| water environment international |

Eastward Flow Through the Mid-Atlantic Ridge at 11oN and Its Influence on the Abyss of the Eastern Basin Dr. M. S. McCartney, Department of Physical Oceanography, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts, USA
Dr. Sara L. Bennett, Water Environment International / Northwest Hydraulic Consultants, Dhaka, Bangladesh / Edmonton, Canada
M.E. Woodgate-Jones, Department of Physical Oceanography, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts USA

Contribution No. 5858 from the Woods Hole Oceanographic Institution.  This paper was originally published in J. Phys. Oceanogr., 21, 1089-1121.  NB: this online version is for general information only.  Published version is authoritative.

2.1 Early Observations
Discovery of the Vema Fracture Zone and Primary Sill
Secondary Sill and Down-Channel Evolution of Bottom Temperature
Transport Estimates
2.2 Observations in 1983
Vema Exit Sill Topography
Transport at the Exit Sill
1983 Transport Compared to Earlier Estimates
3.2 Lateral Extent of Vema Influence
3.3 Abyssal Circulation, Part 1: Gambia Abyssal Plain
  Eastward Current Immediately Outside the Vema
Eastward Flow Along the Plain's Southern Boundary
Kane Gap Throughflow
Cyclonic Gyre in the Eastern Plain
Poleward Flow Along the Mid-Atlantic Ridge

3.4 Abyssal Circulation, Part 2: South Canary and Canary Basins

Northward Flow Across Cape Verde Ridge
South Canary Basin Bottom Water Circulation
Bottom Water Flow into North Canary Basin

ABSTRACT The dilute Antarctic Bottom Water of the North Atlantic eastern trough is supplied from the western trough through fractures in the Mid-Atlantic Ridge. In particular, the influence on eastern trough property distributions of flow through the Romanche and Vema Fracture Zones, near the equator and 11oN respectively, has been noted previously.

Here we report new observations that document the abyssal circulation of the northeastern Atlantic Basins (Gambia Abyssal Plain, South Canary Basin, and North Canary Basin), in particular, the dominance of Vema influence, the absence of Romanche influence, and the existence of a system of deep western boundary currents.

Deep isopycnals slope steeply across the Vema's eastern end near 39oW (previous abyssal observations within the Vema were limited to isolated stations west of its 41oW sill), corresponding to a geostrophic transport through the Vema of 1.4 to 2.3 x 106 m3s-1 colder than 2.0 C. This is half or more of the estimated Bottom Water that flows north into the subtropical western Atlantic in a `transposed' boundary current along the western flank of the mid-Atlantic Ridge. Were this flow not transposed, much warmer water would be flowing through the Vema (the temperature at the primary sill would be 1.73 C instead of 1.40 C).

Further east, where the Vema debouches into the Gambia Abyssal Plain, a deep western boundary current with 1.3 to 3.0 x 106m3s-1 transport colder than 2.0 C flows eastward. This current subsequently bifurcates into (1) a nearly zonal eastward current (< 1.0 x 106m3s-1 transport) along the Plain's southern boundary, the Sierra Leone Rise, and (2) a northward western boundary current along the flank of the mid-Atlantic Ridge (3.0 to 4.5 x 106m3s-1 transport).

Property and shear fields indicate that, for water colder than 2.0 C, basically none of the eastward flow along the Rise passes southward through its deepest passage, the Kane Gap, nor does Romanche-derived water flow north there. The poleward mid-Atlantic Ridge flank flow in the Plain continues northward across the Cape Verde Ridge into the South Canary Basin, and from there poleward into the North Canary Basin.

1.   INTRODUCTION By the 1930's, sufficient abyssal water mass measurements and depth soundings in the North and South Atlantic had accumulated for Wst (1933, 1935) to construct charts and sections of deep and bottom waters and to infer circulation pathways and topographic constraints from them. In the western trough, he found western-intensified Weddell Sea influence (later named ``Antarctic Bottom Water'' or simply ``Bottom Water'') spreading northward from the Antarctic zone as far as the northern subtropics. Along the spreading path, he found the coldest bottom potential temperature QB increasing steadily northward, from < -0.8 C immediately north of the Weddell Sea, to +0.6 C at the equator, and finally to +1.8 C near Bermuda.

In the eastern trough, Wst found a different bottom water configuration, with the coldest bottom temperatures/ located near the equator and bottom temperatures increasing from equator to pole in both hemispheres. Wst (1933) attributed this to eastward flow through the Romanche Fracture Zone, followed by poleward spreading. Later studies (Metcalf, Heezen, and Stalcup, 1964) confirmed the presence of cold water in the Romanche and its influence on the abyssal water of the Guinea and Sierra Leone Basins of the equatorial and southeastern Atlantic (Figure 2A shows regional bathymetry; bathymetry names conform to current international usage which differs from physical oceanography usage as noted in figure caption).

A second eastward flow through the Mid-Atlantic Ridge, near 8oN and weaker than that through the Romanche, was later suggested by Wst (1935) based on new depth soundings and observations of an isolated QB minimum east of the Ridge at 12oN (QB= 1.74 C at Meteor station 305 near 34oW). Much later, Heezen, Gerard, and Tharp (1964) described seven fractures between 7 and 13oN, including one at 8oN, Wst's preferred location.

The deepest of Heezen et al.'s seven fractures is the Vema Fracture Zone at 11o N; they inferred an approximate location for the sill from a west-to-east jump in QB across 41oW which was later confirmed and sharpened by the Seabeam bathymetric survey of Vangriesheim (1980).

