JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 105, NO. C6, PAGES 14,29%14,323, JUNE 15, 2000

Tracing the flow of North Atlantic Deep Water using chlorofiuorocarbons

William M. Smethie Jr., • Rana A. Fine,: Alfred Putzka, • and E. Peter Jones n

Abstract. Chlorofluorocarbon (CFC) and hydrographic data collected in the North Atlantic in the late 1980s and early 1990s are used to confirm and add to earlier work on the large-scale circulation pathways and timescales for the spreading of North Atlantic Deep Water (NADW) components and how these components relate to the hydrographic structure. Throughout the western North Atlantic, high CFC concentrations are coincident with newly formed NADW components of Upper Labrador Sea Water (ULSW), Classical Labrador Sea Water (CLSW), and Overflow Waters (OW). ULSW is marked by a CFC maximum throughout the western subtropical and tropical Atlantic, and CLSW is marked by a CFC maximum north of 38øN in data collected in 1990-1992. Iceland-Scotland Overflow Water (ISOW) splits into two branches in the eastern basin, with one branch entering the western basin where it mixes with Denmark Strait Overflow Water (DSOW) and the densest branch flows southward along the bottom in the eastern basin. DSOW contributes the largest portion of the CFC signal in OW. It is estimated that these NADW components are at 60-75% equilibrium with the CFC concentration in the atmosphere at the time of formation. The large-scale data set confirms that NADW spreads southward by complex pathways involving advection in the Deep Western Boundary Current (DWBC), recirculation in deep gyres, and mixing. Maps of the CFC distribution show that properties within the gyres are relatively homogeneous, particularly for OW, and there is a profound change at the gyre boundaries. The density of the core of ULSW increases in the equatorward direction because of entrainment by overlying northward flowing Upper Circumpolar Water and at the equator, ULSW has the same density as CLSW in the subtropics but is warmer and saltier. The density of OW decreases between the subpolar region and the subtropics. This is caused by the least dense part of OW exiting the subpolar region in the DWBC, while the densest component recirculates in the subpolar basins. Some variability is observed in OW density in the subtropics and tropics because of variability in mixing with Antarctic Bottom Water and changes in the subtropics that are probably related to the transport of different vintages of DSOW. Ages derived from CFC ratios show that the NADW components of northern origin spread throughout the western North Atlantic within 25-30 years. This corresponds to a spreading rate of 1-2 cm s '• and is comparable to the time a climate anomaly introduced into the subpolar North Atlantic will take to penetrate the entire western North Atlantic Ocean.

1. Introduction

Global thermohaline circulation is an important component of the Earth's climate system. This circulation occurs by numerous pathways that transport warm waters to high latitudes where they become more dense, sink, and spread throughout the ocean [e.g., Broecker, 1991; Gordon, 1986]. One of the primary pathways by which this occurs is the formation and spreading of North Atlantic

•Lamont- Doherty Earth Observatory of Columbia University, Palisades, New York.

2Rosenstiel School of Marine and Atmospheric Science, University of Miami, Miami, Florida.

3Institut far Umweltphyik, University of Bremen, Bremen, Germany.

aBedford Institute of Oceanography, Dartmouth, Nova Scotia, Canada.

Copyright 2000 by the American Geophysical Union.

Paper number 1999JC900274. 0148-0227/00/1999JC900274509.00

Deep Water (NADW), the properties of which have been observed throughout much of the deep ocean and extend as far as the North Pacific [Reid and Lynn, 1971 ]. The flow of warm shallow water to the northern North Atlantic to feed NADW formation results in

Europe having an anomalously warm climate for its latitude. Fluctuations between warm and cold climates in the past are thought to have been caused by changes in the formation of NADW [e.g., Broecker, 1995; Manabe and Stouffer, 1988].

During the past half century a number of manmade substances have been introduced into the environment. Some of these

substances enter the surface ocean on a global scale and are chemically stable in seawater. These substances are excellent tracers of deep water formation and circulation since they become incorporated into the deep water when it is renewed from the surface. The substances that have been most widely used in oceanographic studies are bomb tritium and radiocarbon that were injected into the atmosphere by nuclear weapons testing, mostly in the early 1960s, and anthropogenic chlorofluorocarbons (CFCs) that have been entering the atmosphere since the 1930s.

This paper is a synthesis of available CFC data for the North Atlantic Ocean up to 1992. We review the formation processes of NADW, show how CFCs become incorporated in NADW, map spreading pathways of NADW within the Atlantic Ocean using CFCs, and estimate the timescale for NADW to flow along these pathways.

14,297

14,298 SMETHIE ET AL.' TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER

2. Background Information on CFCs

CFCs are primarily anthropogenic substances that have been used extensively in our society. The two CFCs that have been most widely measured in the oceans are CFC-11 (CCI3F) and CFC-12 (CC12F2), and they have no natural sources. These substances are chemically stable in the troposphere but do decompose in the stratosphere where their decomposition products cause the destruction of ozone [Anderson et al., 1991 ]. Because of this global environmental problem, the concentrations of CFC-11 and CFC-12 have been closely monitored in the atmosphere since the mid-1970s [Elkins et al., 1993; Cunnold et al., 1994, 1997]. Prior to that time the atmospheric concentration can be reconstructed from industrial production data [Fisher et al., 1994] and the atmospheric lifetimes that have been determined from the atmospheric measurements referenced above. The concentrations of CFC-11 and CFC-12

continuously increased with time from their initial production in the 1930s until the early 1990s (Figure 1). The rate of increase was different for the two CFCs (Figure 2), resulting in the CFC-11 :CFC- 12 ratio changing with time (Figure 1) until the late 1970s, after which the rate of increase for both CFCs was about the same until

the 1990s. The rate of increase dropped in the 1990s as a consequence of the efforts of countries complying with the provisions of the Montreal Protocol to phase out CFCs. By 1995 the concentration of CFC-11 was beginning to decrease in the atmosphere, but CFC-12 was still increasing slowly [Montzka et al., 1996; Walker et al., this issue].

CFCs enter the surface ocean from the atmosphere by gas exchange, and the average equilibration time between the surface mixed layer and the atmosphere is ~1 month [Broecker et al., 1980]. Most of the surface waters, except in regions of deep convection, are within a few percent of equilibrium [Warner, 1988], and the expected concentration at a given point in time can be calculated from the atmospheric concentration and the solubility [Warner and Weiss, 1985]. Both CFC-11 and CFC-12 are chemically stable in seawater, but CFC-11 does decompose under anoxic conditions [Bullister and Lee, 1995; Shapiro et al., 1997], which are rarely found in the open ocean.

CFCs can be used to estimate ages of water parcels. The simplest way is to convert the measured concentration in water to an atmospheric concentration (using the solubilities [Warner and Weiss, 1985], assuming the water was saturated when it was at the surface) and then comparing this to the time history of CFCs in the atmosphere (Figure 1). This age, subsequently referred to as the

600 f i i ! .

CFC-11 :CFC-12 nd]c•qaøoo:•c•nq:30o,._eeee 500 Ratio _• eee••

CFC-I 2 ß 400 • o o ø

300 • øø o o •••

• ß o •ø 200 • o• ooO•

• o ß o oø •00 • _oo o* •o oø

_coo" _• o ...... • •0 1S•O 1S70 1S80 • SSO

0.6

0.5

o

0.1

0.0 2000

Year

Figure 1. CFC- 11, CFC- 12, and the CFC- 11 :CFC- 12 ratio versus time for the Northern Hemisphere troposphere. The data for these plots were provided by Walker et al., [this issue].

40

35

• 30 • 25

•, 20

•. 15

•: •o

<1: 5

o o o

o CFC-11 o

o

o

O 0 0 000000 eeee e o o Oooo

eeeeeeeeeeeeeøOOo

CFC-12 eeeeeøe•/•ee

'5 , I I I I , 1 •50 1960 1970 1980 1990 2000

Year

Figure 2. Annual percent change in CFC-11 and CFC-12 versus time for the Northern Hemisphere troposphere calculated from the data plotted in Figure 1.

pCFC age [Fine et al., 1988; Doney and Bullister, 1992; Warner et al., 1996], is valid if no mixing occurs after the water parcel leaves the surface, which is usually not the case. The CFC-1 l:CFC-12 ratio changes with time prior to the late 1970s, and it can be used for age estimation from the 1950s to that time [e.g., Weiss et al., 1985]. The ratio age is not affected by mixing with CFC free water. This is the case for the leading edge of the CFC signal as it enters the deep ocean in boundary currents. However, Pickart et al. [1989] and Rhein [1994] have shown that after adjacent water begins to accumulate CFCs by mixing, the ratio is underestimated and the age is overestimated for the core of the boundary current because of this mixing.

3. Observations of Anthropogenic Transient Tracers in NADW

It has been known for some time that NADW is produced at high latitudes in the North Atlantic Ocean from a combination of deep convection of dense water that forms at the surface and inflow of dense water from behind the Greenland-Iceland-Scotland Ridge. Measurement of transient tracers in the Noah Atlantic has enhanced

our knowledge of the pathways by which newly formed NADW enters the Atlantic Ocean and the rate at which newly formed water spreads along these pathways (for review see Fine, [1995]). The first transient tracers to be measured on a large-scale were tritium and bomb radiocarbon. A zonal section of both substances in the western Noah Atlantic was measured in 1972 on the Geochemical

Ocean Sections Study (GEOSECS) expedition and clearly showed the input of recently ventilated water to the deep Atlantic at high latitudes [Stuiver and Ostlund, 1980]. Tritium and bomb radiocarbon were observed throughout the water column noah of 45øN with a maximum at the bottom extending northward to the Greenland-Iceland Ridge [Ostlund and Rooth, 1990] where dense water enters the Atlantic Ocean through Denmark Strait. This same zonal structure was observed again in the early 1980s on the Transient Tracers in the Ocean ('I'TO) expedition, but the deep tritium and bomb radiocarbon concentrations had increased and

extended farther south [Ostlund and Rooth, 1990]. Tritium measurements made at the Blake-Bahama Outer Ridge in 1977 revealed high levels in the Deep Western Boundary Current (DWBC) and demonstrated for the first time that NADW no older than 15 years was transported from the northern formation regions to the subtropics [Jenkins and Rhines, 1980]. The presence of

SMETHIE ET AL.: TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER 14,299

recently formed NADW in the subtropical Atlantic DWBC was confirmed by tritium [Olson et al., 1986; Doney and Jenkins, 1994] and CFC [Fine and Molinari, 1988; Smethie, 1993] measurements made in the early and mid-1980s, and these results will be discussed in detail later in the paper. The presence of recently formed NADW in the tropical Atlantic was shown by Weiss et al., [ 1985], Molinari et al., [1992], Rhein et al., [1995], and Andtie et al., [1998], and in the South Atlantic was shown by Wallace et al., [ 1994].

Before proceeding with the details of the uptake and spreading of transient tracers in NADW an overview is presented of where these tracers are currently found within the NADW components. This can be seen using two zonal sections of CFC-11, one in the subpolar North Atlantic extending from the Irminger Basin eastward and one in the subtropical North Atlantic along 24øN (Plate 1). The highest CFC-11 concentrations are found in the near-surface water along both sections, which is expected since CFCs enter the ocean at the surface; however, here we are interested in the concentrations in the deep water components.

Along the subpolar section, high CFC-11 concentrations extend quite deep (-2000 m) west of the Mid-Atlantic Ridge. The water between -500 and -2000 m is Classical Labrador Sea Water

(CLSW) that most likely advected into the Irminger Sea from its formation region in the Labrador Sea. CLSW is also observed east of the Mid-Atlantic Ridge as a layer of nearly homogeneous CFC- 11 concentration between -700 and 1500 m, although the concentration is less than west of the ridge. In the Irminger Basin there is a maximum at the bottom along the Greenland continental rise extending to the middle of the basin. This is Denmark Strait Overflow Water (DSOW) that flows over the Greenland-Iceland Ridge. In the eastern basin there is a similar maximum at the bottom along the eastern flank of the Reykjanes Ridge. This is Iceland-Scotland Overflow Water (ISOW) that has flowed across the Iceland-Scotland Ridge. It has a lower CFC concentration than the DSOW in the western basin. ISOW flows through the Charlie Gibbs Fracture Zone into the western basin just south of this section. Tritium measurements nmde along a similar section during TTO in 1981 [Ostlund and Root& 1990] show basically the same structure.

There are two prominent subsurface CFC-11 maxima in the western basin along the subtropical section (Plate lb). Both are intensified in the west and extend well into the interior of the

western basin. Initially the upper maximum appeared to be CLSW and the deep maximum DSOW or ISOW. As will be elaborated on below, the deep maximum is a mixture of DSOW and ISOW, which will be referred to as Overflow Water (OW), but the upper maximum lies at a lighter density horizon than CLSW. This latter water mass, which has only recently been recognized, is Upper Labrador Sea Water (ULSW), which appears to form in the southwestern Labrador Sea [Pickart, 1992a; Pickart et al., 1996].

4. Formation of NADW

NADW is a complex of several water masses, as was first demonstrated by Wiist [1935] who, on the basis of salinity and oxygen concentrations, classified NADW into three types, upper NADW, middle NADW, and lower NADW. Today, it is known that there are four primary North Atlantic water masses that make up NADW, two that form by deep convection in the open ocean and two that are derived from the overflow of dense water across the

Greenland-Iceland-Scotland sill. Wiist [1935] also recognized that there are deep waters of non-North Atlantic origin that influence its characteristics. Water from the southern ocean, which is often referred to as Antarctic Bottom Water, and Mediterranean Overflow Water are the two most important.

The least dense components of NADW form by deep open ocean convection during winter. The best known of these components is CLSW, which forms by convection extending to

depths as great as 2000 m in the central Labrador Sea [Wallace and Lazier, 1988; Lazier, 1995; Dickson et al., 1996]. This water mass is characterized by a relatively low salinity acquired from inflow from the Arctic Ocean and a low potential vorticity resulting from the breakdown of vertical stratification caused by deep convection [Talley and McCartney, 1982]. The characteristic potential temperature, salinity, and density are 3.5øC, 34.88, and o•. 5 = 34.66, respectively. However, conditions influencing CLSW formation, which include preconditioning water in the central Labrador Sea during the previous year and the strength of the Noah Atlantic Oscillation [Curry et al., 1998], vary from year to year. Thus the temperature, salinity, and density can vary from year to year, and in some years, surface water does not become dense enough to form CLSW [Lazier, 1995; Dickson et al., 1996].

ULSW has only recently been recognized as a distinct water mass and a component of NADW. Its discovery stems from observations of an extensive subsurface CFC maximum at 1200-

1500 m in the subtropical North Atlantic during the 1980s with a potential temperature range of 4ø-5øC [Weiss et al., 1985; Fine and Molinari, 1988; Smethie, 1993] (Plate lb). This feature is also observed as a maximum in vertical profiles of tritium measured near the western margin of the subtropical Atlantic in the 1970s and 1980s [Jenkins and Rhines, 1980; Ostlund, 1984; Olson et al., 1986; Ostlund and Grall, 1987; Pickart et al., 1996], but their significance was not recognized until the more extensive CFC data sets were collected. Pickart [ 1992a] was the first to recognize that this was not a variety of CLSW but a distinct water mass that formed at a density less than CLSW and hence referred to it as ULSW. (This water mass has also been referred to as Shallow Upper NADW by Rhein et al., [1995]). Recently Pickart et al., [ 1996] observed newly formed ULSW near its source region in the southern Labrador Sea. They observed a small weakly rotating eddy that had an anomalously low temperature and salinity (2.9øC and 34.78) and high CFC and tritium content, embedded in warmer and saltier water of the same density flowing equatorward in the DWBC. Although its potential temperature is similar to that of CLSW, its salinity is significantly fresher, and its density is in the range of the 4ø-5øC water with high CFC and tritium concentrations observed in the subtropical Atlantic. This eddy was being eroded rapidly by lateral mixing, and its lifetime was estimated to be several months. Eddies such as this one apparently form by deep convection during winter near the southwest •nargin of the Labrador Sea, and Pickart et al. [1997] have shown that this could occur in the Labrador Current. There they quickly become entrained into the equatorward flow of the DWBC, are capped by less dense water, and become completely absorbed in the DWBC by the time it flows around the Grand Banks. The end result is a water mass that is warmer, more salty, and less dense than CLSW.

