slope down to the west. The circulations derived by [Bryden et al., 2005b] assume a zero velocity at the base of the lNADW with the northward flowing AABW below. A net shallowing of isopycnals in the west of the basin is consistent with a reduction of the southward transport of lNADW.

[29] Baehr et al. [2007] compare the observed temperature and salinity changes at 24°N between 1957 and 2004 to a control climate simulation of the coupled ECHAM5/MPI-OM global model for which there are no long term trends in MOC strength. Regressing model temperature and salinity against MOC strength at 1000 m replicates the observed pattern of upper ocean warming and salinification with more pronounced changes in the western boundary. In the model lNADW is much shallower than in reality but cools and freshens as in the observations. The magnitude of the observed changes is consistent with model MOC reduction of 8 Sv, which is within the model's range of natural variability. The slowing of the Atlantic overturning reported by Bryden et al. [2005b] appears to be consistent with the warming of the thermocline and cooling of lNADW reported here, however Baehr et al. [2007] suggest that these property changes may be driven by natural interannual variability of the MOC. The observations reported here do show a 50 year trend in decreasing temperatures and salinities for lNADW. These deep property changes we expect to be highly integrating of processes in the north Atlantic and consequently may vary on a quite different timescale from the MOC strength measured at 1000 m. Thus the deep property changes may also provide a way to partition the causes of upper ocean heating due to global warming or to MOC interannual variability or trends.

Acknowledgments

[30] This paper is a contribution to the James Rennell Division for Ocean Circulation and Climate core strategic programme “Ocean Variability and Climate.”

Ancillary