Since Wst's time, the relative impact of Vema and Romanche Fracture Zone flows on the eastern trough has remained unclear, despite the accumulation of new hydrographic and bathymetric data. The Worthington and Wright (1970) atlas [based on the earliest Atlantic data set with salinometer salinity determinations, the IGY data (summarized by Fuglister, 1960) plus a few additional 1960's stations] shows an isolated pool of cold water (QB£  1.8 C) east of the Mid-Atlantic Ridge and north of the Sierra Leone Rise, but does not resolve its connection path to the western trough parent water mass (Figure 1A).

Later, in a comprehensive review of the world ocean's deep circulation, Warren (1981) wrote:

. . . the principal inflow of water below the general crest of the Mid-Atlantic Ridge (3500 m depth, say) occurs from the western Atlantic through the Romanche Fracture Zone on the equator (e.g., Drygalski, 1904; Wst, 1933; Metcalf, Heezen, and Stalcup, 1964). No estimate of the rate of this inflow has been made. Silica values on the surface where potential temperature is 2 C increase northward (Metcalf, 1969), as well as southward (Chan, Drummond, Edmond and Grant, 1977), indicating further that the deep water ``ages'' poleward.

Many additional fracture zones cut across the Mid-Atlantic Ridge, but the existing evidence suggests that there are only three others through which noticeable deep water transport may take place . . . Indications of eastward flow through the Vema Fracture Zone near 11o N were shown by Heezen, Gerard, and Tharp (1964), but the flux must be much less than that through the Romanche Fracture zone, given its much smaller influence on property distributions in the eastern Atlantic ....

In contrast, Mantyla and Reid (1983), drawing on much the same literature and data as Warren and writing only a couple of years later, say: Evidently, the Romansche Gap is the primary passage for bottom water flow into the eastern equatorial and southeastern Atlantic Basins, while the Vema Fracture Zone is the main passage for bottom water flow into the northeastern Atlantic basins. Another matter for debate has been the existence of deep western boundary currents in the northeastern Atlantic Basins (specifically the Gambia Abyssal Plain, the South Canary Basin, and the Canary Basin). Of this area, Warren (1981) states: The dynamical fact that poleward movement from an equatorial source can be accomplished by an interior flow may explain to some extent why no deep boundary currents along the western margin of the eastern trough (Mid-Atlantic Ridge) have been observed in transatlantic sections (Fuglister, 1960). On the other hand . . . the budgetary function of western boundary currents is not only to transport deep water into low latitudes, but also to correct imbalances between interior flows and upward fluxes; and it is not easy to see how continuity can be maintained in the eastern trough of the Atlantic without recirculation boundary currents. The new high-quality CTD hydrographic transects reported here (Figure 2B) - part of a set whose collection began while Warren and Mantyla and Reid were writing their reviews - in large measure settle the question of the relative importance of Romanche and Vema flow on the abyssal characteristics of the eastern trough. The new data rather firmly support Mantyla and Reid's inference that eastward flow through the Vema Fracture Zone is the dominant influence governing the water mass characteristics of the abyssal northeastern Atlantic Basins: the abyss of the Plain is found to be extensively and strongly influenced by water from the Vema, in effect, filled with Vema water. Romanche-derived water is not evident north of Kane Gap. The new data also document the northeastern Atlantic Basins' deep water mass distribution and deep circulation, including their deep western boundary currents.


2.1 Early Observations

Discovery of the Vema Fracture Zone and Primary Sill

Heezen et al. observed that a succession of fracture zones offsets the Mid-Atlantic Ridge axis progressively westward in the northern tropics. These fractures are most easily recognized in echo-sounding transects where they are floored by extensions of the western trough abyssal plain: returns from the flat floors stand out from the generally noisy background signal of the rough Ridge topography. Heezen et al. proposed that Wst's low-latitude eastward leak of Bottom Water across the Ridge occurs through the deepest of these tropical fractures, the Vema Fracture Zone at 11o N.

As Heezen et al. described, the Vema offsets the Mid-Atlantic Ridge axis west from roughly 41oW to 43.5o W. The Demerara abyssal plain of the western trough extends into the Vema about 5o eastward, at about 5200 m depth and 8 to 20 km width. Though the floor breaks up east of 41o, its depth stays below 5000 m as far east as 39o W. The Vema's bounding scarps rise steeply above the floor by a few hundred to 3000 meters; the north wall gradually drops to depths of about 4500 m at the Ridge's eastern flank.

Heezen et al. inferred the existence of a sill between 41 and 40.5oW at approximately 4500 m depth, from a jump in near-bottom (about 5200 m) temperature between these two longitudes. At this depth, the same near-bottom temperature (1.27 C) was observed in the western trough and in the Vema at 41oW, while further east in the Vema at 40.5oW a significantly higher temperature (Q  = 1.48 C) was observed. This warmer temperature corresponds to the temperature at 4500 m depth and 41o W.

The inferred sill configuration was corroborated by a 1977 Seabeam survey (reported by Vangriesheim, 1980; data collected by D. Needham). The sill is a triple one, with depths of 4690, 4650, and 4710 m (all 10 m), at 41o 01'W, 40o 55'W, and 40o 53'W, respectively.

Secondary Sill and Down-Channel Changes in Bottom Temperature

Heezen et al. speculated that the Vema also had a secondary sill, which would account for Wst's observation of warmer (1.74 C) near-bottom temperatures nearby in the eastern trough than at the primary sill. Observed bottom temperatures in the Vema (shown in Figure XX) evolve down-channel from a remarkably uniform value west of the sill (1.30 to 1.34 C) to a sharply higher value just east of the sill (1.40 C). Proceeding eastward from the sill, temperatures warm at a rate of about 0.1 C per 100 km. The available bottom temperature observations by themselves do not reveal if the 0.1 C per 100 km warming east of the sill occurs gradually or in jumps at (an) additional sill(s). We present evidence below that Heezen et al.'s speculation regarding a secondary sill was correct.