The densest components of NADW form behind and overflow the Greenland-Iceland-Scotland Ridge. They are ISOW that enters the eastern Noah Atlantic across the Iceland-Scotland Ridge and DSOW that enters the western Noah Atlantic across the Greenland-

Iceland Ridge. On the basis of tritium measurements from the early 1970s, Swirl et al. [1980] showed that DSOW is much better ventilated than ISOW, which will be elaborated on below.

ISOW originates from a density horizon of ~o0 = 28.06, which lies near 900 m in the Norwegian Sea and approaches the surface in the Greenland Sea [Swirl, 1984]. The primary input of this water (-0.55øC and 34.91 in 1981 [$methie et al., 1986]) to the northeastern Atlantic is thought to be the Faeroe Bank Channel with a sill depth of 850 m, but some flow also occurs across the shallower Iceland-Faeroe Rise. This water has relatively low anthropogenic tracer concentrations [Bullister, 1984; $methie et al., 1986; $methie and Swirl, 1989], and a residence time of ~45 years behind the Iceland-Scotland Ridge has been estimated from these low concentrations [$methie and Swirl, 1989; $methie, 1993]. ISOW entrains surrounding water after entering the eastern North

14,300 SMETHIE ET AL.: TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER

Atlantic. Swirl [1984] deduces that a mixture of roughly 60% pure ISOW/40% northeast Atlantic water forms, which then flows into the western North Atlantic through the Charlie Gibbs Fracture Zone. CLSW is also entrained into this mixture but primarily after the water enters the western basin. Harvey and Theodorou [1986] conclude that CLSW is entrained prior to entering the western basin and that this water is a mixture of 45% pure ISOW, 25% CLSW, 21% northeast Atlantic water, and 9% North Icelandic and Arctic Intermediate Water. Deep water masses can also contribute, and McCartney [1992] has shown that the southward flowing water along the eastern fl•nk of the Reykjanes Ridge contains a component of deep water that has been advected from the Southern Hemisphere. The entrained northeast Atlantic water is well ventilated down to the depths of inflow of pure ISOW (-850 m) by deep convection during winter [Harvey, 1982; Robinson et al., 1980]. Swirl [1984], Srnethie and Swirl [1989], Srnethie [1993] and Doney and Jenkins [ 1994] have shown that this entrainment is also the major source of transient tracers in water flowing into the western basin through the Charlie Gibbs Fracture Zone. This water is less dense than DSOW entering the western basin and overrides it as it flows cyclonically around the Irminger Basin.

DSOW is the densest, coldest, and freshest component of NADW. It forms behind the Greenland-Iceland Ridge and enters the western North Atlantic through Denmark Strait, which has a depth of-600 m. Swirl et al. [ 1980] have provided evidence that DSOW is formed primarily from upper Arctic Intermediate Water (AIW) (-0.5øC and 34.75-34.85). Upper AIW is found in the upper few hundred meters of the western Greenland Sea and the Iceland Sea

where it outcrops at the surface during deep convection in winter. A denser water mass also appeared to contribute -10% to the formation of DSOW. Newly formed DSOW would be expected to be well ventilated and have a high anthropogenic tracer content because of the exposure of upper AIW to the atmosphere. This was confirmed by the observation of high tritium concentrations during the GEOSECS expedition [Swirl et al., 1980] and can be seen in the high CFC-11 concentration at the base of the Greenland continental slope at 60øN (Plate lb). Livingston et al. [1985] and Srnethie and Swift [1989] showed that the DSOW layer 450 km downstream of Denmark Strait was no older than 2 years, demonstrating its rapid input to the North Atlantic from its formation region.

The formation of both ISOW and DSOW discussed above

involves the conversion of surface water to deeper water in the Greenland/Iceland/Norwegian region. This water flows across the Greenland-Iceland-Scotland Ridge with ISOW residing much longer behind the ridge than DSOW before overflow. Recently, Mauritzen [1996] has proposed that the overflow waters originate primarily in the Arctic Ocean. The source water is warm salty surface water that enters the Arctic Ocean from the North Atlantic in

the Norwegian Atlantic Current. This water loses its buoyancy and enters the Arctic Ocean in two branches, the Fram Strait Branch and the Barents Sea Branch [Schauer et al., 1997]. Both branches are modified in the Arctic Ocean and flow back toward the North

Atlantic as denser water masses through Fram Strait. The Barents Sea branch flows into the Norwegian Sea and feeds ISOW. The Fram Strait Branch flows toward the Denmark Strait along with Atlantic water that has recirculated within Fram Strait to feed

DSOW. Mauritzen [1996] refers to this water as Arctic Atlantic Water. This mechanism provides a steady (nonseasonal.] input of water of the appropriate density to the overflow regions. Dickson and Brown [ 1994] showed that the flow of DSOW was steady with respect to season from 1986 to 1991. However, using recent hydrographic and acoustic Doppler current profiler data, Bacon [1997] suggests that this transport decreased significantly after the early 1990s, and Dickson et al., [1999] present more recent data indicating that the characteristics of DSOW have changed since the late 1980s.

The exact mechanism by which DSOW forms is not known, but there is evidence supporting both mechanisms discussed above.

It could also be formed by a combination of both processes. For example, Strass et al., [ 1993] have presented evidence that a portion of DSOW forms by mixing, within the East Greenland Current, between upper AIW and Atlantic water that has recirculated within Fram Strait.

All of the components of NADW either form in the subpolar western North Atlantic or are advected to this region. DSOW entrains overlying Gibbs Fracture Zone Water (GFZW) as it enters the North Atlantic and further mixes with it as both water masses

flow along the west Greenland continental slope and then around the Labrador Sea in a deep western boundary current (also referred to as the Deep Northern Boundary Current by McCartney [ 1992]). DSOW also entrains some Antarctic Bottom Water that has entered

the region either through the Charlie Gibbs Fracture Zone or from a deep flow along the western flank of the Mid-Atlantic Ridge [McCartney, 1992]. DSOW may also entrain some CLSW and less dense local water masses as it flows into the Atlantic just south of Denmark Strait. GFZW is sandwiched between DSOW below and CLSW above and thus mixes with both of these water masses.

Finally, ULSW enters the system near the outflow from the subpolar Atlantic to the subtropical Atlantic.

5. Construction of Sections and Maps

These results are based on a synthesis of most of the CFC data collected in the North Atlantic through 1992. Since our analysis is focused on the basin-wide distribution of CFCs, we have relied mainly on data collected in the late 1980s and early 1990s to obtain a quasi-synoptic picture of the three-dimensional distribution of CFCs in the North Atlantic Ocean. A listing of all of the data used in this analysis is given in Table 1.

To illustrate how CFCs track the flow of the various

components of NADW in the DWBC, 12 CFC-11 sections normal to the flow of the DWBC (Figure 3) taken between 1988 and 1992 are presented (Plate 2). These sections start near the origin of the DWBC in the subpolar North Atlantic and extend to 10øS. Each section is accompanied by a plot of O/S and 0/CFC-11 (Figure 4) to illustrate how the various components of NADW fit into the large- scale hydrographic structure and are modified during the equatorward transit. In these plots the cores of the various water masses are highlighted as described in the Figure 4 caption, and a summary of the core properties is given in Table 2. These CFC-11 data have not been normalized to a common date, which must be

taken into account when comparing CFC concentrations between sections. CFC-12 data for these sections show essentially the same features and are not presented.

Although CFCs have been measured in much of the North Atlantic, there has not been a synoptic survey, and the basin-wide distribution can only be mapped using data from a number of different cruises taken during different years. This presents a problem when using CFCs and other transient tracers since the input varies with time and the distribution is not in steady state. To investigate the basin-scale circulation pattern of recently formed NADW, we prepared maps of CFC-11 for the core of ULSW and the core of Overflow Water (OW) (Figure 5) using data collected between 1988 and 1992 and one cruise in 1986 (Table 1). The maps were constructed by finding the maximum concentration at each station between 34.5 and 34.7 Ol.5 with 0 >3.5øC for ULSW and between 45.8 and 45.9 o4 below 2500 m for OW. The maximum concentrations were then multiplied by a correction factor (given in Table 1) to normalize the concentrations to 1990. All the data used were adjusted to 1990 because much of the data were taken in that year, and it is also the midpoint of the 1988-1992 time period. The normalization factor was determined as follows. The profiles of CFC-11 and CFC-12 in each of the upper and lower maxima were vertically integrated, and the CFC-1 l:CFC-12 ratio was calculated for the integrated values. This seawater ratio was

SMETHIE ET AL.: TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER 14,301

Table 1. Data Used in This Study

Correction Principal Cruise Reference a Stations Dates U/L b Use • Investigator

STACS 3 1 1-61 2/89-3/89 1.22/1.20 CU,CL,AU,AL,VS R. Fine 65-79 1.17/1.17

STACS 4 2 1-67 6/90-7/90 1.00/1.00 CU,CL,AU,AL R. Fine Trident 3 1-39 8/92-9/92 0.74/0.74 CU,CL, AU,AL, VS R. Fine

40-61 0.71/0.70 62-71 0.74/0.74

OCE 134 4 1-76 6/83-7/83 NA AU,AL W. Smethie WBEX 4 1-51 4/86-5/86 __a CU,CL W. Smethie

18-51 AU,AL

EN 214 5,6 1-44 6/90 1.0/1.0 CU,CL,AU,AL,VS W. Smethie EN 223 7,8 1-10 3/91-4/91 0.95/0.86 CU,VS W. Smethie

11-44 0.95/0.95 CL,AL HE 06 9 1-74 7/92-8/92 0.71/0.70 CU,CL,AU,AL, VS W. Smethie

75-112 0.74/0.74 METEOR 14 10 627-682 10/90 1.0/1.0 CU,CL,AU,AL M. Rhein METEOR 16 10 286-343 5/91-6/91 NA AU,AL. M. Rhein METEOR 18 11 558-622 9/91 0.95 CL,VS W. Roether

A. Putzka

METEOR 22 10 474-537 10/92-11/92 0.70/0.70 CU,CL, AU,AL,VS M. Rhein 3'TO NAS_leg 7 12,13 220-249 9/81-10/81 NA AU,AL. R. Gammon TTO TAS_leg 1 13 4-32 12/82 NA AU,AL. R. Weiss TFO TAS_leg 2 13 55-94 12/82-1/83 NA AU,AL. R. Weiss SAVE 1 14 11-170 11/87-3/88 1.39/1.39 CU,CL, AU,AL R. Weiss (Legs 1,2,3) W. Smethie SAVE 2 14,15 309-379 4/89 1.20/1.20 CU,CL, AU,AL R. Weiss (Leg 6) W. Smethie HUDSON 92014 16 1-52 6/92 0.91 CL,VS E.P. Jones OCE 202 17 1-17 7/88-8/88 NA VS J. Bullister

al, Molinari et al., [1992]; 2, Johns et al., [1997]; 3, R.A. Fine (personal communication, 1999); 4, Smethie [1993]; 5, Pickart and Srnethie [1993]; 6, Pickart et al., [1992]; 7,Pickart et al., [1996]; 8, McKee et al., [1995]; 9, Bryden et al., [1996]; 10, Rhein et al., [1995]; 11, W. Roether and A. Putzka (personal communication, 1999); 12, Physical and Chemical Oceanographic Data Facility [1986]; 13, Weiss et al., [1991]; 14, Weiss et al., [1993]; 15, Srnethie et al., [ 1992]; 16, E.P.Jones (personal communication, 1999); and 17, Doney and Bullister [ 1992]. •dultiplication factors used to adjust the CFC concentrations to 1990, for the upper CFC maximum (U) and lower CFC

maximum (L). See the text for detailed explanation. cC denotes that the cruise was used in a map of CFC-11 concentration; A denotes usage in a map of CFC-11/CFC-12

derived age; U denotes usage in a map of the upper CFC maximum; L denotes usage in a map of the lower CFC maximum; and VS denotes usage in a vertical section of CFC-11. dMultiplication factors used to adjust the WBEX CFC concentrations to 1990 for the upper CFC maximum: stations 1-9,

correction 1.54; stations 10-18, correction 1.74; stations 19-22, correction 2.08; stations 23-25, correction 1.98; stations 26-36, correction 2.05; station 37, correction 1.83; stations 38-49, correction 1.92; and stations 50-51, correction 1.82; and for the lower CFC maximum: stations 5-12, correction 1.80; stations 13-17, correction 1.90; stations 18-23, correction 2.00; station 24, correction 2.10; station 25, correction 2.30; station 26, correction 2.10; stations 27-28, correction 2.00; station 29, correction 1.90; stations 30-31, correction 2.00; stations 32-33, correction 2.10; and stations 34-50, correction 1.95.

then converted to an atmospheric ratio by dividing by the CFC- 1 l:CFC-12 solubility ratio [Warner and Weiss, 1985], and the year of formation was determined from Figure 1. The annual percent change for the year of formation was estimated from Figure 2, and this rate of change was multiplied by the time difference between i990 and the cruise date to determine the normalization factor. For

water formed since 1978 it was not possible to calculate the year of formation using the CFC-1 l:CFC-12 ratio because the ratio was nearly constant from 1978 to 1990. The average annual rate of change for CFC-11 from 1978 to 1990 was -4.5%, and this rate of change was used to determine the normalization factor for water formed after 1978.

Age maps (Figure 6) were constructed using data acquired between 1981 and 1992 at the maximum CFC- 11 concentrations in ULSW and OW for the intervals given above. The seawater CFC- 11 :CFC-12 ratio was calculated for each station and converted to an

atmospheric ratio of the partial pressure of CFC-11/CFC-12 as described in the previous paragraph. The resulting partial pressure ratios were then compared with the Northern Hemisphere atmospheric time history of the ratios (Figure 1) to give the year of formation. The year was subtracted from the cruise date to give an age. Ages were not estimated when the concentration of either CFC was <0.02 pmol kg '• because of the large error in the ratio at these low concentrations. Also, ages were not calculated for the subpolar

14,302 SMETHIE ET AL.' TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER

-...--...--.:.z..•'? :::..-..-•: ': -':•. •.• .......

.... ...... .........

. -. '" ' 7:;:;". -• ': '• .... :*:6•":':'.:. " .. .

"7.:. •. ';'-:?'"': -;: . .... ..

•:: '•. .

../½ ;..-;.*:' ::-. - %

5O

**• :-.,:. ,:-:'?-- ::

**7******* '• '•**':"}'::": .

......... •i:i;•!' ..-•-.x-.-..--9• • ":::::.:'•: '

' .:::• :: • ..,:½'.*' ::-:;::(;• •- .-%.,

,..:,.,... :--. .....

