Variability of AABW properties in the equatorial channel at 35°W



[1] Hydrological data and tracer (silicate, oxygen and CFCs) analyses have been performed in order to describe the time and space variability of the characteristics and transport of the Antarctic Bottom Water (AABW) entering the equatorial channel at 35°W. The 35°W sections were sampled in the tropical Atlantic during several cruises from 1993 to 1999 as part of the international WOCE and CLIVAR efforts. Data for the 1990s reveal a slight warming trend superimposed on significant variability due to the seasonal changes of AABW transports. Data from previous cruises made in the 1970s and 1980s complete the time-series and show that, over the last three decades, the maximal AABW input occurred in the early 1990s.

1. Introduction

[2] The Antarctic Bottom Water (AABW) flowing northwards over the Atlantic ocean floor below 4000 m, west of the Mid-Atlantic Ridge (MAR), strongly affects the southward export of the lower North Atlantic Deep Water (LNADW) which is part of the thermohaline circulation. Using a global modelling approach, Stössel and Kim [2001] described a possible link between convection in the Weddell Sea and decadal variability of deep-water flows in the (sub)tropical Atlantic.

[3] Orsi et al. [1999] described the AABW as a mixture of Lower Circumpolar Deep Water (LCDW) and dense water fed from Antarctic margins. In addition to the Weddell Sea Bottom Water (WSBW), other AABW sources were reported, based on measurements of CFCs which are tracers of water ventilation [Smythe-Wright and Boswell, 1998; Meredith et al., 2000; Doney and Hecht, 2002]. The unventilated LCDW is characterised by low oxygen and low CFC content and high silicate concentrations, while the recently formed Weddell Sea Deep Water (WSDW) is colder and less rich in silica but richer in oxygen and in atmospheric tracers [Weiss et al., 1979]. At low latitudes (10°S–5°S), there is evidence of “young” WSDW components (σ4 > 46.06) in CFC data [Roether et al., 1993; Andrié et al., 1998; Rhein et al., 1998] but this water mass has not yet reached the equatorial channel [Andrié et al., 2002].

[4] The variability of the AABW properties in the southern Atlantic over the last three decades has been widely documented. The AABW warming, at the end of the 1980s, reported by Coles et al. [1996] seems somewhat questionable according to data collected at the entrance to the Argentine and the Brazil Basins during the early 1990s [Hogg and Zenk, 1997; Hogg and Owens, 1999; Arhan et al., 1999].

[5] In our study we adopted the isopycnal σ4 = 45.90 as the boundary between the LNADW and the AABW inferred from tracer data [Rhein et al., 1995, 1998; Hall et al., 1997; Andrié et al., 1999]. This boundary is somewhat different to the 1.9°C boundary originally defined by Whitehead and Worthington [1982] North of 4°N on the basis of hydrological and current measurements.

[6] In the following section we describe the distributions of AABW characteristics in the deepest part of the equatorial channel from 1993 to 1999. In section 3, we discuss the variability of the AABW properties, by referring to the seasonal variability of transports. In section 4 we extend the study to the last 30 years for the tropics and the South Atlantic basin.

2. Data and Water Mass Vertical Distributions

[7] Hydrographic and geochemical tracers were collected during the CITHER 1 (February 1993, henceforth referred to as CIT1), ETAMBOT 1 (October 1995, ET1) and ETAMBOT 2 (May 1996, ET2) cruises and the EQUALANT 99 (July 1999, EQ1) cruise along the 35°W meridian with a horizontal resolution of less than 30 nm. These data enabled us to describe the overall variability and the change in properties of the AABW at the equator. The precision for temperature, salinity, and oxygen measurements obtained with CTD-O2 probes was respectively less than 0.005°C, 0.003 psu and 1.5 μmol kg−1 [Arhan et al., 1998; Bourlès et al., 1999; Gouriou et al., 2001]. The precision for dissolved oxygen measurements was 0.5 μmol kg−1 and 0.2 μmol kg−1 for silicate. The standard analytical precision for CFC-11 was within 1% [Andrié et al., 1999]. Taking into account the blank value and analytical detection limits, CFC-11 data differences between all three cruises could not be significant for variations lower than 0.02 pmol kg−1 [Andrié et al., 2002].