Transport Estimates

Both Vangriesheim and Eittreim et al. estimated cold water transport in the Vema from cold water cross-sectional area multiplied by the time-averaged down channel velocity component from short-duration near-bottom current meter records. The resulting transport estimates for water colder than 1.5 C ranged from nearly zero to 0.46 x 106m3s-1. All of Vangriesheim's and Eittreim et al.'s current meters were deployed on the flanks of the Vema or on the sill itself; instrument pressure limitations precluded placing instruments in the central channel. Deployment durations were short, between 8 and 33 days. Neither Vangriesheim or Eittreim et al. have data suitable for geostrophic calculations, though the latter paper includes a temperature section at 42o 35'W showing sloping isotherms suggestive of eastward geostrophic flow.

2.2  1983 Observations

We first occupied seven CTD stations near the Vema in early 1983 while aboard the R/V Oceanus, after completing a 20 station transect of the western trough near 13o N. These seven stations cross the Mid-Atlantic Ridge eastward, then cross the Vema southeastward about 240 km east of the primary sill, following the Ridge's eastern flank from 13o to 10o N (Figure 2B).

We found very cold Bottom Water (QB = 1.66 C at 4800 m on station 26), 0.08 C colder than Wst's observation some 500 km further east, at a station in a deep north of the Vema and separated from it by 4600 m topography (Figure 3); the station missed the deep's maximum depth by at least 200 m. A station was not obtained in the maximum depths of the Vema, due to time restrictions and a deteriorating echo-sounding system.

To investigate further the distribution and flow of Bottom Water near and through the Vema, we returned on the R/V Knorr during a north-south Atlantic transect near 35o W in late 1983, deflecting the cruise track to intersect the Vema near 38o30'W (Figure 2B). We first ran the Knorr on a triangular reconnaissance survey (Figure 4) of the Vema's eastern end; our only other bathymetric data at the time was a low-resolution chart (Searle et al., 1982). Without time for any real reflection, we placed a station (station 76) at the base of the Vema's northern wall in 4765 m depth, then proceeded east-southeastwards across the Vema deep channel to the south wall, placing a station (station 79) there in 4630 m depth, finally rejoining the main transect at station 81.

Here we found Bottom Water in a thick layer with low vertical gradients lying beneath a transitional layer between Bottom and Deep Water; Vangriesheim and Eittreim et al. found a similar layer further west. We found bottom temperatures on the northern wall to be as cold (QB = 1.58 C) as in the depths of the channel 600 m deeper. Temperature, s4, nitrate, and silicate sections are shown in Figure 5.

Vema Exit Sill Topography

The topography of the Vema's eastern end is shown in Figure 4 with areas shallower than 4700 m shaded. The data comes from several sources: a Lamont-Doherty Geological Observatory database (D. Martinson, pers. comm., 1986) our own 1983 Knorr and Oceanus tracklines, and 1989 Oceanus stations (Roemmich and Hall, pers. comm., 1989).

Our findings confirm Heezen et al.'s inference of a second sill. We found the eastern Vema to consist of a linear deep (5000 to 5500 m) channel extending east to 38o45'W where depths begin rising to less than 4000 m. The north wall height drops at 39oW from almost 4000 m down to  4700 m, judging from the Knorr bathymetry. The Vema is connected to the eastern trough across this secondary, exit sill, which seems to have fallen exactly at the Knorr triangular survey. Consistent with this topographic configuration, the coldest waters of the Vema proper were not found west of the exit sill.

It must be noted that the interpretation of the topographic data is not entirely straightforward. Around sharp topography, like fracture zone walls, small navigation errors are troublesome, as are spurious echo returns from nearby shallower soundings: the echo sounding at Knorr station 77 registered 4490 m, but we were able to sweat the instrument to 5357 m, registering 5 m above the bottom from the attached pinger. The true bottom eventually emerged from the ghosts after we had steamed about 2 miles towards the next station.

Transport at the Vema Exit Sill

We will argue in this section that the cold (Q  < 2.0 C) water transport eastward out of the Vema is most likely between 1.4 and 2.3 x 106m3s-1, based on reference levels between 2.0 and 2.5 C. These transport values are of clear significance to both eastern and western trough abyssal circulations, representing a bleeding-off of about a third or more of the estimated Bottom Water (Q  < 1.9 C) supply to the western tropical Atlantic, which is 2.3 to 2.8 x 106m3s-1 at 16oS to 8oN (Wright, 1970) and 3.5 to 4.5 x 106m3s-1 at 13oN (McCartney and Speer, 1990). The implications of this leak on western trough abyssal circulation dynamics are examined in McCartney and Speer (1990).

We define four layers such that the flow directions of all but Layer 2 are constrained by property field characteristics:

(1) Bottom Water: < 2.0 C, net eastward flow demanded, eastward flow of all sublayers most plausible.

(2) Transitional Layer: 2.0 to 2.3 C water, net eastward flow most plausible because (a) net eastward flow is demanded in bracketing layers 1 and 3 and (b) evidence for westward Layer 2 flow is lacking.

(3) Deep Water: 2.3 to 4.5 C water, net eastward flow demanded because nitrate minimum must come from western trough.

(4) Intermediate Water: 4.5 C to 9.0 C water, net westward flow demanded because nitrate maximum must come from near Africa.

Net flow means summed over the layer temperature range and over all stations.

The Deep and Intermediate Water layers are defined and their flow directions constrained by the nitrate field (Figure 7D). Two laterally isolated vertical nitrate extrema are seen at the Vema stations. The colder extremum, a minimum with values falling below 20 µmol l-1, extends from Q  = 2.6 to 3.9 C (1600 to 2800 m). This is western basin Deep Water, and its isolated position in the section, surrounded by higher nitrate eastern basin waters, requires that it flow eastward from the western trough. The second extremum, a maximum with values exceeding 40 µmol l-1 extends from Q  = 5.5 to 9 C (400 to 900 m). This seems to be a form of Antarctic Intermediate Water (broadly speaking) that has nitrates exceeding 37 µmol l-1 and is found in the eastern trough off western Africa. In the Knorr 35oW section, it is found only within the Vema. There is no western source of such high nitrate, and thus it must be flowing westward from the eastern trough.