:...:.. •..•.•..,.

::' :,- *%X'"'"::::•.: ;: .*' -- •:*•. ...,z;,-,,':.•s,•-,:::•:? ...,'•**:, ,: ,.

..,::.,:..

.'.....x:...-- ........

. . ,. :.,.

•. •.:..:,1,:•;,½,:.::---:-• .: ::

2O

.......

......:

10

2O

80 70 60 50 40 30 20 10 0 10

Figure 3. Station location map for 12 sections that cross the flow of North Atlantic Deep Water (NADW) in the Deep Western Boundary (DWBC). The sections have been labeled A-L. Data sources are as follows: section A, Oceanus-202, July 1988 [Doney and Bullister, 1992]; sections B and C, Meteor 18, September 1991 (A. Putzka and W. Roether, unpublished, 1999); section D, Hudson 92014, June 1992 [Pickart et al., 1996]; sections E and F, Endeavor 223, April 1991 [Pickart et al., 1996]; sections G and H, Endeavor 214, July 1990 [Pickart and Srnethie, 1993]; section I, Trident, August 1992 (R.A. Fine, unpublished, 1999), section J, STACS 3, February 1989 [Molinari et al., 1992]; and sections K and L, Meteor 22, November 1992 [Rhein et al., 1995]. Also plotted are station locations (stars) for the subpolar (Meteor 18) and subtropical (Hesperides 06) trans-Atlantic sections presented in Plate 1.

region because ULSW is not found in this region and the CFC- 1 l:CFC-12 ratios in OW for this region indicated that this water formed after 1978 when the ratio did not increase monotonically with time.

Maps of dilution factors (Figure 7) were constructed using the age data presented in Figure 6. CFC ratio-derived ages were converted to concentrations using the Northern Hemisphere atmospheric time history (Figure 1) and the Warner and Weiss [1985] solubility function with the estimated potential temperature and salinity at the time of formation. For ULSW a formation temperature and salinity of 2.9øC and 34.78 were used on the basis of the observations of Pickart et al. [1996]. For OW the O/S characteristics of pure DSOW (-0.5øC and 34.80) were used because most of the tracer contribution is from the DSOW component (see discussion that follows on OW and Smethie and Swirl [1989] and Smethie [1993]). These calculated concentrations were divided by the measured concentrations to get a dilution factor. Only data with CFC-11 and CFC-12 concentration >0.02 pmol kg '• were used. The dilution factors were then multiplied by 0.7 for both ULSW and OW to take into consideration the best estimate of the percent equilibration with the atmosphere at the time of formation [e.g., Pickart et al., 1996] (see also the following discussion on DSOW).

Data from the 20øW section occupied in 1988 [Doney and Bullister, 1992] are not used for the maps because ULSW and the mixture of GFZW and DSOW that form OW is confined mainly to the western basin. At the far northern end of the 20øW section there

is water with a CFC maximum near 36.56 o•.s, and it has OIS properties similar to ULSW. However, the water derives its properties from incomplete mixing between ISOW and northeast Atlantic water [Doney and Bullister, 1992]. Also, Pickart et al. [ 1996] find that the eastward extent of ULSW is largely blocked by the North Atlantic Current. Along 20øW between 45 ø and 25øN, water at the density of oks = 36.56 is warmer and saltier than ULSW and is apparently mostly Mediterranean Water. Although there is a CFC maximum at the bottom in the northeastern basin, the OW in the western basin derives most of its CFC signal from DSOW, so these concentrations are not included in the map.

6. Discussion

The equatorward pathways followed by the NADW components are quite complex. There is a decrease in the CFC concentrations in the equatorward direction that reflects

SMETHIE ET AL.' TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER 14,303

(A) M18

CFC-11 (pmol/kg) , ........... ., ,•- 2.50 ß .il ' : .' : '.• . 'i

ß i.. 0% 2.00 ß

ß . .. 211 .• . . ß :

1.50

12.00 .- : •.00

0.50

0 1000

DISTANCE FROM STN. 558 (km)

0.00

(B) He-06

CFC-11 (pmol/kg)

-2

-4

'"1 i i ß ß ß - ',, , .n==a, 'ql.•5' ' ' " ,.,- :.. . . 0.10 . . ,. ' ' .

ß • 0 ' i, ß . . ß .. ß

0.2 0:. . .

'' i

,, l,

ß ß

ß

ß '0.05' ' ' o 5

-5 . . . .... .o •1o

ß

-6

,t. 5 o .

o 1000 2000 3000 4000 5000 DISTANCE FROM STATION 101 (km)

Plate 1, Vertical sections of CFC-11 in (a) the subpolar North Atlantic Ocean and (b) the sub-tropical North Atlantic Ocean. The subpolar section is from the Meteor 18 cruise in 1991, and the subtropical section is from the Hesperides 06 cruise in 1992. See Figure 3 for station locationsß

14,304 SMETHIE ET AL.: TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER

0

-I

124

:

O0 .

-2

-3

-1

I00 2oo 300 4o0 50o 600

DISTANCE (kin) 700

-5

3.5O

I

1oo 200 3.00 400 500

DISTANCE 0tin)

(D)

.35

,

3.L66 . ß

3.00 o

-.3 ß

3.5

3.0

2.5

2.O

1.5

- 1.0

O.5

-5

100 200 300 400 500 600 0 100 200 300 400 DISTANCE (km) DISTANCE (km)

oo

Plate 2. Vertical sections of CFC-l I (pmol kg -•) across the DWBC. See Figure 3 for section locations. Five density lines are also plotted. They are, from shallowest to deepest: 34.56 and 34.66 c•.5, 45.82, 45.86, and 45.90 c•4.

SMETHIE ET AL.' TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER 14,305

-1

.-4

-1

-5

41

- 0.80

-3

-5

.

t .60

No

IOO 200 300 400 500

DISTANCE (kin)

(G) 38

2 ß

.

' O0 C60

100 200 300 DISTANCE (kin)

0 100 200 300 DISTANCE (kin)

(H) 10 7 4 1

o

•.:• -oo- • -

0.60 o

'"' --"'• i U

-4 0.0

0 100 200 300 DISTANC• (kin)

o) 1 s 8 11

o

,. . .

-I • '•

!:•0 ' •.' 0.40

-2 1•'' ., • . ß

• i

-3

4 m

-5

0.30 ß

0. 5.90

ß .

o IOO 200

DISTANCE (km)

Plate 2. (continued)

14,306 SMETHIE ET AL.: TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER

0 j

-1

-5

0

0 S

-1

(J) (L) 48 46 44 $37 $36 $34 $32

n " " , , 0 [ ' ' ' .......... .

No [ No Data ß Dntn .

ß

0.10 • •: ' ß

.,4

0.04 () 04

08 o.o

104 -2

0.08

01 i !

-5

100 200 300 400 500 600 700

DISTANCE (km)

508 5

.

503 501 499 (K)

No Data

495

I t,

0.04

100 200

DISTANCE (km)

492

No Data

53o

300

-5

100 200 300 400 500 600 700

DISTANCE (kin) 800 100o

Plate 2. (continued)

SMETHIE ET AL.' TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER 14,307

a 9 Section A (Northeast Atlantic) Section A (Northeast Atlantic)

8

7

6

.-. 5 o .•.•

f4 3

2

b 8 Section B (Northeast Atlantic) Section B (Northeast Atlantic)

CFC-11 (pmol/kg) '•4.7 34.8 34.9 35.0 3•.10 ! ' •

Salinity (psu)

Figure 4. Potential temperature/salinity and potential temperature/CFC-11 plots for the sections presented in Plate 2. Lines of constant density are plotted on the potential temperature/salinity plots, oi.5 is used for the upper portions and o4 is used for the lower portions. Potential temperature/salinity ranges are plotted for (a) pure Iceland-Scotland Overflow Water (ISOW) and Northeast Atlantic Water (NEAW), and (c) Upper Arctic Intermediate Water (UAIW) and Arctic Atlantic Water (AAW). Note that the salinity scale for Figure 4a is shifted to a higher range than in Figures 4b-41 to include NEAW. Points for the cores of the following water masses are designated by different symbols as follows: Classical Labrador Sea Water (CLSW), square; Upper Labrador Sea Water (ULSW), circle; section E western boundary slope water, triangle; ISOW, inverted triangle; Gibbs Fracture Zone Water (GFZW), star; incompletely formed ISOW, diamond; Denmark Strait Overflow Water (DSOW), hexagon; and Antarctic Bottom Water (AABW), plus sign. See text for discussion.

the temporally increasing atmospheric source function, advection, mixing, entrainment, and recirculation in gyres. The sections presented are from Iceland to 10øS (Plate 2) and have several features in common. The highest concentrations are observed in the inshore region coincident with the DWBC. Intermediate concentrations are observed in the

adjacent region near the interior, where the spreading of tracers is predominantly in deep gyre recirculations. Seaward of the recirculations, concentrations are near blank levels, and the tracer distribution is probably dominated by lateral mixing. The discussion below includes references to both the

sections and maps (Plate 2 and Figures 4-7). Note, however, that the map concentrations were normalized to the year 1990, but the section concentrations are presented as observed.

6.1 Upper and Classical Labrador Sea Water

6.1.1. Subpolar to 38øN. CLSW has high CFC concentrations and is observed as a maximum along all of the sections to 38øN (Plate 2, C-F). The highest CFC concentrations and lowest 0 and salinity values in CLSW are found on section D extending from the

14,308 SMETHIE ET AL.: TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER

(2 8 Section C (Irminger Sea)

5

34.8 35.0

Section C (Irminger Sea)

34.9 35.10 1 2 3 4 5 Salinity (psu) CFC-11 (pmol/kg)

Section D (Labrador Sea)

6

5

7•-• -34'5

4.7 34.8 35.0

Section D (Labrador Sea)

• I * ' GFZW

i

34.9 35.10 I 2 3 4 5 Salinity (psu) CFC-11 (pmol/kg)

Section E (Newfoundland Basin) Section E (Newfoundland Basin) e 8 ' ' i . ' I ' i

' T

,

ß . ... :... i .. , 4 •' '"• ' 7: '

'-" ' ' -*--O'--"-:. :;: • • . o:; ..'" ':i •'"•

] • •'."":' :.' '.""•':", i. ,%,

' I 5 Salinity (psu) CFC-11 (pmol/kg)

Figure 4. (continued)

SMETHIE ET AL.: TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER 14,309

Section F (55' W)

I- •'-• I J.-• • --•,

2

Salinity (psu)

g 8

t4'.7

Section G (North of Gulf Stream Crossover)

]4.8 ]4.• ½•.o •.8.o

Salinity (psu)

h 8

-•4'.7

Section H (South of Gulf Stream Crossover)

....... i F- 34.1• i, _.-• ...... ,.

34.8 34.9 35.0 35.0.0

Salinity (psu)

Section F (55' W) t '

Section G (North of Gulf Stream Crossover) i

0.5 1.0 1.5

CFC-11 (pmol/kg)

Section H (South of Gulf Stream Crossover)

• © ß DSOW•

,. e ß UL-SWI

•.•o CFC-11 (pmol/kg)

2.0

2.0

Figure 4. (continued)

14,310 SMETHIE ET AL.: TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER

Section I (24' N)

34.7 34.8 34.9 35.0 35.0.0

Salinity (psu)

Section I (24' N)

I.

I

, I

CFC-11 (pmol/kg) 2.0

j 8 Section J (5-10 ø N)

-h'.7 35fl.00

Section J (5-10' N)

i

,

, © 'IDSO,W •ULSW

I •

I 0.•$ 0.12 0.04

CFC-11 (pmol/kg) 0.16 0.20

k 8 Section K (35' W)

, •' 't 3 .9 !'- - ......

_ _--.-34.1 _L ........ •"

!"'' '!': ' _ % - "-7:"•"•: _; 34.5 -4

_ 4-34.7 ] ,;--_• ,,•• J ....... '-'•I - . ....... ---r------ -

.

-•--.•..•,..•46.0 •-' , ......

...... :•__. _-i --, ; .- 46.2

34.8 34.9 35.0

Salinity (psu)

Section K (35' W)

ß '•:.:

--;':.•;".""..t" ::' '

I ,

', !

i 6 i i 35 fl.00 0.04 O. 8 0.12

CFC-11 (pmol/kg) 0.16 0.20

Figure 4. (continued)

SMETHIE ET AL.: TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER 14,311

I 8 Section L (10' S)

-•4'.? 34.8 34.9 35.0 35 Salinity (psu)

Section L (10' S)

ß ' I I I ' ; ,, i c'DSOW • I

''4-+ I / e ' ULSW I I

2.00 0.04 0.08 0.12 0.16 0.20 CFC-11 (pmol/kg)

Figure 4. (continued)

Table 2. Core Potential Temperature, Salinity, Density and CFC-11 Concentration for the Vertical Sections Crossing the Deep Western Boundary Current

Potential

Water Temperature, Salinity, CFC- 11, Density Section Mass øC psu pmol kg' • (71.5 (74

A Incompletely Formed ISOW 4.3-4.6 35.00-35.01 2.5-2.8 34.51-34.59

CLSW 3.5-3.8 34.90-34.91 1.4-1.8 34.60-34.65 ISOW 2.7 34.98 2.4 45.81

B CLSW 3.3-3.6 34.88-34.89 1.7-2.3 34.62-34.65 ISOW 2.6-3.0 34.97-34.98 1.6-1.9 45.75-45.81

C CLSW 2.9-3.2 34.85-34.87 3.3-3.6 34.65-34.66 GFZW 3.0-3.3 34.99-35.00 1.3-1.6 45.68-45.71 DSOW 1.1-2.0 35.87-35.88 2.9- 3.1 45.88-46.02

D CLSW 2.9 34.88 3.6-3.8 34.68 GFZW 2.8-3.0 34.92 1.2-1.4 45.72-45.75 DSOW 1.8 34.89 2.1-2.3 45.91

E Slope Water 3.0-3.7 34.85-34.87 3.2-3.7 34.50-34.65 ULSW 3.2-4.0 34.80-34.82 3.1-3.5 34.49-34.60 CLSW 3.0-3.2 34.86-34.88 2.6-3.0 34.65-34.66 GFZW 3.2-3.4 34.94-34.95 0.5-0.8 45.65-45.69 DSOW 1.8-1.9 34.895 1.5-2.0 45.91-45.92

F ULSW 4.4-4.6 34.89-34.93 2.2-2.4 34.49-34.52 CLSW 3.4-3.9 34.88-34.92 2.4-2.5 34.59-34.64 OW 2.0 34.905 0.8-1.0 45.89

G ULSW 4.1-5.0 34.96-34.98 1.4-1.6 34.49-34.60 CLSW 3.8-3.9 34.95 0.9-1.3 34.62-34.65 OW 2.0-2.4 34.91-34.92 0.6-0.75 45.83-45.88

H ULSW 4.1-4.8 34.99-35.04 0.8-1.0 34.55-34.62 OW 1.9-2.5 34.895-34.93 0.4-0.5 45.82-45.90

I ULSW 4.2-4.9 35.005-35.04 0.6-0.7 34.55-34.62 OW 2.0-2.2 34.90-34.91 0.3-0.4 45.85-45.88

J ULSW 4.0-4.4 34.97-35.005 0.08-0.17 34.58-34.65 OW 1.9-2.1 34.900-34.905 0.06-0.09 45.86-45.89

K ULSW 3.6-4.3 34.98-34.99 0.08-0.11 34.59-34.69 OW 1.9-2.2 34.89-34.91 0.06-0.08 45.86-45.90

L ULSW 3.4-4.0 34.975 0.04-0.06 34.62-34.71 OW 1.9-2.3 34.89-34.91 0.02-0.03 45.84-45.88 AABW 0.2-0.4 34.70-34.71 0.03-0.04 46.03-46.04

14,312 SMETHIE ET AL.' TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER

...... ---:•? . ...•.