[8] We completed data from the previously mentioned French WOCE 1993–1999 time-series during the following cruises: GEOSECS (October 1972), TTO Knorr104 (August 1983), SAVE A16S (March 1989), Meteor cruises (from October 1990 to February/March 1994) [Rhein et al., 1998], WOCE A17 (CIT2, February/March 1994 at 32°30′W) and WOCE A15 (April/May 1994 at 36°W). The hydrological and chemical characteristics of deep water flowing westward across the 35°W meridian are strongly constrained by the topography of the equatorial channel between 1°S and 0°30′N. Figure 1a shows a vertical CTD profile of oxygen versus σ4 (density anomaly relative to 4000 m) for the deepest cast for each cruise. The density range corresponds to deep waters below about 3600 m depth down to the bottom. LNADW is well identified above the σ4 = 45.90 interface through an oxygen maximum (264 μmol kg−1; Figure 1a) and a silicate minimum (36 μmol kg−1; Figure 1b). The maxima of oxygen and of CFC-11 (Figure 1c) are observed at the same density level and can be attributed to the Denmark Strait Overflow, since the CFC maximum increased regularly from 1993 to 1999 as a result of its transient origin.

Figure 1.

Oxygen (a), silicate (b) and CFC-11 (c) vertical profiles versus σ4 (density anomaly relative to 4000m) for the LNADW/AABW layers below 3600 m. The profiles correspond to the CIT1 (blue squares), ET1 (red diamonds), ET2 (green triangles) and EQ1(yellow dots) cruises in the deepest part of the channel near 0°20′N at 35°W.

[9] The CFC minimum (Figure 1c) observed at the bottom confirms that only the old LCDW component flows through the equatorial channel (without any ventilated WSDW component). The linear trends observed for each characteristic between σ4 = 45.90 and the bottom reflect the mixing between LNADW and AABW.

3. Variability of AABW Properties and Transports Over the 90s Decade

[10] Figure 2 shows the value of the potential temperature (a) and silicate (b) at the greatest pressure for each cast between 1°S and 0°30′N along 35°W. The shape of the meridional distribution (the Y axis is reversed for silicates) is very similar to the shape of the bathymetry along 35°W. The deeper the sampled depth, the greater the AABW influence. Figure 2a shows a steady temperature increase between the CIT1 and EQ1 cruises. A similar evolution can be observed in silicate distributions (Figure 2b): silicate concentration decreases from CIT1 and ET1, to ET2 and EQ1.

Figure 2.

Latitudinal distributions of potential temperature (a) and silicate (reversed Y axis) (b) for the CIT1, ET1, ET2 and EQ1 cruises. The shape of the channel bathymetry has been superimposed along 35°W.

[11] Due to the detection limit of the measurements and the “blank” levels, we did not observe any CFC enrichment between 1993 and 1999 (Figure 1c), although this could have been expected from the mixing of WSDW as observed at 5°S, east of 32°W- in 1994 [Rhein et al., 1998]. Our measurements confirm that the bottom topography prevents the WSDW from reaching the equator.

[12] The salinity/temperature and silicate/temperature diagrams in Figure 3 compare the deepest CIT1 CTD profile, made at 0°20′N, with the deepest bottle measurements made between the equator and 0°30′N and between 35°W and 37°W since 1972. The pressure corresponding to the deepest samples generally falls within an interval of less than 10m. In Figure 3a the bottom properties stand very near to the CTD mixing line (except in 1983 where the sampled depth is about 30m above the bottom). In order to point out possible seasonal effects, boreal autumn cruises are highlighted in pink and boreal spring cruises in green in Figure 3.

Figure 3.

(a) Salinity versus temperature diagram for the deepest samples of the published data from 1973 to 1999. The CTD data from CIT1 are given as a reference (purple circles). (b) The same for silicate versus potential temperature (the purple dashed curve is the silicate versus temperature mixing line for CIT1). The Y axis has been reversed for silicate. Cruise dates are given in month-year. The same symbols used in Figures 1 and 2 are used for CIT1, ET1, ET2, EQ1. Boreal spring situations are highlighted in green, autumn situations in pink.

[13] Salinity and temperature variability is dominated by the vertical mixing between AABW and LNADW. The same conclusion can be drawn from the silicate versus temperature diagram (reversed Y axis, Figure 3b).

[14] The temporal variability of AABW transport has previously been described as partly seasonal [Hall et al., 1997; Rhein et al., 1998]: the maximum AABW transport occurs in September–October, associated with an uplift of the LNADW/AABW transition layer, while minimum AABW transport occurs in February–March. It is difficult to determine precisely the effect of seasonal variability on AABW properties from our series: the CIT1-winter (Feb. 1993) situation seems to correspond to an inter-annual event, strongly marked by AABW properties (lowest temperature, highest silicate concentration, Figure 2); however, ET1 (Oct. 1995) measurements showing low temperature, salinity, oxygen and high silicate concentration are consistent with a strong AABW input characterising an autumn situation.