The potential temperature/nitrate relationships for the Vema station group and for the nearest station group to its north are shown as mean curves and standard deviation envelopes (Figure 8). The standard deviation envelopes are distinct for two potential temperature ranges: 2.3 to 4.2 C, part of our Deep Water layer, and 4.7 to 9.0 C, part of our Intermediate layer. For simplicity, the intervening 4.2 to 4.7 C layer is split at 4.5 C and added to the Deep and Intermediate Water ranges.

The deep vertical shear field (proportional to isopycnal slope times the vertical density gradient) is characterized by isotherms in the Deep to Bottom Water transitional layer rising about 400 m to the south, indicative of eastward flow relative to a shallower reference level. The vertical temperature gradient and thus vertical shear are greatest near Q  = 1.9 C; the shear persists up through the water column.

Given the observed shear field, transports in each layer hinge on the reference level choice (Figure 6A). In particular, net Bottom Water transport increases strongly as the reference level moves upward through it, stabilizes at a value of about 2 x 106m3s-1 as the reference level moves through the Deep Water, and then increases again to > 4 x 106m3s-1 as the reference level moves through the lower thermocline. Part of the Bottom Water moves westward for reference levels < 1.85 C.

Reference levels in two temperature ranges produce layer flows in the demanded and plausible directions. The cold range (2.1 C <= Q <= 2.5 C) gives Bottom Water transports between 1.4 and 2.3 x 106m3s-1. The warm range (4.2 <= Q  <= 4.8 C) gives Bottom Water transports between 2.4 and 3.9 x 106m3s-1. The cold and warm reference level ranges are separated by a band where the Deep Water moves westward, not eastward as demanded. Above the upper limit of the warm range, the Intermediate layer moves east.

We have chosen to exclude the warm range transport values from our most likely estimate of Vema Bottom Water transport, because the cold range produces a sensible eastern basin transport field (as discussed below), whereas the warmer range produces boundary current transports that seem rather large. This does not mean that we advocate a single regional reference level, however.

The transport range associated with the cold reference level range expands to 0.35 to 2.3 x 106m3s-1 if the Transitional Layer is allowed to move westward (reference levels 1.85 < Q  < 2.5 C); the transport lower bound drops to zero if, in addition, part of the Bottom Water is allowed to move westward (reference levels 1.85 < Q  < 2.5 C).

The transport for a reference level analogous to Wright's western trough level (top of the transition layer, here roughly Q  = 2.2 C), is 1.4 x 106m3s-1.

In the remainder of the paper, transports associated with the cold reference level range will be given in the text, while the transport graphs analogous to Figure 6A will continue to show transports over the full reference level range. In the entire paper, transport below the deepest common level of two stations is computed from the isopycnal slope at the deepest common level and the density gradient along the deeper station.

1983 Transport Compared to Earlier Estimates

The 1983 transport range lower bound (1.4 x 106m3s-1 for water colder than 2.0 C) is over three times the largest Vangriesheim estimate (0.46 x 106m3s-1 for water colder than 1.5 C). Can these be reconciled? The answer seems to be yes: the channel area occupied by water colder than 1.6 C is 19% of the area occupied by < 2.0 C water and velocities are bottom-intensified, bringing the <1.6 C transport proportion to 25 or 30%, in line with the ratio between Vangriesheim's and our observed transports. Bottom speeds at our exit sill section (Figure 5) are about 20 cm s-1 for reference levels in the preferred range; Vangriesheim's current meter velocity near the bottom was 33 cm s-1.

Another estimate, of 0.0 to 0.7 x 106m3s-1 eastward transport through the Vema, was produced by the box model inversion of Schlitzer (1987) using physical and chemical tracers. It appears that these low values, compared to our observations, resulted from an inaccurate representation of Vema inflow characteristics, in particular an excessively low d14C value.

The model flows are quite sensitive to the 14C balance. As Schlitzer (p. 2962) says, the Vema flow upper bound is constrained by ". . . the 14C balance of the box into which the [Vema] inflow is directed." Also, the model's East Atlantic d14C balance drives equatorial downwelling: this peculiar circulation is required ". . . because of the low d14C content of the West Atlantic Water flowing through the Romanche and Vema Fracture Zones into the East Atlantic." As part of this equatorial downwelling system, the model requires 2.6 to 5.1 x 106m3s-1 flowing eastward through the Romanche and 2.5 to 5.0 x 106m3s-1 flowing northward through Kane Gap.

Schlitzer uses a value of -122 o/oo for the Vema inflow, lower than that in all model boxes except the equatorial deep water box; values of -117, -119, and -118 o/oo are used for the Cape Verde and Canary Basin boxes. A better d14C value for the Vema Bottom Water inflow would be in the range -115 to -110 o/oo, judging from two pairs of d14C observations. One pair (-125 o/oo at 1.18 C and -104 o/oo at 1.76 C) comes from GEOSECS station 37 (12oN, 51oW) in the Vema's western trough source region; the station has oxygen and nutrient profiles similar to stations in the Vema (Bainbridge, 1981; stlund et al., 1987). The second pair (-105.6 at 1.86 C and -119.5 at 1.36 C ) comes from TAS station 20, located less than 20 km south of our 13oN western basin transect station 15 (stlund et al., 1987). Interpolating linearly to the Vema primary sill temperature (1.40 C) gives d14C values of -117 o/oo and -118.5 o/oo at the GEOSECS and TAS stations respectively - both significantly higher than Schlitzer's value. And these provide in effect a lower bound: Vema Bottom Water inflow mean temperature must be higher than the sill temperature, and thus a mean d14C value of -115 to -110 o/oo is likely.