... .• .-..•

•: • ":.',•'•,.-•:.:.•,.:....•:•: . • ..... :: .:• :. : ..... --:.•::::.:--:: •.... ............. . .... --' . .;/•' :::-. • .•::.- ...... . ......... :• ........ ,....-:- •. •..:.•:•.:

-'•: • •:•":•-•-': • / • • .. .:'•' ? --:?..:...::...:%.:: .• : ,. •••• .: ...... •:::-:•:: ..... . .... ::•? ..... •..•:.?.•::.'.:.•..•:,?... •:•..•.•:-,.

--:::•::.:'-.•.; • * • • • :..:•:"•:',- .. :.-. --. , ........... - •:.•j-[•% ß . ..... : :?:.::-•:•: ..... •..• . .......... • *. •:• • ß ..... ;. '-:.• :? •:'•' ....L::.

:•:•.•:...•::•.'"•,' .'.•• •-.• '.:•--::' •:. .•.,.•.:• .......... • • • ..• :•:-::

..%..:...:-..•:•: .:•% o.• .. ..... ß ..• [..:.•:•: .:.•:'

,.. •..:::.... ,

i .... .....:?... '•%:. ':'::.•::L•'... .• .•:j'"'"':•:::.:•.. •:. • : . . •:._ . .

.... , :': ..:.' ....•.. - .... . • * • ...... :•:::.:•'

,.. . • (pmol/kS) /:- ...... -•--::• ::::::::::::::::::::::::::: .... ...... :•::::.•. •?'? ': •::•..:.:•::•:•:•: :: .•j::•i-•:::.

,%•:•:•::• ...... ...•:. . o o o ......... ..: .. ::•:•.-,•.

80 70 60 50 40 30 20 10 0 10

5O

4O

3O

2O

10

10

2O

Figure 5a. Lateral map of the maximum CFC-11 concentration in Upper Labrador Sea Water (ULSW). See Table 1 for data sources. These data have been adjusted to a common date of 1990 as described in the text.

80 70 60 50 40 30 20 10 0 10

Figure 5b. Same as Figure 5a but for overflow waters (OW).

:-:.

ß :.

.,

:::

5O

4O

3O

;::.:"-"•:'..-.-.'*-•;- 20 ß --• -.- --

•,,-•:-:•;•;:;..:: ............ ::,:;:-..•.-.• ....... "-- ,, - :; :.•.•;•'...

.... "Z'. ...... -• • •'"-";.:•: :. -• .... --,.*,,-...: •.,,

.... '4 .'.:%•& ß .,-..'-"' *

.. :':.•. .-....:::.:. :% • '•;'•:½•-%•:•:• .. ½ .: ....... .-.. . ..,•..• . ................. ,•..if:, ½- . ..::::.

....... •::;:• .......... : :.:,.,. •:::,-:-:::;;;•;::•-:::-:-... v,- ................. .,.;•:.-":•;:- .... ::'•;::•:::•2:•'•,:-: • •':½½;t4•::½;½ : ' • ½½•½' ' .......

-(:,-..• ,•, •:=- :::::•...%. - •:•; x :,.;::T;:. =.:; .' ½ ..... :. • •.•:•%:::..•.;•:;:,,.•:•.•-:"•...,. :.

,.'• . :..--•.•.•---:•.,: •:•.. . ...... ;.:: '%;;:,' :•.,:;.;...•: .:':'::•:•;&•,...: ............. .• •:.:. ........ . __

CFC age ?7' ,-•-- :...;. lO (years) .•-•? :• .... ;?•"

........ . ::

80 70 60 50 40 30 20 10 0 10

2O

Figure 6a. Lateral map of the CFC-11 :CFC-12 ratio age for ULSW. See Table I for data sources and the text for the procedure used to calculate ages. The dashed line represents the 0.02 pmol kg -I isoline. Ages were not calculated for CFC concentrations <0.02 pmol kg 'l because of the large uncertainty in the CFC-I I:CFC-12 ratio at low concentrations.

14,313

"o ,<,•....5•'•*,.•ii•i:t'::-'-'-':..:'"•.',;-'•'•i, ;.:."..•., .. .:•::•':'"•---' ..t !'•'-':"'•:%:.;;•:•'-.•:::;;:,::;iiii :.."::.Z::i::• •.., ..• •- '•,'.; ............. ii•;:":•ii• '""•ii:: :• -i: ,..:.-:'• -:½ 5i•-•-'-•.'•$....--'!•i::;!•i!E::E:::•:'•2:'4;-•:.: '-.4

' •..;• ""%::•"=:• .......... •='•:;7:.:: .......... '•,'•ii!iigii,.?• *' •':- • 50

-.½- -.½li.';½iii:;::i;:....--"::•i•Sii;i:.;•:.:;;•i:;::.i:½':-, *;. ':.. 7 •'• •.-:.•4;•.: ===================== %?. , .?........:•. • . ,.. :'¾ ".%d?•?:•½?-g"½*:---';•½•:•: :' ':..:.:?½•;:;j•'::½':;•%•;•,:::,.;•:.•::: •' ..... ;•:•;•.:*:.•;:•:':.."., '" ................ ,'.•-, t½

:--5•i. % -';:..::.:.:,:•:•}-" ' ß • .... ', ß • •.-..-•:- " '- ";;;;i•iii:½;'i;i•;;•D!!•:½•::;:• '.•*' '•.½ ::;•:' ,-•:;'5:::5....-:. '•":•::'•;.:;.;..':'•4;•i•i;-i,::•;•i:;'!.::•'•i :•:•:• ? •k':•o•;• ;'"" .• '• ß -.; .. • , o e -- 7:7:'•;::•' •i'•,•,•:ili½:;=:•:•;• ' ;.: •-;•,• ::;¾.•;•::,:•;•:::-:•..':':;:•"•:-•-• :- . • •. , .:.•:•..• e. . • • • ............ •.. ........... . . :: ,•,:::::... ..... . :, ..... :•:,:.,:::,..... • ... "";½i;;½4' •'.,:•i•;•:•?'" ½"..: • •' ............. d-,•;;;:'•':•'::";:;:; - ' '?'::':• :-";*" ............ ½0

...... ' .•':'•4:;½';,- '::•: ß . ß •-•:,:::;•:•J•i;•;::::;:;½:;•½;;5'*•'::::';"':;;;..:. ;;•:::55E5 •::' ........ ½•g4; ........ "•:•;:::::{"

.7 ?&" 7;:..t .. ½, -:?;;i;.'•:;;•:::;•:•;.;:.; :'.--:,-,• .-i5: .......

•:::•::•::•½•:E.'"':"--..5:-' .:-::'-• ",-'•.: '?:. :.. ": •:•::•:•, ,:g::J .......... :'?":,"'; ..... .:•-½g"•-';•.. 25 • ::""•.:.;•E :• .:•::::,. .•..•.%•..,......:•::•:: •.•:. . • .... .. •.,::•:•.•:•:•:;;•.•;•:::;•E•.•:? • "--•:-•5•%::;4:( •-•L-- ß ß • -'.• •..':•;•i•:."': ?•g;•;•: •'

• ;,:..Z 30 '• ':.-,.

7? . .,.:.:•: •.,•:.,,, ...... ::%:.,. . •. .- ,•.. • .•,.

ß ,•;• ......... -% . • •-;:•:½•'•:•:•.• ........... •:.:•:.:- •:::: ..•..-'-:

•:'::• .- :,,:•. • .:.•. -•. -.-:;F:;•::•;•Z:•- :,•:•;•-':•,. •g•;.:•.. • :..,:::::.:..-:, .................. .%.,•:.?::..

.... :•.?½•,• :•..•. ß -' • ......................

CFC age •. ..,/•.•,:•:•::•;;;•?:':. • ...... ? - :. :•- .: .•.:. ..::: ' ?:"•:::•-.: .....

.

. . .

• (years) ..d? >. " -:- ....... : .... •;::. % ... :., •:.•:.

80 70 60 50 40 30 20 10 0 10

Figure 6b. Same as Figure 6a but for OW.

14,314

• ....... ß .... • ?•-,:.•': ....•a• .:2.::•i%:'?

, ß "=., :.•:-•'=.;•;.:'.•:. '=• <•}F"•"• '•;=•m:.•/•:•:'• ............. -=:•:.==.:: '-. .... %; . .; -•?'"'•='•'•= '"'77.":C:::::::::::::::::::::::::::::::::::::::::::::•.' ........ "•}•.:•..

,• ........ ,,...:=.•:•::==<:i '•.•k.':. • , •'57•?' . •-*.-=a•*•()•:g•'•?•D;. • .......... -'•:• = =• '=•:"• ß • ' • '- '" • ß s :":•<=•-• • •g?.•;•:'= •½:=•gS•=•BE•:•B ' '=:::::•' .==37•-• ß-. ' t,. •, ............ . q •e=::=•'T'.:??•:'•:[75 ....

.........................................

• . • •...-

....

ß Z q:•:•='?•:==- ............... • .......... '5•:•:•:•'•:;: - .:,?:=::' *•:•= .x .-:.::•.-•?-,':,

ß y'=•-...:::....=..' ....... ----•=.•= ?=:, • ... .• .%• ...... :•=•. .... x •-.--.-%.. ;:

.. :.•..;--.

Dilution :.; .... ,- .•:•-•.• •:•,:-. .... ?:---: • [ ................... : ..... - .........

:-• .....

ß -•. •:.,: ULSW .',-:% ,/ :.,:•.:":•:•':'•':':'•'•;•i•'•:•:-•5• •':'•:::" .... ,, 20

80 70 60 50 40 30 20 10 0 10

Figure 7a. Lateral map for dilution factors for ULSW. See Table I for data sources and the text for the procedure used to calculate dilution factors. The dashed line represents the 0.02 pmol kg -I isoline. Dilutions were not calculated for CFC concentrations <0.02 pmol kg 'l because of the large uncertainty in the CFC- 11 :CFC-12 ratio and hence age at low concentrations.

..::..,• 1'•/... :..'...' '-.•i::"'::. '. :" -':•iC•'" o •, ø o ' o .... : ø t ........ • '•:. /' "t,. .,:

. . _- - _

:.. "--"'if i .. 4k,,,•... I •:• ..•..

i.:!:: I '• ...... ..

.... ß :..."'•... •%• •-- ß .::.

"-• '. "7": :::.'•' ß .... '"'•X -. •:... '• ...

........ ,.•

': 4j•..i:•. i

Dilution Factor

ow

80 70 60 50 20 10 0 10

Figure 7b. Same as Figure 7a but for OW.

5O

4O

3O

2O

10

10

2O

SMETHIE ET AL.' TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER 14,315

30

25

< 15 o

lO

e/

ß CFC-11.CFC-12 Age ß CFC-113:CFC-11 Age

0 ' i i i i i i i I i i i i i i 0 5000 10000 15000

Distance from southern Labrador Sea (km)

30

25

.-. 20

• 15

10

0 0

o ß ß

ß o Tritium/He-3 Age ß CFC-11:CFC-12 Age ß CFC-113:CFC-11 Age

i ! i ! ! i ! i , , i i

5000 10000 15000

Distance from Denmark Strait (km)

Figure 8, CFC- 11 :CFC- 12 ratio ages and tritium/He-3 ages [Doney and Jenkins, 1994] for the core of (a) ULSW and (b) OW in the DWBC versus distance from the source region.

Labrador coast into the central Labrador Sea (Plate 2d, Figure 4d, and Table 2), which is not unexpected since these stations are closest to the formation region in the central Labrador Sea. High CFC concentrations extend down to 2000 m depth indicative of the deep convection that forms CLSW. The most extreme values are located just northeast of the western boundary, indicating that the recently formed CLSW has become entrained in the equatorward flowing DWBC. However, high CFC concentrations and low potential temperature and salinity values are also found at the station in the central Labrador Sea. Potential temperature, salinity, and density of these waters are colder, fresher, and greater, respectively, than the typical values because of the intense winters that occurred in the region starting in the late 1980s with the increase in the North Atlantic Oscillation (NAO) index [Curry et al., 1998]. The CFC-11 concentration is at -62% saturation, similar to that reported by Wallace and Lazier [1988] for the summer of 1986. In the Irminger Sea (Plate 2c and Figure 4c) the CFC-11 concentration is similar to, and potential temperature and salinity slightly greater than, the values in the Labrador Sea. In the northeastern basin (sections A and B), CLSW is warmer and saltier and has a much lower CFC-11 concentration (Plates 2a and 2b, Figures 4a and 4b, and Table 2). Most of the difference in CFC-11 concentration, potential temperature, and salinity between the eastern and western basins is due to the increased convective

activity in the late 1980s. While the newly convected water is observed in the western basin in 1992, it had not yet reached the eastern basin in the 1988 observations [Sy et al., 1997].

Along section E in the Newfoundland Basin, high CFC concentrations extend down to -2000 m and both CLSW and

ULSW can be identified as CFC maxima in the upper 1800 m (Plate 2e). CLSW is warmer and saltier than is observed in the Labrador Sea, and the CFC-11 concentration is lower (Figure 4e and Table 2). These properties are all consistent with transport of CLSW in the DWBC from its source region and mixing laterally with older, warmer, and saltier CLSW. ULSW is observed at a single station, 15, in an eddy such as the one discussed in the section on NADW formation, but farther downstream from the formation site. Its CFC- 11 concentration and potential temperature are slightly higher, and its salinity and density are lower than the underlying CLSW (Table 2).

The southern tip of the Grand Banks represents the boundary between the subpolar and subtropical Atlantic. Along section F at 55øW just south of the Grand Banks the CFC distribution is similar to that in the Newfoundland Basin (Plate 2f). High CFC concentrations extend down to -1800 m with two maxima, one corresponding to ULSW and one to CLSW. Potential temperature and salinity for CLSW have increased moderately (Figure 4f and Table 2) as a result of lateral mixing with warmer, saltier water in the interior. However, there is a sharp change in O/S properties for ULSW compared to the eddy observed along section E in the Newfoundland Basin. The water south of the Grand Banks is much

warmer and saltier but retains its density, indicating extensive isopycnal mixing with warmer and saltier water flowing northward in the North Atlantic Current [Pickart et al., 1996].

It is important to note that during the 1980s there was only one CFC maximum at 55øW, and it was in ULSW [Smethie, 1993]. There was no CFC signal in CLSW, although Talley and McCartney [1982] had mapped its low-salinity and potential vorticity characteristics into the subtropics along the western margin on the basis of hydrographic data collected between 1954 and 1964. The flow of CLSW with relatively high CFCs around the Grand Banks appears to have begun about 1990 [Pickart and Smethie, 1998] after an increase in its production during the late 1980s [Lazier, 1995].

The earlier data, which had a better spatial resolution [Smethie, 1993, Figure 17] also suggested a split in the core of the ULSW at the Grand Banks with one branch flowing to the southwest along the western boundary and the other flowing more directly southward, similar to the circulation pattern shown by Schmitz and McCartney [ 1993]. The differences in pathways could be related to temporal variability or to the spatial resolution of the data collected in 1990 not being sufficient to show the split pathway (Figure 5a).