4. Variability During the 1970s, 1980s and 1990s

[15] Previous works [Whitehead and Worthington, 1982; Whitehead, 1989] have discussed the variability of AABW characteristics and transports North of 4°N near 40°W from 1977 to 1978. In this area of rough topography, the mixing effects seem to dominate the seasonal or inter-annual variabilities. On the other hand using currentmeter time-series for 1992 to 1994, for the topographically- constrained equatorial channel, Hall et al. [1997] described a strong seasonal variability together with a long term transport decrease associated with a warming trend of the AABW.

[16] To discuss the possible influence of a global climatic trend on AABW properties, we extend the time-series to include the last three decades by using GEOSECS (1972) and TTO (1983) data (Figure 3). We observe a low (silicate minimum and temperature maximum) and relatively stable AABW input from the early 1970s to 1980s (during boreal autumn situations) and a strong AABW input in February 1993 and October 1995 (silicate maximum, temperature minimum) (Figure 3b). The AABW input seems to decrease from 1995 to 1999 but such a general trend is difficult to distinguish from seasonal signals for a relatively short time series.

[17] The warmest, saltier and poorest silicate enriched AABW was observed in the 1970s/1980s, with relatively unchanged characteristics, and the coldest, freshest and most silicate enriched AABW during the mid 1990s in agreement with Hall et al. [1997]. The deviations from the AABW/LNADW mixing line in Figure 3b could correspond to inter-annual or decadal events affecting AABW characteristics, independently of local transport and/or vertical mixing variability.

[18] Such large-scale temporal variability has been noticed upstream of the AABW flow in the southern hemisphere, in the Argentine and Brazil Basins. Coles et al. [1996] described the warming of the abyssal layer between 1987 and 1989 at the entrance to the Argentine Basin following a stable situation between the 1970s and the late 1980s. The warming of the deep part of the WSDW may be observed together with the cooling (and freshening) of the overlying LCDW component [Arhan et al., 1999]. Zenk and Hogg [1996] and Hogg and Zenk [1997] reported similar features during the early 1990s at the entrance of the Brazil Basin in the Vema and Hunter Channels after a stable situation from 1972 to 1991. The available data appears to show that a WSDW warming event, without any transport decrease, propagated from the southern ocean to the Argentine and Brazil Basins over a few years, at a rate estimated at 2.3 cm s−1 at the entrance to the Argentine Basin [Coles et al., 1996]. The CFC profiles (Figure 1c) showed that the AABW at the equator never contained any ventilated WSDW, but only the LCDW component. We suggest that the cold LCDW signal observed in 1993 at the equator is directly linked to the warm WSDW anomalies observed in the Argentine Basin during the late 1980s and at the entrance to the Brazil Basin during the early 1990s. The variability of AABW properties depends strongly on the history of the different AABW cores. For the early 1990s the AABW warming at the bottom of the Hunter Channel has been linked to a decreased (or stopped) flow of LWSDW [Arhan et al., 1999]. In 1993, the LCDW cooling, which appears a few years after the warming signal of the WSDW in the South Atlantic, could be due to concomitant opposite features concerning the LCDW and WSDW components of the AABW.

5. Conclusion

[19] We have discussed the variability of the AABW flow entering the equatorial channel (35°W) from observations collected between 1990 and 1999. The weak AABW input observed before and during the early 1980s, and the strong input observed in the early 1990s, suggest an increase in AABW reaching the equatorial channel at 35°W during the WOCE period.

[20] Such a trend is superimposed upon significant variability related to the seasonal changes of AABW transports. Consequently, it is difficult to draw any conclusion about a global trend of the characteristics and transports of the AABW during the 1990s.


[21] The CITHER 1, ETAMBOT and EQUALANT cruises were supported by the Institut de Recherche pour le Développement (IRD), the Institut Français de Recherche pour l'Exploitation de la Mer (IFREMER) and the Institut National des Sciences de l'Univers (INSU) through the Programme National d'Etude de la Dynamique du Climat (PNEDC), French contribution to WOCE. The authors would like to thank the Captains and crews of the N.O. L'ATALANTE, LE NOROIT, EDWIN LINK and THALASSA for helping to make the cruises a success. We are grateful to S. Freudenthal, F. Baurand and R. Chuchla for the tracer measurements.