We predict rather different model Romanche, Kane Gap, Vema, and upwelling flows, if the Vema d14C value is changed to a more realistic value. The other Vema inflow characteristics (average temperature, silica, etc.) also need to be adjusted.

Relationship with the Western Trough Transposed Bottom Water Current

A well-known curiosity of the abyss of the North Atlantic eastern trough is its `transposed' Bottom Water current (Warren, 1981, Wright, 1970, McCartney and Speer, 1990), so called because it is found on the eastern side of the western trough. Its geostrophic signature, an eastward 500 m rise in the depth of isotherms near 2.0 C (Figure 3), means that the water at the sill depth of the Vema is colder than it would be if the Bottom Water were not transposed. Temperatures at low-latitude stations along the western flank of the Ridge (e.g. station 18, Figure 3) have temperatures of 1.45 C at the depth of the Vema primary sill, while stations further west have temperatures of 1.73 C (station 11, Figure 3).

As a result, we are tempted to say that the existence of a transposed distribution of Bottom Water in the western trough is responsible for the magnitude of eastward transport of water colder than 2.0 C through the Vema. But could it be the other way around? Does the Vema eastward leak (which supplies deep water for eastern trough upwelling) in some way help to cause the western trough Bottom Water to bank eastward up against the Ridge western flank? This question is explored in a companion study (McCartney and Speer, 1989).


3.1 Eastern Trough Bottom Water Sources

Eastern trough bottom water is derived from flow through the Vema Fracture Zone into the Gambia Abyssal Plain and from flow through the Romanche Fracture Zone into the Sierra Leone Basin. We argue below that the Vema influence determines the abyssal water mass characteristics of the northeastern Atlantic Basins north of the Sierra Leone Rise (the Gambia Abyssal Plain, the South Canary Basin, and the North Canary Basin), and that the influence of Romanche waters on these basins is slight (Figure 2A).

Northward flow of the coldest Romanche waters into the Gambia Abyssal Plain is obstructed by the Sierra Leone Rise. The deepest passage through the Rise is Kane Gap. The Gap's sill depth is between 4389 and 4572 m [Hobart et al. (1975) and Egloff (1972); their charts' 2400 and 2500 uncorrected fathom contours], separated along-channel by a deep of about 4700 m. There is a second, longer and shallower passage through the Rise, west of Gap, that consists of a broad region shallower than 4500 m silled at about 4300 m (Figure 2A).

Hobart et al. (1975) were the first to suggest that deep eastward and northward flow from the Vema might predominate over flow through the Gap in determining the characteristics of the Plain abyss. They based this idea on the eastward warming observed in two thermoprobe measurements in the Plain well west of the Gap (QB = 1.829 C at 22oW and 4815 m, and QB = 1.799 C at 25oW and 5641 m). Indeed, Hobart et al. wondered if flow through Kane Gap might be southward, noting that sediments there had been disturbed by flow, but they were not able to deduce the flow direction.

The Kane Gap sill temperature is about 1.89 C (estimating from the temperature at the approximate sill depth, 4500 m, on a station just to its south, Tropical Atlantic Study station 99, Scripps Institution of Oceanography, 1986). Bottom temperature observations in the Gap deep channel are 1.851 C at 4707 m, 1.854 C at 4673 m (at two additional Hobart et al. thermoprobe measurements, accuracy quoted as 0.01 C), and 1.867 C at 4660 m on the TTAS station mentioned above.

Thus Kane Gap sill water (1.89 C) is much warmer than Vema sill water (1.40 C): indeed, a 1100 m thick layer of water colder than Gap sill water was found well north of the Vema (Knorr station 72, Figure 5) with a bottom temperature of 1.754 C. A similar layer 1200 m thick was found about halfway between the Gap and the Vema (Oceanus station 100), with QB = 1.772 C at 5704 m.

3.2 Lateral Extent of Vema Influence

The extent of the Vema's influence on the Gambia Abyssal Plain's and the Canary Basins' abyssal characteristics can be seen clearly in the near-bottom temperature field (Figure 9). The data used to produced this detailed picture include Mantyla and Reid's (1983) carefully examined and edited data; the 1983 Oceanus and Knorr stations; the 1981 Atlantis II 24oN and 36oN sections; a 1988 Oceanus section near 25oW plus 5 stations near 22oW (made by Talley, Tsuchiya, and McCartney); the few Tropical Atlantic Study stations that reach to < 100 m above the bottom; the 1989 Oceanus 11oN section (kindly supplied by Dean Roemmich and Mindy Hall); the GEOSECS and TTO stations in the area; some Meteor and Rambler stations included in Wst's original study (1933); and, lastly, some early Crawford stations that fill out the picture near the Sierra Leone Rise. This data set is not intended to be all-inclusive, just adequate for perceiving basin-scale trends. The isotherm contours are noisy because they are closely spaced (0.02 C) and unsmoothed, and because of real temperature variability introduced by small-scale topographic features.

Near-bottom temperatures in the Gambia Abyssal Plain generally increase eastward from the Vema to the Gap, confirming Hobart et al.'s isolated thermoprobe observations. Just 300 km northwest of the Kane Gap sill (at our short 22oW section) QB is 0.02 to 0.04 C colder than the Gap sill temperature. The coldest temperatures ( 1.70 C) are observed at the Vema exit section and on the Ridge flank north and east the Vema (Oceanus station 26, Figure 3; Knorr station 75, Figure 5; and six of the 1989 11oN Oceanus stations). The eastward spread through the Gambia Abyssal Plain of Vema cold water is shown to be strikingly narrow and zonal by the 1.70 to 1.76 C contours. The next heavy contour (Q  = 1. 80 C) encompasses a large area of the western and central Gambia Abyssal Plain, bounded in the south, north, and east respectively by the Sierra Leone Rise, the Cape Verde Ridge, and the shoaling depths towards Kane Gap.