Farther south along section G at 38øN, 72øW in 1990, there is only a single CFC maximum in the upper 2000 m (Plate 2g), which is ULSW. This water is warmer and saltier (Figure 4g and Table 2) than at 55øW, but its density range is similar, indicating further isopycnal mixing. CLSW is present in the O/S plot as a minimum in salinity, but there is not a CFC maximum associated with it. However, the CLSW salinity minimum is not present farther south along section H (Figure 4h) suggesting that the leading edge of newly formed CLSW extended no farther than -38øN in 1990. Recently collected data show the signal had arrived at 26.5øN in 1996 [Molinari et al., 1999].

6.1.2. 38øN to the tropics. At -36øN the northeastward flowing Gulf Stream crosses over the southwestward flowing DWBC, and them is a distinct boundary in the ULSW CFC distribution. Contrasting section G (Plate 2g) and O/S and 0/CFC-11 plots (Figure 4g) north of the crossover with section H (Plate 2h) and O/S and 0 /CFC-11 plots (Figure 4h) south of the crossover reveals a 40% drop in CFC-11 concentration (these sections were taken within 2 weeks of each other), an increase in density from o•.5 = 34.57 to o•.5 = 34.63, and an increase in the depth of the maximum from -800 to 1300 m. This abrupt change in properties is due to a combination of the interaction between the Gulf Stream and

the DWBC and the increased contributions of Southern Hemisphere

14,316 SMETHIE ET AL.: TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER

and Mediterranean Water south of the crossover. Pickart and

Smethie [1993] showed that when the Gulf Stream and DWBC collide, the ULSW is split, with water less dense than o•.5 = 34.57 being entrained by the deep Gulf Stream and recirculated into the interior and water with greater density mostly flowing beneath the Gulf Stream. These results confirm the modeling studies of Spall [1996a,b] that suggest the properties of both the Gulf Stream and DWBC should be altered at the crossover.

ULSW is transported into the interior in the subtropics in two ways, cold core Gulf Stream rings [Smethie, 1993] and the deep Gulf Stream recirculation gyre [Schmitz and McCartney, 1993]. The latter appears to be the major pathway. Water recirculating in the gyre is found south of the Bermuda Rise (Figure 5a) with a high CFC-11 concentration [M. Baringer, personal communication, 1999]. Johns et al., [1997] show that several sverdrups of ULSW from the Gulf Stream recirculation are carried westward across the

axis of the Blake Bahama Outer Ridge. From there it is channeled back into the equatorward flowing DWBC at ~29øN. The same flow pattern is seen in Reid's [1994] adjusted steric height map on the 34.64 o•.5 surface [Reid, 1994, Figure 13b], which is slightly deeper than the CFC maximum at this latitude. Along section I at 24øN (Plate 2i and Figure 4i) the O/S properties are essentially the same as just south of the crossover and this marks the highest salinity (-35.03) that the ULSW obtains along the western boundary (see discussion below).

The decrease in the CFC concentration of ULSW equatorward and eastward is the result of two processes. First, water farthest from the source region has a lower concentration because it formed at an earlier time when the atmospheric concentration was less. Second, the water farther downstream has had more time to mix with low

tracer or tracer-free water along the flow path. The effect of mixing and dilution can be examined by calculating dilution factors as described in section 6.1.2. The map of dilution factors for ULSW (Figure 7a) has a distribution similar to the CFC-I 1 concentration map (Figure 5a) but the relative gradients are less since the effect of the temporally changing atmospheric source has been removed. North of ~30øN, the dilution factors are 2 or less, indicating that the water has mixed with an equal volume of water formed prior to the input of CFCs. The high dilutions south of the Gulf Stream Crossover are consistent with the conclusions of Pickart and

Smethie [1993] regarding its effect on the properties of the upper part of the DWBC and the recirculation of the upper part of ULSW into the interior with the Gulf Stream.

6.1.3. Tropics. In the tropical Atlantic the ULSW CFC signal remains a prominent feature (Plates 2j, 2k, and 21). There is a strong decrease in CFC-11 concentration from-0.6 pmol kg '• at 24øN to -0.15 pmol kg '• at 6øN for the core of ULSW, but this difference is exaggerated by temporal differences because the 24øN measurements were made in 1992 and the 6øN measurements were

made in 1989. In the map (Figure 5a) where concentrations have been normalized to 1990 the concentration change is more gradual, decreasing by about a factor of 2. The most prominent feature in the map and the section along 35øW (Plate 2k) is the splitting of the ULSW into two branches with one extending eastward along the equator and the other southward along the western boundary, as first observed by Weiss et al., [ 1985] in CFC data collected in 1983. In agreement with C. Andtie et al., [1998], the CFC maximum is located just south of the equator in the western basin coincident with the salinity maximum [Tsuchiya et al., 1992]. However, the CFC maximum appears to shift northward to the equator in the eastern basin. A more subtle feature is the northward bulging of the CFC-11 isopleths between 40 ø and 50øW. In agreement with Andtie et al., [ 1998] this suggests that some ULSW is recirculating from the western boundary to the interior before reaching the equator, supporting the existence of a deep recirculation gyre [Johns et al., 1993; McCartney, 1993; Schmitz and McCartney, 1993; Friedrichs and Hall, 1994] in the Guiana Basin. Reid [ 1994, Figure 13d] shows a very similar pattern for oxygen (splitting at the

equator and northward bulge between 40 ø and 50øS) plotted on the 34.64 Ol. 5 density surface, which is close to the density of the ULSW CFC maximum in the equatorial region.

There is a dramatic change in the O/S structure of ULSW between the subtropics and the tropics. In the subtropics the ULSW is observed at either a slight salinity minimum or in a region where salinity increases as temperature increases (Figures 4fi 4g, 4h, and 4i), but in the tropics it is observed at a salinity maximum (Figure 4j, 4k, 41). Also, the salinity of the ULSW, which increases monotonically in the southward direction in the subtropics, decreases in the southward direction in the tropics (Figure 4j, 4k, 41; Table 2). The maximum salinity of ULSW in the data presented here occurs at 24øN, which is the latitude of the high-salinity tongue extending westward from the Mediterranean Sea at this depth [Reid, 1994, Figure 13c]. Lateral mixing with this higher-salinity water to the east can account for the increase in salinity with decreasing latitude to this point. The abrupt change between the subtropics and tropics is the result of intermediate waters of North Atlantic origin being displaced by northward spreading intermediate waters from the South Atlantic. Fine and Molinari, [1988] suggested the water from the South Atlantic was Antarctic Intermediate Water, but it

appears to be Upper Circumpolar Water which lies between Antarctic Intermediate Water and ULSW. The decrease in salinity in the southward direction is caused by the upper part of ULSW being entrained into the northward flowing Upper Circumpolar 'Water, thus eroding the top portion of the O/S structure (Figures 4j, 4k, 41), not by addition of fresher water into ULSW by diapycnal mixing with the Upper Circumpolar Water. This also causes the density of the ULSW core to increase, which is the opposite of what would result from diapycnal mixing with the less dense Upper Circumpolar Water. The density of the ULSW core in the tropics has increased to Ol.5 = 34.66, which is coincident with the density of CLSW.

In the tropics the core of ULSW is associated with salinity and oxygen maximums in the water colurn. This water has been widely referred to as upper NADW, and its origin has been thought to be CLSW [Wiist, 1935]. This is not surprising since it has the same density as CLSW, although it is warmer and saltier. Using CFCs to trace its O/S evolution, we have shown that at the time of our observations the upper NADW CFC signal is not derived from CLSW but instead from the less dense ULSW.

The dilution factor increases from 2 to 10 along the western boundary to the equator with high values in the equatorial plume (Figure 7a). This is consistent with the reversing currents along the equator observed with SOFAR floats [Richardson and Schmitz, 1993], which would result in enhanced mixing with older water in the equatorial band.

6.2 Overflow Water

6.2.1. Subpolar. As discussed previously, there are two overflow water masses that feed NADW: ISOW, which enters the northeastern basin, and DSOW, which enters the northwestern basin.

6.2.1.1. ISOW: Only ISOW is observed in the northeastern basin. Sections A and B extend across the ISOW that flows

southward along the Reykjanes Ridge in the eastern Atlantic (Figure 3 and Plate 2). Three varieties of ISOW are observed along these sections. (1) There is a lobe of high CFC water at the base of the Iceland continental lope (section A, station 4) that has a potential temperature and salinity ~4.3øC and 35.00, which is incompletely formed ISOW [Doney and Bullister, 1992] (discussed earlier). Although its potential temperature, salinity, and density character- istics are similar to those of ULSW, it is not thought to be a source of this water mass. (2) At section A, station 6, ~75 km farther south, them is a slight CFC-11 maximum just above the bottom that has O/S values of 2.66øC/34.976 and a density of o2 = 37.058 (04 =

SMETHIE ET AL.: TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER 14,317

45.825). Although there is only one data point in this feature, there is a similar feature at stations 580-584 along section B (Plate 2b and Figure 4b), which has nine data points. This water is slightly warmer and saltier (Table 2) and has a density of c•2 = 37.03-37.05 (c•4 = 45.77-45.81). Swirl [1984] reports that the maximum density that extends from the eastern basin through the Charlie Gibbs Fracture Zone is c•2 = 37.04, so this water appears to be the water that flows through the fracture zone. Although the O/S changes between sections A and B are small, the difference in CFC-11 concentration is fairly large, 2.4 pmol kg '• for section A compared to 1.8 pmol kg -• for section B. This difference would be even greater if the concentrations were normalized to the same date (section A was taken in 1988, and section B was taken in 1991). This decrease in CFC concentration indicates mixing with adjacent older water with similar O/S properties as the water flows southward along the Reykjanes Ridge. (3) At stations farther soath and east on both sections there is a CFC maximum along the bottom. This bottom maximum has O/S values of 2.2ø-2.3øC/34.975 and a density of c•2 >37.08 (c•4 greater than 45.85), which is too dense to flow through the Charlie Gibbs Fracture Zone. Thus it appears that the densest variety of ISOW formed does not flow directly into the western basin but fills the bottom of the northeastern basin.

Water that flows through the Charlie Gibbs Fracture Zone is fresher than the mixing line between northeast Atlantic water and pure ISOW (Figure 4a), although both of these water masses must be major components. Thus it appears that pure ISOW entrains CLSW in addition to northeast Atlantic water as shown by Harvey and Theodorou [1986]. From Figure 4a and 4b, a composition of 45% pure ISOW, 20% northeast Atlantic water, and 35% CLSW can be estimated for water expected to flow through the Charlie Gibbs Fracture Zone, which is in good agreement with Harvey and Theodorou's [1986] estimate, except that they conclude that North Icelandic and Arctic Intermediate water contribute 9% and CLSW

contributes only 25%. In the lrminger Basin, three of the major components of

NADW are clearly evident (Plate 2c). The ISOW horizon is relatively well mixed horizontally, is saltier, and has a lower CFC concentration than the CLSW and DSOW that it is sandwiched

between. The O/S characteristics for ISOW between the eastern and

western basins is different (Figures 4b and 4c) with the western basin variety being fresher and slightly warmer; this can be explained by mixing with DSOW and CLSW after entering the western basin through the Charlie Gibbs Fracture Zone. There is not a vertical CFC-11 maximum in the ISOW horizon along the western flank of the Reykjanes Ridge, which would be expected if the water flowing through Charlie Gibbs Fracture Zone flowed northward along the ridge undiluted. This suggests that older water from the ISOW horizon is rapidly entrained into the newer water entering the basin from the fracture zone. This modified ISOW will be referred to as GFZW [Broecker and Peng, 1982].

6.2.1.2. DSOW: The DSOW underlying the GFZW in the Irminger Basin has a density greater than c•4 = 46.00 and is tagged with a high level of CFCs (Plate 2c) reflecting its recent input from Denmark Strait. It is observed as a CFC-11 maximum and

temperature and salinity minimum at the bottom extending eastward from the Greenland continental rise to beyond the center of the basin. Although the DSOW is the coldest water observed, it is not as cold or as fresh as upper Arctic Intermediate Water (UAIW) (Figure 4c), which has been proposed to be its precursor [Swirl and Aagaard, 1981; Swirl, 1984]. Its O/S properties along this section could be explained by roughly a 45:55 mixture of UAIW and GFZW. If Arctic Atlantic Water (AAW) is the precursor as suggested by Mauritzen [ 1996], it would also have to entrain GFZW to obtain the observed O/S characteristics but in a smaller amount, 60% AAW:40% GFZW. For either formation mechanism the

majority of the CFC signal comes from the DSOW precursor since its CFC concentration is much higher than the GFZW concentration.

Assuming the mixtures estimated in the above paragraph form rapidly with respect to the changing input of CFC-11 (-1 year), the concentration of CFC-11 in pure DSOW can be estimated from the CFC-11 concentrations measured in DSOW and GFZW on section

C (Plate 2c and Table 2). For the 45:55 mixture of UAIW and GFZW the estimated concentration in pure DSOW is 5.4 pmol kg '•, which corresponds to a percent saturation of 74%. For the 60:40 mixture of AAW and GFZW the pure DSOW CFC-11 concentration is estimated to be 4.3 pmol kg '•, which corresponds to 60% saturation. The fraction of the CFC-11 signal that is derived from the pure DSOW component can also be estimated and is 78% for the UAIW:GFZW mixture and 85% for the AAW:GFZW mixture, which agrees with earlier findings that DSOW is the main source of CFCs to OW that flows into the subtropical Atlantic [Srnethie, 1993].

Along section D in the Labrador Sea the GFZW CFC-11 concentration is similar to that in the Irminger Sea, but the core is cooler and fresher (Figure 4d). This is probably caused by the upper portion of the GFZW layer mixing with CLSW during the wintertime deep convection. The maximum CFC concentration occurs in the DSOW layer at the base of the continental slope (Plate 2d). It has decreased by -30% from values in the lrminger Sea

ß

(Figure 4d), and the DSOW core is warmer and less dense. It is possible that the densest water observed in the Irminger Sea has not been transported to the Labrador Sea, which would occur if the densest DSOW follows deeper isobaths east of the Labrador Basin. Diapycnal mixing with overlying GFZW would also decrease the density, and both mechanisms may be at work.

Along section E in the Newfoundland Basin the CFC concentrations in both the GFZW and DSOW layers (Plate 2e and Figure 4e) have decreased relative to the Labrador Sea concentrations (Plate 2d) as expected in the downstream direction. The O/S properties for DSOW are essentially unchanged (Figure 4e). However, potential temperature and salinity are warmer and saltier for GFZW and are close to the values for the Irminger Sea. This suggests that some GFZW bypasses the Labrador Basin enroute to the Newfoundland Basin.

There is an abrupt decrease in the density and CFC-11 concentration between the subpolar and subtropical basins, which is a result of the higher-density, higher CFC bottom water in the subpolar basins not flowing southward around the Grand Banks in the DWBC (see section 6.2.2). This indicates that the denser water recirculates in the subpolar basins with water siphoned off the top into the DWBC and replenished at the bottom by inflow from Denmark Strait. Some bottom water also flows around the Grand

Banks at deeper depths than the DWBC as discussed previously. A deep recirculating gyre has been proposed by a number of investigators. Worthington [1976] proposed a deep anticyclonic gyre in the Newfoundland Basin, but McCartney [1992] and Schmitz and McCartney [ 1993] suggest that this recirculation gyre does not exist in the very bottom water. Reid's [1994] plots of adjusted steric height for this region indicate that a cyclonic gyre circulates between the Labrador and Newfoundland Basins for deep and bottom water. Although the direction of the recirculation gyre cannot be determined from the CFC data, the data do strongly suggest that a deep recirculation gyre does exist in this region.