Northward western-intensified flow in the northern Plain and in the South Canary Basin is also evident in the crowded band of isotherms that follows the Mid-Atlantic Ridge eastern flank northward from the Vema to at least 30oN. The coldest South Canary Basin observation (QB = 1.83 C) falls in its southwest corner, and is again colder than Kane Gap sill water.

3.3 Abyssal Circulation 1: Gambia Abyssal Plain

Here and in Section 3.4 below we discuss the abyssal circulation of the northeastern Atlantic basins. A circulation cartoon annotated with our favored transport estimates is given in Figure XX.

Eastward Current Immediately Outside the Vema

Abyssal flow exits the Vema moving roughly north-northeast (25o true) and then turns slightly past east (100o true). The Vema exit flow, shear, and topography (see Section 2.2) are adequately resolved by our high-resolution section (Figure 5, Knorr stations 76 to 81).

We have a much fuzzier image of the eastward current: just three stations at low resolution (Figure 5, Knorr stations 74, 75, and 80) lying on a line that runs almost north, just east of the high-resolution Vema exit section. The dramatic northward descent of deep isotherms along this line (300 to 800 m, about twice the descent of isotherms across the Vema) indicates eastward flow relative to a shallower reference level.

Transport estimates for the eastward current are very uncertain. Transport is in the range 1.3 to 3.0 x 106m3s-1 for reference levels in our favored range (2.0 C  Q   2.5 C), including an adjustment for topographic blocking. These values overlaps the most likely Vema exit transport range (1.4 to 2.3 x 106m3s-1, Section 2.2).

Transport is significantly larger (5.1 to 8.4 x 106m3s-1) if the blocking adjustment is omitted. The unadjusted transport computation ignores the topography and uses a simple bottom triangle (defined by the deepest common level of a station pair and deeper station's maximum depth). The blocking adjustment involves scaling the triangle area to account for the blocked portion (or for any additional area, i.e. if two shallow stations bracket a deeper gap). Here we adjust for blocking by computing transport only for water above (i.e. warmer than that found at the level of) the highest significant intervening topography. This level is 1.85 C at station pair 74-75 and 1.80 C at pair 75-80 (filled circles, Figure 6B).

The higher, unadjusted transport range may be wrong and reflect nothing more than the inability of our few widely-spaced stations to resolve the shear with respect to the topography (the section crosses three 30 km-wide fractures parallel to the Vema, whil station spacing is 70 and 102 km). Alternatively, the larger transports could be real and reflect augmentation of Vema exit transport from some source. Three physical mechanisms come to mind: cold water flow through other fractures, entrainment of warmer water, recirculation, or ageostrophic dynamics. We examine these alternatives in turn.

First, we examine the evidence for cold water flow through the fracture zone, choosing the fracture at 12oN since the pathway from the Vema to station 74, which lies in the 12oN fracture zone, is not obvious given the complexity of the intervening topography. The 12oN fracture's estimated sill temperature (roughly in the range 1.49 to 1.89 C), as cold as station 74's bottom temperature (1.749 C), does not rule out eastward transport through the fracture, but the westward warming of its bottom temperatures probably does (Table 2). The fracture's sill temperature was estimated using the same method as in the Vema (see Section 2.1), by reading off temperatures (i) at the estimated sill depth range on the nearest western trough station (Oceanus station 18; 1.489 to 1.785 C), and (ii) at the bottom at a station nearby in the channel, in this case in a rift valley to the north (1.89 C).

Entrainment is another possible source for cold (Q  2.0 C) eastward transport outside the Vema, since this transport's average temperature is warmer than the Vema exit transport average. A recirculating component of the kind mentioned by Warren (1981, see Section 1, second quote) could also contribute: boundary current transport is determined by basin-wide mass balance requirements driven by both point-like sources such as the Vema and distributed ones such as mixing-driven upwelling (Stommel and Arons, 1960). Finally, ageostrophic dynamics may be contributing significantly. A familiar aspect of dense overflows (e.g. the Denmark Strait and the Mediterranean overflows) is that isopycnal slopes normal to the velocity field are geostrophically balanced, while slopes along the axis of the flow are not. To explore this possibility further, we would need a two-dimensional station grid around the Vema's eastern end.

Eastward Flow Along the Plain's Southern Boundary

(i) Mass and property fields at 29oW Some of the eastward current flow, we argue, feeds into flow along the southwestern and southern boundaries of the Gambia Abyssal Plain. This boundary flow can be seen as a cold (Q  < 1.76 C) water tongue extending eastward from the Vema (Figure 9B). The Mid-Atlantic Ridge flank and the Sierra Leone Rise bound the Plain south of the Vema (Figure 2A): south of the Vema at 4500 m depth, the Ridge flank is oriented east-southeastward, turning eastward 3o further south, and meeting the eastward-oriented Rise about 5o further east.

The mass and property field signatures of the eastward flow along the Rise flank are illustrated by the Oceanus ``29oW'' section (Figures 10A, B, and C; the Oceanus data delineates the 1.78 C QB contour of Figure 9B). In the temperature section, isotherms < 1.95 C descend northward from the Rise flank (station 103) to the Cape Verde Ridge (station 89) indicative of eastward flow relative to a shallower reference level. Cold water along the Rise flank is younger than water further north, judging from the northward decrease in oxygen and increase in nitrate (northern stations 89 to 99 vs. southern stations 99 to 103).

Property extrema in the Rise flank flow and in the Vema exit flow are similar:

(a) a slight oxygen minimum layer between the Bottom and Deep Waters, defined at the Vema by the 5.8 ml/l contour (Figure 7C) and at the Rise flank by the 5.7 ml/l contour (Figure 10B);

(b) a nitrate minimum layer in the Deep Water, defined at the Vema by the 20 µmol l-1 contour at 2.6 < Q  < 3.9 C (Figure 7D) and at the Rise flank by the 21 µmol/l contour at similar temperatures (Figure 10C, stations 99 to 102); and

(c) a nitrate maximum layer in the Intermediate Water, reaching values at the Vema of ³  37.5 µmol l-1 (Figure 7D) and at the Rise flank of 35.5 µmol l-1 (Figure 10C, stations 102, next to the strongest Deep Water nitrate minimum at station 101).

Both of the nitrate extrema were used to constrain the reference level choice at the Vema exit (Section 2.2). Paradoxically, the nitrate maximum core has higher values at 35oW (1983 Knorr data) than at 25oW (1988 Oceanus data), even though the core's source in the eastern trough off Africa is closer to 25oW. This may reflect interannual variability in the upwelling region further east that is the source for this layer.

The maximum estimated Bottom Water transport of the Rise flank flow is about 1.0 x 106m3s-1, computed from stations 96 to 104 using a reference level near 2.1 C (Figure 11A). Transport dependence on reference level at the Rise flank (Figure 11A) is different than at the other locations (Figure 6A and 6B): the Bottom Water (Q  <2.0 C) and lower Deep Water (2.0 < Q  <3.0 C) move in the same direction (eastward for levels 1.9 £Q £ 2.35 C) even when the reference level is between them, because isopycnal slopes reverse at about 2.2 C.

The Rise flank cold water transport is significantly less than the transport of the eastward current just outside the Vema (1.3 to 8.4 x 106m3s-1; Section 3.3). We argue below (this Section) that most of this decrease occurs where a northward deep western boundary current branches off the eastward current just outside the Vema.

(ii) Flow dynamics The eastward flow in the southern Gambia Abyssal Plain could be interpreted as any one (or any combination) of three dynamical entities: (i) a topographically channelled jet; (ii) the southern boundary current of the Plain's abyssal circulation, or as (iii) a free jet supplying cold water from the Vema to the Kane Gap where it flows southward into the Sierra Leone Basin. We consider each of these possibilities in turn.

The first possibility is that the eastward flow is channelled by topography. The cold water tongue (Figure 9B) extends eastward from the Vema exit along two fracture zones (Figure 2A) as they deepen eastward from 38oW to about 35oW, then follows the maximum depth pathway (> 5500 m and sometimes > 6000 m) to the Kane Gap, even though maximum depths decrease from 26oW to the Gap. It appears that cold water trapping occurs only in the western part of this pathway: in the eastern part, at 29oW, the coldest temperature (QB= 1.764 C) at 29oW (Figure 10A) does fall at the deepest station (station 99, 5959 m) but is not significantly colder than at an adjacent station (station 98, QB= 1.766 C at 5729 m), and only slightly colder (by < 0.033 C) than at the other five deep stations at this longitude.

The second possibility is that the eastward flow is a southern boundary current that feeds the Plain's interior abyssal circulation. A simple Stommel-Arons (1960) model for the eastern basin abyss consists of a rectangular basin with a mass source in the southwest corner and basin boundaries at the Mid-Atlantic Ridge in the west, Africa in the east, the Cape Verde Ridge in the north, and 10oN in the south. Given this geometry, plus upwelling through the thermocline and northward meridional velocity everywhere, the source flow bifurcates into (i) an eastward-flowing southern boundary current that supplies interior streamlines emerging from the (non-equatorial) southern boundary and (ii) a northward-flowing western boundary current that supplies streamlines emerging from the western boundary. The observations are qualitatively well explained by the model, but the observed eastward flow extends further eastward and is stronger relative to the northward flow than in the model.

The third possibility is that the eastward flow is, wholly or in part, a free jet that supplies Vema water to the Kane Gap where it flows southward into the Sierra Leone Basin. Southward flow at the Gap can be represented in the model by second, southeastern sink at or north of the southern boundary which causes a free eastward jet at the sink latitude. As it turns out, however, the observations, discussed in the next section, indicate that Kane Gap throughflow is negligible.

Kane Gap Throughflow

Traditionally, Bottom Water from the Romanche Fracture zone has been thought to flow northward across the Sierra Leone Rise, through Kane Gap, into the northeastern Atlantic Basins. We will now argue that the flow through Kane Gap is most likely negligible, neither northward nor southward, based geostrophic calculations in the Plain and at the Gap, and on a comparison of Plain and Gap water mass characteristics.

The Vema and Romanche Fracture Zones and their throughflows have rather similar characteristics (Table 3). The Romanche throughflow magnitude not yet been determined directly, but water mass characteristics clearly indicate that it supplies Bottom Water to the eastern trough basins south of the Sierra Leone Rise (the Sierra Leone, Guinea, and Angola Basins).

How much, if any, Romanche influence passes northward across the Rise through Kane Gap (sill temperature 1.89 C; Section 3.1) into the Gambia Abyssal Plain? Apparently, cold (1.89 < Q  < 2.0 C) water transport through Kane Gap is essentially zero for reference levels colder than 2.7 C [Figure 20, Scripps Inst. Oceanogr. (1989), Tropical Atlantic Study cruise Knorr 99, stations 96 to 100]. Shear is negligible in the lower Deep and Bottom Waters at the Gap, in contrast to all previously discussed shear and transport fields where significant cold water transport resulted from deep shear in the cold layers themselves. Water mass characteristics also seem to rule out northward flow: Bottom Water in the Gap is more like the Bottom Water found in the Plain than in the Sierra Leone Basin.