Between 35 ø and 42øN and 52 ø and 68øW, OW recirculating in the Northern Recirculation Gyre causes the eastern bulge of the 0.4 pmol kg -• isopleth (Figure 5). This recirculation feature shows up much better in the more extensive OCE-134 data set collected in

this region in 1983 [Hogg et al., 1986; Srnethie, 1993]. The 0.2 pmol kg -• isopleth extends much farther south than the 0.4 pmol kg '• isopleth in the interior of the western basin. There is not sufficient data in the northeastern part of the subtropical basin to determine the pathway for this CFC signal, but it could have been transported to this location by a southward interior flow that splits from the western boundary flow at the point where OW rounds the Grand Banks as proposed by Schmitz and McCartney [1993]. Another

14,318 SMETHIE ET AL.: TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER

possibility is the deep cyclonic subtropical gyre shown by Reid's [ 1994] adjusted steric height maps, which extends northward to the subpolar gyre in the Newfoundland and Labrador Basins. If there is an exchange of water between these two gyres, then the subtropical gyre would transpoa relatively high CFC water southward in its eastern limb.

The map of dilution factors for OW (Figure 7b) reveals that the water in the DWBC just south of the Grand Banks has been diluted by a factor of 2-3. The water circulated to the interior by the Noahern Recirculation Gyre has been diluted by a factor of 3-4. This dilution is greater than for ULSW at the same location (Figure 7a), demonstrating the strong influence the Gulf Stream has on recirculating ULSW into the interior. Along the western boundary the dilution factor increases monotonically from south of the Grand Banks to values greater than 10 in the tropical Atlantic.

6.2.2. Subtropics. In the subtropical Noah Atlantic the deep CFC maximum occurs adjacent to the continental slope between -3000 and 3800 m, and the concentration continues to decrease in the equatorward direction (Plates 2f-2g). The CFC-11 isolines of the OW maximum (Figure 5b) are nearly parallel to the western margin and extend beneath the Gulf Stream DWBC crossover, in contrast to the ULSW maximum whose isopleths are normal to the western boundary in the region of the crossover. This is caused by ULSW being recirculated into the interior at the crossover region as discussed previously, while OW flows more easily beneath the Gulf Stream [Pickart and Srnethie, 1993].

Although the effects of the DWBC/Gulf Stream crossover and recirculation gyre appear to be less for OW than ULSW, the opposite is the case for the Abaco gyre. A cyclonic gyre centered at 26.5øN, offshore of Abaco Island, the Bahamas [e.g., Lee et al., 1990], is outlined by the 0.2 pmol kg -• contour (Figure 5b). Johns et al.'s [1997] detailed study of the deep water flow paths in the region shows that they follow the isobaths and recirculate in the Abaco gyre more faithfully than does ULSW.

Although OW with a density greater than o4 = 45.9 is not found in the subtropical Atlantic DWBC, it is found at the bottom farther offshore. $rnethie [1993] observed lobes of high CFC/low silica water alternating with lobes of low CFC/high silica water at the bottom along 55øW between 37 ø and 42øN. This was attributed to OW that had a greater component of DSOW and hence was denser than OW found in the DWBC, interleaving with low CFC/high silica water of southern origin flowing into the region from the mid-ocean ridge to the east. The 0/$ properties of the offshore water were 1.75ø-1.88øC/34.884-34.896, and its density ranged from o4 = 45.908 to c• 4 = 45.915.

Within the DWBC the deep CFC maximum is observed at different densities. A comparison of the 0/$ plot for section F at 55øW (Figure 4f) with the O/S plot for section E in the Newfoundland Basin (Figure 4e) reveals that the CFC maximum has shifted to a shallower density. In the Newfoundland Basin as well as the other subpolar sections the DSOW maximum occurs at a density greater than c• 4 = 45.90 (Plates 2c, 2d, and 2e and Figure 4c, 4d, and 4e). Water denser than o4 = 45.90 does not enter the subtropical North Atlantic from the Newfoundland Basin in the DWBC. Along section F the CFC maximum occurs at a density of

-c• 4 • 45.88. However, farther south in the subtropical western Atlantic the CFC maximum is found at a lower density of-45.86 o4 (Plates 2g, 2h, and 2i). In the tropics the CFC maximum is again found at 45.88 c• 4 (Plates 2j and 2k). During the mid-1980s the CFC maximum in the DWBC occurred at 45.88 c• 4 throughout the subtropics [Fine and Molinari, 1988; $rnethie, 1993]. This vari- ability in density with time in the subtropics is probably due to density variations in different vintages of OW being transpoaed in the DWBC [Pickart, 1992b]. The variability between the subtropics and tropics may also be due to temporal variability in the AABW [Hall et al., 1997].

In the area between the Bermuda Rise and the mid-Atlantic

Ridge, there is evidence for modification of OW by mixing with AABW (M.S. McCartney, personal communication, 1999). Speer and McCartney [ 1991] suggest that the salinity minimum between 15 ø and 25øN over the mid-Atlantic Ridge could be direct evidence of veaical mixing between OW and AABW. A poaion of this modified water joins the DWBC at the Blake Bahama Outer Ridge (M. Baringer, personal communication, 1999) to flow equatorward [Amos et al., 1971]. The effect of AABW on modifying the OW (estimated to contain 20% AABW by Wiist [ 1935] and Broecker et al., [1991]) is the reason for the large zonal gradients in CFC concentrations (Figure 5b) in the subtropics. The effect can be observed in the O/S diagram throughout the subtropical and tropical western Atlantic Ocean. There is an inflection at -2øC, often referred to as the 2 ø discontinuity [Broecker et al., 1976] with water colder than the inflection point falling on a straight line between the inflection point and AABW. The deep CFC maximum in the DWBC occurs either at this inflection point or at a slightly warmer temperature throughout the subtropical and tropical western Atlantic (Figures 4f-41).

6.2.3. Tropics. Overflow water is transpoaed in the DWBC from the subtropical Atlantic to the tropical Atlantic. Reid [1994] and Schmitz and McCartney [1993] both present evidence for an elongated recirculation gyre in the Guiana Basin with flow from the western boundary into the interior near the equator and flow from the interior to the western boundary near 25øN. This recirculation has been related to an increase in transpoa of the DWBC in the Guiana Basin [Johns et al., 1993]. The CFC distribution reflects the Guiana Basin gyre. The CFC concentration gradient along the western boundary is low (Figure 5b) because of the rapid transpoa in the DWBC, and the isopleths turn eastward near the equator. The return flow of the gyre extends to the noaheast along the flank of the Mid-Atlantic Ridge and transpoas a mixture of AABW (which contains essentially blank level CFC concentrations) and the recirculated OW. The low CFC influence of AABW appears to be most pronounced between 20 ø and 30øN with a westward pointing minimum centered at 25øN reflecting the region where the gyre circulation tums westward toward the western boundary. Similar patterns are observed in tritium [Doney and Jenkins, 1994] and salinity, nutrients, and oxygen [Fukumori et al., 1991; Reid, 1994; Speer, 1993].

In the equatorial region the CFC distribution (Figure 5b) reveals spreading of OW eastward along the equator and southward along the western boundary. This is similar to the distribution for the ULSW, but the isopleths do not extend as far to the east because at this depth eastward flow is affected by the Romanche Fracture Zone [Andrie et al., 1998]. Along section K that crosses the equator at 35øW the deep CFC signal is confined to the channel, which runs east-west along the equator and is the conduit for bottom water flow between the western North and South Atlantic Oceans (Plate 2k). The CFC-11 concentration in the OW CFC maximum decreases by a factor of 3 between section K at the equator and section L at 10øS (Table 2), and these two sections were both taken in 1992. This sharp decrease may be the result of the deep flow splitting at the equator, as is the case for the ULSW, with a significant poaion of the flow entering the Romanch Fracture Zone [Andrie et al., 1998].

As discussed in the previous paragraph, the O/S propeaies of the OW CFC maximum are the same as in the subtropical North Atlantic. However, this water is now underlain by water enriched with Antarctic Bottom Water (AABW) as can be seen from the O/S trend below the 2 ø discontinuity (Figure 4k and 41). The CFC-11 concentration also increases with decreasing temperature beneath the 2 ø discontinuity, and this is particularly evident at 10øS (Figure 4k-41). This CFC signal is from AABW, which originated in the Southern Ocean, and in 1992 it approached entry to the Noah Atlantic. See Smythe-Wright and Boswell [1998] for additional information on CFCs in AABW.

SMETHIE ET AL.: TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER 14,319

7. CFC Ages and Spreading Rates

The previous discussion has demonstrated how newly formed components of NADW are transported and mixed into the North Atlantic Ocean from their source regions and how this affects the hydrographic structure. Since this discussion was built around the spreading of the CFC signal in the North Atlantic, one can readily see that the timescale involved is several decades, i.e., the time of the measurable CFC transient. However, as discussed previously, the CFCs can be used to estimate water mass ages on a finer scale than several decades.

Maps of the CFC-11 :CFC-12 ratio age (hereafter referred to as the CFC age) for ULSW and OW are presented in Figure 6. These ages are calculated assuming that the CFC-1 l:CFC-12 ratio (not concentration) in the source water was in solubility equilibrium with the atmosphere during the year of formation. However, if the surface water does not reach equilibrium during the formation process, as occurs during deep convection, the ratio in newly formed water may not be the same as is predicted from the atmospheric ratio and the solubilities, and the newly formed water will have an apparent age. For ULSW an apparent age at formation has been estimated to be .-.3-5 years from the CFC-113:CFC-11 ratio (in a manner similar to the one used to calculate ages from the CFC-11:CFC-12 ratio [Schlosser and Smethie, 1995]) and from the tritium:He-3 ratio [Pickart et al., 1996]. Newly formed OW may also have an apparent age. On the En223 cruise (Table 1), CFC-113 was measured as well as CFC-11 and CFC-12 at four stations

crossing the flow of DSOW around the southern tip of Greenland. The age of this water based on the CFC-113:CFC-I 1 and CFC- 113:CFC-12 ratios was 3.6 years, which would represent an upper limit for the apparent age at its entry into the North Atlantic. Thus the ages presented in the maps include a "relic age" acquired at the time of formation that could be as high as 3-4 years. Also, mixing can result in the CFC age of the DWBC being overestimated [e.g., Pickart et al., 1989; Rhein, 1994]. It should be kept in mind that these ages are not the average age for the entire water mass but are for the most recently formed component that has been tagged with CFCs during the past few decades.

The CFC age for ULSW is ~ 10 years in the DWBC just south of the Grand Banks (Figure 6a). It increases more or less monotonically to between 20 and 25 years at 10øN and to greater than 25 years in the equatorial plume and along the western boundary south of the equator. Ages in the interior north of 30øN range from <15 years to 20 years, reflecting the circulation of ULSW into the interior at the Gulf Stream/DWBC crossover. The

15 year isopleth extends to 30øN near the Mid-Atlantic Ridge, suggesting flow directly to the interior from the subpolar region. However, this flow pattern is not supported by the 1990 CFC-11 distribution (Figure 5a) perhaps because there are no data east of 63øW for this time. The age data east of 63 ø W are from earlier cruises. The age difference between the Grand Banks and the tropics suggests a 10 year transit time for ULSW to flow through the subtropical Atlantic along the western margin. Recently, Molinari et al., [1999] have used a time series of hydrographic and CFC concentrations at 26.5øN in the DWBC to show that the transport time of the recently formed CLSW (discussed previously) is 8-10 years from the Labrador Sea to 26.5øN, which is about the same as ULSW transit time derived from the maps.

The CFC age of OW (Figure 6b) is 12-15 years in the DWBC south of the Grand Banks and increases along the western margin to -25-30 years at 8øN. The OW ages in the interior north of 30øN are higher than for ULSW and range from 20 to 25 years. A comparison of Doney and Jenkins' [1994] tritium/He-3 ages with Figure 6 reveals similar ages along the western boundary, but in the interior the tritium/He-3 ages are greater than the CFC ages by 5 to 10 years. This difference may be caused by mantle-derived He-3, which is transported into the North Atlantic by the northward

flowing AABW. Although a correction is made for this, this correction becomes large and difficult to make accurately in regions where the mantle component is high and the tritiogenic component is low, which is the case in the interior and far downstream from the source region.

Apparent current speeds can be calculated by dividing the distance from the source by the tracer age. For ULSW and OW these speeds range from 1 to 2 cm s 't [Weiss et al., 1985; Srnethie, 1993; Doney and Jenkins, 1994]. This can be easily seen in a plot of tracer age in the DWBC core verses distance from the source region (Figure 8). For ULSW the slope in the subtropics indicates an apparent current speed of .-.1.1 cm s 't. In the tropics the slope appears to decrease, which would indicate a higher apparent current speed, but the ages in the tropics have large errors because of uncertainties in the CFC ratio at low CFC concentrations. For OW

there is very good agreement between tritium/He-3 ages based on measurements made between 1981 and 1983 [Doney and Jenkins, 1994] and CFC ages based on measurements made between 1983 and 1992. The overall slope based only on CFC ratio ages yields an apparent current speed of 1.9 cm s 'l compared to 1.7 cm s '• on the basis of the tritium/He-3 age [Doney and Jenkins, 1994]. The slope of the combined CFC ratio ages and tritium/He-3 ages (Figure 8b) yields an apparent current speed of 1.8 cm s 'l, extending from the subpolar region to the tropics.

Measured current speeds in the DWBC are considerably higher than 1-2 cm s '1. Long-term current meter measurements in the DWBC along the North American continental slope show speeds generally from 5 to 10 cm s '1 [Watts, 1991], and speeds as high as 40 cm s -1 have been observed [e.g., Mills and Rhines, 1979; Johns, 1993]. The primary reason for the difference between the tracer- derived speed and absolute velocity measurements is that the tracers give an effective spreading rate [Doney and Jenkins, 1994; Fine, 1995], which includes the effects of mixing and recirculation. In the mean the water does not flow directly equatorward in the DWBC but makes excursions into the interior, mixes with older water, and then comes back to the boundary; the DWBC transports both newly formed water and older recirculating water. The effective spreading rates derived from tracer data clearly demonstrate the importance of exchange with the interior in aging the water. The tracer-derived age represents the rate at which a climate anomaly or perturbation in the formation of NADW would enter the ocean.

8. Summary and Conclusions

The compilation of available CFC data for the western North Atlantic provides an opportunity to contribute details to our knowledge of the large-scale circulation pathways that the components of NADW follow and the related timescales. The formation of the major co•nponents of NADW are reviewed with a bias toward understanding the sources of the most recently formed and hence high tracer components. A summary of this review and synthesis follows.

The shallowest component of NADW is ULSW, which forms by deep convection in the southern Labrador Sea, possibly in the Labrador Current [Pickart et al., 1997]. Its existence was discovered only recently [Pickart, 1992a] from observations of a CFC maximum [Fine and Molinari, 1988; Smethie, 1993] and a tritium maximum [Jenkins and Rhines, 1980; Olson et al., 1986; Pickart et al., 1996] obtained during the 1980s in the North Atlantic subtropical DWBC at a temperature too warm to be CLSW. It has a core density of •1.5 = 34.56. Newly formed ULSW, which has a CFC saturation of-70%, advects southward from the source region in small eddies that become rapidly entrained into the DWBC [Pickart et al., 1996]. This maximum has also been observed in the DWBC in the tropics [Rhein et al., 1995] and in a plume extending eastward along the equator [Weiss et al., 1985]. The upper portion

14,320 SMETHIE ET AL.: TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER

of the ULSW recirculates into the interior at the Gulf Stream/DWBC crossover [Pickart and Smethie, 1993], but some of this flow appears to rejoin the DWBC at the Blake-Bahama Outer Ridge [Johns et al., 1997; M. Baringer, personal communication, 1999]. From the combined CFC data set presented here we show the following.

1. ULSW is a major water mass in the western North Atlantic Ocean. It is a continuous feature in the DWBC from the Grand

Banks to just south of the equator and extends far into the interior of the western subtropical Atlantic.

2. The O/S properties of the ULSW CFC maximum in the DWBC show that the salinity gradually increases in the southward direction to ~24øN because of mixing with higher-salinity water from the east. South of this latitude, them is an abrupt change caused by the upper portion of the ULSW being entrained into northward spreading upper Circumpolar Water, and the salinity and temperature of the CFC maximum decrease while the density increases. In the equatorial region the density of the ULSW CFC maximum is essentially the same as CLSW in the subtropics, but the temperature and salinity are greater than that of CLSW. Although CLSW may have been important in the past for formation of upper NADW, during the 1980s, ULSW was the major source of upper NADW.

3. The equatorial plume is much better defined than in earlier work, extending eastward to at least 3øE with the maximum generally occumng at 1 ø-2øS.

4. The high CFC concentrations in the interior of the subtropics indicate extensive recirculation from the boundary to the interior, particularly with respect to the Gulf Stream Recirculation. The data also support recirculation at the Blake Bahama Outer Ridge and in the Guiana Basin.

5. Dilution factors of ULSW range from -2 along the western boundary of the subtropics to 10 at the equator, with values >10 in the equatorial plume. The large region in the subtropics where the dilution ranges from 2 to 4 reflects the extensive recirculation in this region.

CLSW forms by deep convection extending to depths of >2000 m, and its historical core density is o•. 5 = 34.66 [Talley and McCartney, 1982]. The very deep convection results in undersaturation of CFCs in newly formed water, and Wallace and Lazier [1988] reported 60% saturation in recently formed CLSW in the central Labrador Sea in 1986. However, in the subtropics during the 1980s the 34.66 Ol. 5 density surface was close to a deep CFC minimum, indicating that recently formed CLSW had not entered the subtropics in significant quantities [Smethie, 1993]. Yet, in the Newfoundland Basin, CLSW was observed near the western boundary, and as reported by Pickart and Smethie [1998], it was observed just south of the Grand Banks underlying ULSW in 1991. This demonstrated that CLSW, which formed during the intense winters associated with the rise in the NAO index in the late

1980s and early 1990s, was just beginning to flow into the subtropics in the early 1990s. From the combined CFC data set presented here we show the following for CLSW.

1. High CFC concentrations extended down to 2000 rn in both the Labrador Sea, which had a CFC saturation of 62%, and the Irminger Sea. Lower concentrations were observed in CLSW in the eastern basin.

2. The leading edge of the CLSW formed during the intense winters of the late 1980s appeared to extend to 38øN in the DWBC in 1990, but there was no CFC maximum in CLSW. The CFC concentration in CLSW at 38øN was well above the minimum observed in the 1980s but was less than that of ULSW.

The densest components of NADW are the overflow waters that enter the North Atlantic from behind the Greenland-Iceland- Scotland Ridge. ISOW is the least dense of the overflow waters, and it enters the western basin through the Charlie Gibbs Fracture Zone where it ovemdes and is entrained into the denser DSOW that

enters through Denmark Strait [Swirl, 1984]. DSOW has a much higher CFC signal than ISOW and is the primary source of CFCs in the bottom water of the subpolar western basin [Smethie and Swirl, 1989; Smethie, 1993]. Both ISOW and DSOW flow around the Irminger and Labrador Basins in a deep boundary current [Swirl, 1984; McCartney, 1992], and a mixture of these two water masses enters the subtropics in the DWBC [Smethie, 1993] producing a deep CFC maximum that is observed in the DWBC through the subtropics [Fine and Molinari, 1988; Smethie, 1993] and tropics to 10øS [Molinari et al., 1992; Rhein et al., 1995]. The OW recirculates into the interior of the subtropical Atlantic with the northern recirculation gyre [Hogg et al., 1986; $methie, 1993], but south of this recirculation, it crosses beneath the Gulf Stream with little impedance [Pickart and Smethie, 1993]. Farther south, at ~26øN, there is another recirculation gyre off Abaco Island [Johns et al., 1997]. From the combined CFC data set presented here we show the following for lower NADW.

1. The maximum density of recently formed ISOW that flows through the Charlie Gibbs Fracture Zone to the western basin to form GFZW is 44.77-45.81 o4 (37.03-37.05 o2), which is in good agreement with Swifi's [1984] determination of 37.04 o2 on the basis of data collected in the 1970s. It is a three-component mixture of 45% pure ISOW, 20% northeast Atlantic water, and 35% CLSW. The densest ISOW, which was observed to be 45.85 o4 in the combined data set, does not enter the western basin but spreads along the bottom of the eastern Noah Atlantic where it produces a bottom CFC maximum.

2. The percent CFC saturation of newly formed DSOW was estimated to be 74% taking AIW to be the source of pure DSOW [Swirl and Aagaard, 1981 ]. However, recently, Mauritzen [ 1996] proposed that AAW is the source for DSOW, and 60% saturation was estimated for this case. For either source, DSOW provides -80% for the CFC signal to the DSOW/GFZW mixture that is transported southward in the DWBC.

3. The dense, high CFC bottom water in the Newfoundland Basin (45.92 04] was not observed south of the Grand Banks and apparently recirculates in the Newfoundland Basin, with some of the water flowing southward along the bottom offshore of the DWBC as observed previously [Smethie, 1993].

4. The density of the deep CFC maximum in the subtropics and tropics ranged from 45.86 to 45•88 04 both spafially and temporally where there were reoccupations of sections across the DWBC. This variability could be caused by temporal variability in the source water as suggested by Pickart [1992b] or by vaciability in mixing with AABW encountered as the NADW flows southward.

5. In addition to the recirculation that occurs in the subtropics, the CFC distribution in the tropics supports recirculation in the Guiana Basin gyre.

6. The dilution factor for OW is estimated to be 2-3 along the western margin of the subtropical North Atlantic and 3-4 in the subtropical interior. It increases to values >10 in the tropics.

The CFC ages reveal that ULSW and OW both spread from their subpolar source regions to the tropics in 25-30 years with younger ages generally found along the western boundary rather than in the interior. Effective spreading rates in the DWBC were estimated from plotting age against distance to be -1.1 cm s 'l for ULSW and 1.8 cm s -• for OW, in good agreement with previous estimates of 1-2 cm s '• from tracer studies [Weiss et al., 1985; Smethie, 1993; Doney and Jenkins, 1994]. These effective spread- ing rates are low compared to direct long-term current meter measurements in the DWBC, which reveal speeds of 5-10 cm s 'l [Watts, 1991 ]. The tracer-derived effective spreading rates integrate the effect of mixing and recirculation processes on the equatorward flow. This difference demonstrates the importance of recirculation gyres in transporting newly formed NADW into the ocean interior, where it mixes with older water before reentering the DWBC to continue its southward journey. This process retards the spreading

SMETHIE ET AL.: TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER 14,321

rate of NADW in the relatively fast flowing DWBC and renews the interior water of the North Atlantic Ocean.

Acknowledgments, We would like to thank Debbie Willey, Hoyle Lee, and Frank Zheng for combining all of the data sets in a common format for us to work with and for preparation of the figures presented in this manuscript. This work was supported by the NOAA Atlantic Climate Change Program, NOAA grants NA26GP0231 and NA46GP0168 to W. Smethie and NA67RJ 1049

to R. Fine, and NSF grants OCE 89-17801 and OCE 90-19690 to WMS and OCE 94-13222 and OCE 98-11535 to RAF. This is

Lamont-Doherty Earth Observatory contribution number 6028.

References

Amos, A.F., A.L. Gordon, and E.D. Schneider, Water masses and circulation patterns in the region of the Blake-Bahama Outer Ridge, Deep Sea Res. Oceanogr. Abstr., 18, 145-165, 1971.

Anderson, J., D. Toohey, and W. Brune, Free radicals within the Antarctic vortex: The roll of CFCs in Antaractic ozone loss, Science, 251, 39-46, 1991.

Andrie, C., J.-F. Ternon, M.-J. Messias, L. Memery, and B. Bourles, Chloro- fluoromethane distributions in the deep equatorial Atlantic during January-March 1993, Deep Sea Res., Part I, 45, 903-930, 1998.

Bacon, S., Decadal variability of the North Atlantic overflows, WOCE lnt. WOCE Newsl., 26, 29-30, 1997.

Broecker, W.S., The great ocean conveyor, Oceanography, 4, 79-89, 1991. Broecker, W.S., Chaotic climate, Sci. Amer., 273, 62-68, 1995. Broecker, W.S. and TQ-S. Peng, Tracers in the Sea., Lamont-Doherty Geol.

Obs., Palisades, N.Y., 1982. Broecker, W.S., T. Takahashi, and Y.-H. Li, Hydrography of the central

Atlantic, I, The two-degree discontinuity, Deep Sea Res., Oceanogr. Abstr., 23, 1083-1104, 1976.

Broecker, W.S., T.H. Peng, G. Mathieu, R. Hesslein, and T. Torgersen, Gas exchange rate measurements in a natural system, Radiocarbon, 22, 676- 683, 1980.

Broecker, W. S., S. Blanton, and W.M. Smethie Jr., Radiocarbon decay and oxygen utilization in the deep Atlantic Ocean, Global Biogeochem. Cycles, 5, 87-117, 1991.

Bryden, H.L., M.J. Griffiths, A.M. Lavin, R.C. Millard, G. Parrilia, and W.M. Smethie Jr. Decadal changes in water mass characteristics at 24øN in the subtropical Atlantic Ocean, J. Clim., 9, 3162-3186, 1996.

Bullister, J.L., Atmospheric chlorofluoromethanes as tracers of ocean circulation and mixing: Measurement and calibration techniques and studies in the Greenland and Norwegian Seas, Ph.D. thesis, 172 pp., Univ. of Calif., San Diego, La Jolla, 1984.

Bullister, J.L. and B.-S. Lee, Chlorofluorocarbon-I 1 removal in anoxic marine waters, Geophys. Res. Lett., 22, 1893-1896, 1995.

Cunnold, D.M., P.J. Fraser, R.F. Weiss, R.G. Prinn, P.G. Simmonds, B.R. Miller, F.N. Alyea, and A.J. Crawford, Globaltrends and annual releases of CCI3F and CC12F2 estimated from ALE/GAGE and other measurements from July 1978 to June 1991, J. Geophys. Res., 99, 1107- 1126, 1994.

Cunnold, D.M., R.F. Weiss, R.G. Prinn, D. Hartley, P.G. Simmonds, P.J Fraser, B. Miller, G.N. Alyea, and L. Porter, GAGE/AGAGE measurements indicating reductions in global emissions of CCI3F and

CC12F2 in 1992-1994, J. Geophys. Res., 102, 1259-1269, 1997. Curry, R.G., M. McCartney, and T.M. Joyce, Oceanic transport of subpolar

climate signals to mid-depth subtropical waters, Nature, 391, 575-577, 1998.

Dickson, R.R., and J. Brown, The production of North Atlantic Deep Water: Sources, rates and pathways, J. Geophys. Res., 99, 12,319-12,341, 1994.

Dickson, R.R., J.R.N. Lazier, J. Meinke, P. Rhines, and J. Swift, Long-term coordinated changes in the convective activity of the North Atlantic. Prog. Oceanogr., 38, 241-295, 1996.

Dickson, R.R., J. Meincke, I. Vassie, J. Jungclaus, and S. Osterhus, Possible predictability in overflow from the Denmark Strait, Nature, 397, 243- 246, 1999.

Doney, S.C., and J.L. Bullister, A chlorofluorocarbon section in the eastern North Atlantic. Deep Sea Res., Part A, 39, 1857-1883, 1992.

Doney, S. C., and W. J. Jenkins, Ventilation of the Deep Western Boundary

Current and Abyssal Western North Atlantic: Estimates from tritium and 3He distributions, J. Phys. Oceanogr., 24, 638-659, 1994.

Elkins, J., T. Thompson, T. Swanson, J. Butler, B. Hall, S. Cummings, D. Fisher, and A. Raffo, Decrease in the growth rates of atmospheric chlorofluorocarbons - 11 and - 12, Nature, 364, 780-783, 1993.

Fine, R. A., Tracers, time scales, and the thermohaline circulation: The lower limb in the North Atlantic Ocean, U.S. Nat. Rep. lnt. Union of Geod. Geophys. 1991-1994, Rev. Geophys., 33, 1353-1365, 1995.

Fine, R.A., and R.L. Molinari, A continuous deep western boundary current between Abaco (26.5øN) and Barbados (13øN), Deep Sea Res., Part A, 35, 1441-1450, 1988.

Fine, R.A., M.J. Warner, and R.F. Weiss, Water mass modification at the Agulhas Retroflection: Chlorofloromethane studies, Deep-Sea Res., Part A, 35, 311-332, 1988.

Fisher, D.A., T. Duafala, P.M. Midgley, and C. Niemi, Production and emission of CFCs, halons, and related molecules, in Report on Concentrations, Lifetimes and Trends of CFCs, Halons and Related Species, chap. 2, NASA Ref. Publ., 1339, 2.1-2.35, 1994.

Friedrichs, M.A.M., and M.M. Hall, Deep circulation in the tropical North Atlantic, J. Mar. Res., 51,697-736, 1993.

Fukumori, I., F. Martel, and C. Wunsch, The hydrography of the North Atlantic in the early 1980s: An Atlas, Prog. Oceanogr., 27, 1-110, 1991.

Gordon, A.L., hater-ocean exchange of thermocline water, J. Geophys. Res., 91, 5037-5046, 1986.

Hall, M.M., M. McCartney, and J.A. Whitehead, Antarctic Bottom water flux in the equatorial western Atlantic, J. Phys. Oceanogr., 27, 1903- 1926, 1997.

Harvey, J., 0-S relationships and water masses in the eastern North Atlantic, Deep Sea Res., Part A, 29, 1021 - 1033, 1982.

Harvey, J.G., and A. Theodorou, The circulation of Norwegian Sea overflow water in the eastern North Atlantic, Oceanol. Acta, 9, 393-402, 1986.

Hogg, N.G., R.S. Pickart, R.M. Hendry, and W.M. Smethie Jr., The northern recirculation gyre of the Gulf Stream, Deep Sea Res., Part A, 33, 1139-1165, 1986.

Jenkins, W.J., and P. Rhines, Tritium in the deep North Atlantic Ocean, Nature, 286, 877-880, 1980.

Johns, W.E., D.M. Fratantoni, and R.J. Zantopp, Deep western boundary current variability off northeastern Brazil, Deep Sea Res., Part 1, 40, 293-310, 1993.

Johns, E., R.A. Fine, and R.L. Molinari, Deep flow along the western boundary south of the Blake Bahama Outer Ridge, J. Phys. Oceanogr., 2 7, 2187-2208, 1997.

Lazier, J.R.N. The salinity decrease in the Labrador Sea over the past thirty years. in Natural Climate Variability on Decade-to-Century Time Scales, edited by D.G. Martinson et al., pp. 295-304, Nat. Res. Counc., Nat. Acad., Washington, D.C., 1995.

Lee, T.N., W. Johns, F. Schott, and R. Zantopp, Western boundary current structure and variability east of Abaco, Bahamas at 26.5øN, J. Phys. Oceanogr., 20, 446-466, 1990.

Livingston, H.D., J.H. Swift, and H.G. Ostlund, Artificial radionuclide tracer supply to the Denmark Strait Overflow between 1972 and 1981, J. Geophys. Res., 90, 6971-6982, 1985.

Manabe, S., and R.J. Stouffer, Two stable equilibria of a coupled ocean- atmosphere model, J. Clim., 1, 841-866, 1988.

Mauritzen, C., Production of dense overflow waters feeding the North Atlantic across the Greenland-Scotland Ridge, 1, Evidence for a revised circulation scheme, Deep Sea Res., Part 1, 43, 769-806, 1996.

McCartney, M.S., Recirculating components to the deep boundary current of the northern North Atlantic, Prog. Oceanogr., 29, 283-383, 1992.

McCartney, M. S., Crossing of the equator by the Deep Western Boundary Current in the western Atlantic Ocean, J. Phys. Oceanogr., 23, 1953- 1974, 1993.

McKee, T.A., R.S. Pickart, and W.M. Smethie Jr., Hydrography data from Endeavor, 223: Formation and spreading of the shallow component of the North Atlantic Deep Western Boundary Current, Tech. Rep. WH01- 95-07, 127 pp., Woods Hole Oceanog. hast., Woods Hole, Mass., 1995.

Mills, C.A., and P.B. Rhines, The Deep Western Boundary Current at the Blake-Bahama Outer Ridge: current meter and temperature observations, 77 pp., WHOI Tech. Rep. WH01-79-85, Woods Hole Oceanogr. hast., Woods Hole, Mass., 1979.

Molinari, R.L., R.A. Fine, and E. Johns, The Deep Western Boundary Current in the tropical North Atlantic Ocean, Deep Sea Res., Part A., 39, 1967-1984, 1992.

Molinari, R.L., R.A. Fine, W.D. Wilson, J. Abell, M.M. McCartney, and

14,322 SMETHIE ET AL.: TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER

R.G. Curry, The arrival of recently formed Laborador Sea Water in the Deep Western Boundary Current at 26.5øN., Geophys. Res. Lett., 25, 2249-2252, 1999.

Montzka, S.A., J.H. Butler, R.C. Myers, T.M. Thompson, T.H., Swanson, A.D. Clarke, L.T. Lock, and J.W. Elkins, Decline in the tropospheric abundance of halogens from halocarbons: Implications for stratospheric ozone depletion, Science, 272, 5266, 1318-1322, 1996.

Olson, D.B., H.G. Ostlund, and J. Sarmiento, The Western Boundary Undercurrent off the Bahamas, J. Phys. Oceanogr., 16, 233-240, 1986.

Ostlund, H.G., NAGS Tritium: North Atlantic Gyre Studies and associated projects, Tritium Lab. Data Rep. 13, Rosenstiel School of Mar. and Atmos. Sci., Univ. of Miami, Miami, Fl., 1984.

Ostlund, H.G., and C. Grail, Transient Tracers in the Ocean North and Tropical Atlantic tritium and radiocarbon, Tritium Lab. Data Rep. 16, Rosenstiel School of Mar. and Atmos. Sci., Univ. of Miami, Miami, Fl., 1987.

Ostlund, H.G., and C.G.H. Rooth, The North Atlantic tritium and radiocarbon transients, J. Geophys. Res., 95, 20,147-20,166, 1990.

Physical and Chemical Oceanographic Data Facility, Transient tracers in the ocean, North Atlantic study, 1 April - 19 October, 1981, shipboard physical and chemical data report, 714 pp., Scripps Inst. of Oceanogr., La Jolla, Calif., 1986.

Pickart, R.S., Water mass components of the North Atlantic Deep Western Boundary Current, Deep Sea Res., Part A., 39, 1553-1572, 1992a.

Pickart, R.S., Space-time variability of the Deep Western Boundary Current oxygen core, J. Phys. Oceanogr., 22, 1047-1061, 1992b.

Pickart, R.S., and W.M. Smethie Jr., How does the Deep Western Boundary Current cross the Gulf Stream?, J. Phys. Oceanogr., 23, 2602-2616, 1993.

Pickart, R.S. and W.M. Smethie Jr., Temporal evolution of the Deep Western Boundary Current where it enters the sub-tropical domain, Deep Sea Res., Part 1,45, 1053-1083, 1998.

Pickart, R.S., N.G. Hogg, and W.M. Smethie Jr., Determining the strength of the Deep Western Boundary Current using the chlorofluoromethane ratio, J. Phys. Oceanogr., 19, 940-951, 1989.

Pickart, R.S., T.K. McKee, and W.M. Smethie Jr. Hydrographic data from Endeavor 214: A study of the Gulf Stream-Deep Western Boundary Current crossover, Tech. Rep., WH01-92-23, 108 pp., Woods Hole Oceanogr. Inst. Woods Hole, Mass., 1992.

Pickart, R.S., W.M. Smethie Jr., J.R.N. Lazier, E.P. Jones, and W.J. Jenkins, Eddies of newly formed upper Labrador Sea Water, J. Geophys. Res., I01, 20,711-20,726, 1996.

Pickart, R.S., M.A. Spall, and J.R.N. Lazier, Mid-depth ventilation in the western boundary current system of the sub-polar gyre, Deep Sea Res., Part 1, 44, 1025-1054, 1997.

Reid, J.L., On the total geostrophic circulation of the North Atlantic Ocean: Flow patterns, tracers, and transports, Prog. Oceanogr., 33, 1-92, 1994.

Reid, J.L. and R.J. Lynn, On the influence of the Norwegian-Greenland and Weddell Seas upon the bottom waters of the Indian and Pacific Oceans, Deep Sea Res., Oceanogr. Abstr., 18, 1063-1088, 1971.

Rhein, M., The Deep Western Boundary Current: Tracers and velocities, Deep Sea Res., Part 1, 41,263-328, 1994.

Rhein, M., L. Stramma, and U. Send, The Atlantic Deep Western Boundary Current: Water masses and transports near the equator, J. Geophys. Res., 100, 2441-2457, 1995.

Richardson, P.L., and W.J. Schmitz Jr., Deep cross-equatorial flow in the Atlantic measured with SOFAR floats, J. Geophys. Res., 98, 8371-8387, 1993.

Robinson, M.K., R.A. Bauer and R.H. Schroder, Atlas of North Atlantic- Indian Ocean monthly mean temperatures and mean salinities of the surface layer. Ref Publ. 18, U.S. Navy Oceanogr. Off., Washington, D.C., 1980.

Schauer, U., R.D. Muench, B. RudeIs, and L. Timokhov, Impact of eastern Arctic shelf waters on the Nansen Basin intermediate layers, J. Geophys. Res., 102, 3371-3382, 1997.

Schlosser, P., and W.M. Smethie Jr., Transient tracers as a tool to study variability of ocean circulation, in Natural Clintate Variability on Decade-to-Century Time Scales, edited by D.G. Martinson, et al., pp. 274-288, Nat. Res. Counc., Nat Acad., Washington, D.C., 1995.

Schmitz, W.S., and M.S. McCartney, On the North Atlantic circulation, Rev. Geophys., 31, 29-49, 1993.

Shapiro, S.D., P. Schlosser, W. M. Smethie Jr., and M. Stute, The use of 3H and tritiogenic 3He to determine CFC degradation and vertical mixing rates in Framvaren Fjord, Norway, Mar. Che•n., 59, 141-157, 1997.

Smethie, W.M., Jr., Tracing the thermohal.ine circulation in the western North Atlantic using chlorofluorocarbons, Prog. Oceanogr., 31, 51-99, 1993.

Smethie, W.M. Jr., and J.H. Swift, The tritium: Krypton-85 age of Denmark Strait Overflow Water and Gibbs Fracture Zone Water just south of Denmark Strait, J. Geophys. Res., 94, 8265-8275, 1989.

Smethie, W.M., Jr., H.G. Ostlund, and H.H. Loosli, Ventilation of the deep Greenland and Norwegian Seas: Evidence from krypton-85, tritium, carbon-14, and argon-39, Deep Sea Res., Part A, 33, 675-703, 1986.

Smethie, W.M., Jr., M.T. Benjamin, and S.C. Sutherland, Chlorofluoro- carbon measurements made by LDGO on South Atlantic Ventilation Experiment Leg 5 and Hydros Leg 4, LDGO-92-2, 55 pp., Lamont- Doherty Geol. Obsv., Palisades, N.Y., 1992.

Smythe-Wright, D., and S. Boswell, Abyssal circulation in the Argentine Basin, J. Geophys. Res., 103, 15,845-15,851, 1998.

Spall, M.A., Dynamics of the Gulf Stream/Deep Western Boundary Current crossover, I, Entrainment and recirculation, J. Phys. Oceanogr., 26, 2152-2168, 1996a.

Spall, M.A., Dynamics of the Gulf Stream/Deep Western Boundary Current crossover, II, Low frequency oscillations, J. Phys. Oceanogr., 26, 2169- 2182, 1996b.

Speer, K.G., The deep silica tongue in the North Atlantic, Deep Sea Res., Part 1, 40, 925-936, 1993.

Speer, K.G., and M.S. McCartney, Tracing lower North Atlantic deep water across the equator, J. Geophys., Res., 96, 20,443-20,448, 1991.

Strass, V.H., E. Fahrbach, U. Schauer, and L. Sellmann, Formation of Denmark Strait Overflow Water by mixing in the East Greenland Current, J. Geophys. Res., 98, 6907-6919, 1993.

Stuiver, M., and H. G. Ostlund, GEOSECS Atlantic Radiocarbon, Radiocarbon, 22, 1-24, 1980.

Swift, J.H., The circulation of the Denmark Strait and Iceland-Scotland Overflow Waters in the North Atlantic, Deep Sea Res., Part A, 31, 1339-1355, 1984.

Swift, J.H., and K. Aagaard, Seasonal transitions and water mass formation in the Iceland and Greenland Seas. Deep Sea Res., Part A, 29, 1107- 1129, 1981.

Swift, J.H., K. Aagaard, and S.V. Maimberg, The contribution of the Denmark Strait Overflow to the deep North Atlantic, Deep Sea Res., Part A, 2 7, 29-42, 1980.

Sy, A., M. Rhein, J.R.N. Lazier, K.P. Koltermann, J. Meincke, A. Putzka, and M. Bersch, Surprisingly rapid spreading of newly formed intermediate waters across the North Atlantic Ocean, Nature, 386, 675- 679, 1997.

Talley, L.D., and M.S. McCartney, Distribution and circulation of Labrador Sea water, J. Phys. Oceanogr., 12, 1189-1205, 1982.

Tsuchiya, M., L.D. Talley, and M.S. McCartney, An eastern Atlantic section from Iceland southward across the equator, Deep Sea Res. Part A, 39, 1885-1917, 1992.

Walker, S.J., R.F. Weiss, and P.K. Salameh, Reconstructed histories of the annual mean atmospheric mole fractions for the halocarbons CFC-11, CFC-12, CFC-113, and carbon tetrachloride, J. Geophys. Res., this issue.

Wallace, D.W.R., and J.R.N. Lazier, Anthropogenic chlorofluoromethanes in newly formed Labrador Sea Water, Nature, 332, 61-63, 1988.

Wallace, D.W.R., P. Beining, and A. Putzka, Carbon tetrachloride and chlorofluorocarbons in the South Atlantic Ocean 19øS, J. Geophys. Res., 99, 7803-7819, 1994.

Warner, M.J., Chlorofluoromethanes, F-I 1 and F-12: Their solubilities in water and seawater and studies of their distributions in the South

Atlantic and North Pacific Oceans, Ph.D. thesis, 124 pp., Univ. of Calif., San Diego, La Jolla, 1988.

Warner, M.J. and R.F. Weiss, Solubilities of chlorofluorocarbons 11 and 12 in water and seawater, Deep Sea Res., Part A, 32, 1485-1497, 1985.

Warner, M.J., J.L. Bullister, D.P. Wisegarver, R.H. Gammon, and R.F. Weiss, Basin-wide distributions of chlorofluorocarbons CFC-I 1 and CFC-12 in the North Pacific: 1985-1989, J. Geophys. Res., I01, 20,525-20,542, 1996.

Watts, D.R., Equatorward currents in temperatures 1.8-6.0øC on the continental slope in the Mid-Atlantic Bight, in Deep Convection and Deep Water For•nation in the Ocean, edited by P.C. Chu and J.C. Gascard, pp. 183-196, Elsevier Sci., New York, 1991.

Weiss, R.F., J.L. Bullister, R.H. Gammon, and M.J. Warner, Atmosperhic chlorofluoromethanes in the deep equatorial Atlantic, Nature, 314, 608- 610, 1985.

SMETHIE ET AL.: TRACING THE FLOW OF NORTH ATLANTIC DEEP WATER 14,323

Weiss, R.F., J.L. Bullister, F.A. Van Woy, M.J. Warner, P.K. Salameh, and R.H. Gammon, Transient tracers in the ocean, in Tropical Atlantic Study: Chlorofiuorocarbon Measurements, Ref. Ser., Vol. 91-1, 159 pp., Scripps Inst. of Oceanogr., La Jolla, Calif., 1991.

Weiss, R.F., M.J. Warner, P.K. Salameh, F.A. Van Woy, and K.G. Harrison, South Atlantic Ventilation Experiment: SIO Chlorofiuorocarbon Measurements, Ref. Ser., Vol. 93-49, 466 pp., Scripps Inst. of Oceanogr., La Jolla, Calif., 466 pp., 1993.

Worthington, L.V., On the North Atlantic Circulation, Johns Hopkins Oceanogr. Stud., 6, 110 pp., 1976.

Wiist, G., Schichtung und Zirkulation des Atlantischen Ozeans: Die Stratosphare, in Wissenschafiliche Ergebnisse der Deutschen Atlantischen Expedition auf dem Forschungs-und Vermessungsschiff "Meteor" 1925-1927, vol. 6, 1935. (English translation, 112 pp., Amerind, New Delhi, India, 1978.)

R. A. Fine, Rosenstiel School of Marine and Atmospheric Science, University of Miami, 4600 Rickenbacker Causeway, Miami, FL 33149. (rfine @ rsmas.miami.edu)

E. P. Jones, Bedford Institute of Oceanography, P.O. Box 1006, Dartmouth, Nova Scotia, Canada B2Y 4A2. Oonesp@mar.dfo-mpo.gc.ca)

A. Putzka, Institut far Umweltphysik, University of Bremen, FB 1, D28359 Bremen, Germany. (putzka@physik. uni-bremen.de)

W. M. Smethie Jr., Lamont-Doherty Earth Observatory of Columbia University, Palisades, NY 10964. (bsmeth@ ldeo. columbia.edu)

(Received April 7, 1998; revised April, 1999; accepted June 22, 1999.)