Negligible Kane Gap throughflow of water < 2.0 C is the only solution consistent with the Gambia Abyssal Plain transport field when the same isotherm reference level is used both at the Gap and in the Plain. The argument is as follows. Physically, the eastern Plain is a cul-de-sac below the Kane Gap sill temperature (1.89 C). The net east-west cold water (< 1.85 C) transport across 29oW goes to zero as required (Figure 17A) at a reference level of 2.3 C; incidentally the eastward current transport reaches a maximum (Figure 11A) at this reference level. A reference level colder than this generates net eastward cold flow across 29oW, which for mass balance demands intolerably large upwelling in the eastern Plain (on the order of 20 cm s-1 for net east-west transport of 0.5 x 106m3s-1, given that the area of the 1.85 C isotherm west of 20oW is only 0.3 x 106 km3). A reference level warmer than this produces a southern boundary current flowing unacceptably west from warm to cold. Negligible Gap throughflow implies that the eastern Gambia Abyssal Plain is effectively a cul-de-sac between the sill temperature, 1.89 C and the 2.0 C isotherm.

Cyclonic Gyre in the Eastern Plain

Additional support for and details of this image of a cyclonic gyre in the eastern Plain is given by the tracer fields. A relatively direct return path for the coldest part of the eastward current (<1.85 C) is indicated by the fact that at 29oW the return flow's near-bottom oxygen and nitrate values differ little from those of the eastward flow to the south; proceeding through the gyre, oxygen and nitrate change by only -0.05 ml l-1 and +0.2 µmol l-1. Indeed, the cyclonic gyre colder than 1.85 C is confined west of 22oW except for a small bubble (Figure 18).

The tracer fields suggest that the warmer Bottom Water (1.85 £ Q £  2.0 C) takes a longer path through a gyre that extends further eastward. Whereas for the coldest Bottom Water little tracer contrast between the eastward and westward flows was found at 29oW, substantial changes (about -0.3 ml l-1 and +1.0 µmol l-1 for oxygen and nitrate respectively) are found following the warmer Bottom water, associated with the formation in the westward flow of a curious oxygen minium (< 5.5 ml l-1)/nitrate maximum (>23 µmol l-1) layer centered at about 1.95 C. Related tracer signals can be seen at 22oW (Figure 18). Unfortunately the transport of the warmer Bottom Water at 22oW cannot be computed as the section does not span the basin at the 2.0 C level. Slight southward rising of the relevant isotherms at this longitude appears only when an adjacent station from the Kane Gap section is appended to the 22oW section.

These tracer changes could be partially explained by a longer path length for the warmer layer through a gyre extending further eastward. The residence time of the warmer layer may also be longer, given the larger volume of this layer and the eastward current's transport as a function of temperature. Finally, the eastern Plain continental Rise, as described by Egloff (1972), is covered by "thick wedge of sediments" that "appears to be receiving biogenic pelagic sediment," possibly from the upwelling regime off west Africa. Contact with this layer could be causing elevated oxygen consumption and nitrate production; a low oxygen feature at 11oS beneath the Congo River plume was attributed to contact with river-derived organic sediment on the continental rise (Warren and Speer, 1990).

Whatever the cause of the low oxygen high nutrient character of the warmer Bottom Water of the eastern Gambia Abyssal Plain, its area of influence seems rather limited. It is not visible in the Kane Gap section (Figure 19), north of the Cape Verde Ridge along the 25oW section (figure 10), nor anywhere along the 35oW (Figure 7) and 24oN sections. It does appear on the eastern part of the 16oN section atop the Cape Verde Ridge (Figure 13) where it intersects the 29oW section (Figure 10): the low oxygen feature extends from about 32oW eastward to the continental rise where the layer intersects the sea floor. The `adjusted' oxygen values of the 16oN section are open to question, however.

Poleward Flow Along the Mid-Atlantic Ridge

Poleward flow along the Mid-Atlantic Ridge north of the Vema can be seen as a cold water tongue extending northward from the Vema (Figure 9B). The mass and property field signatures of this flow are illustrated by Knorr stations 70 to 74 (Figure 5A through E). At these stations isotherms rise northwards, indicative of west-northwestwards flow following the Ridge flank, relative to a shallower reference level.

Our best estimate at this station group of the current's cold water transport is 3.0 to 4.5 x 106m3s-1. This estimate was computed with bottom triangle transports reduced to account for topographic blocking and reference levels (2.35 < Q  < 2.6 C) such that the lower Deep Water (2.0 < Q  < 3.0 C) moves in the same direction as the Bottom Water (Q  < 2.0 C) (transport reference level dependence and unadjusted transports for an expanded group of stations, 66 to 74, is shown in Figure 11B).

Evidence that lower Deep Water and Bottom Water are moving in the same direction is provided by the oxygen and nitrate distributions (Figure 7C and 7D). Stations 66 to 74 have higher Deep Water oxygen levels (> 5.8 ml/l at station 73) and lower nitrate levels than the next eight stations to the north; station 66 to 74 values are close to those in the Vema exit flow and in the eastward flow just outside the Vema. A northern origin for station 66 to 74 lower Deep Water seems precluded by the approximately 0.1 ml/l lower oxygen levels observed at 24oN (Atlantis II 109, Figure 12C; 0.1 ml/l is on the edge of significance for cruise-to-cruise comparisons) and the higher nutrient levels there (Figure 12D).

3.4 Abyssal Circulation 2: South Canary and North Canary Basins

Northward Flow Across the Cape Verde Ridge

Further north, Bottom Water from the Vema appears to flow northward across the Cape Verde Ridge into the South Canary Basin, as can be seen in the cold (Q  < 1.8 C) water that extends northward along the Mid-Atlantic Ridge flank and terminates atop the Cape Verde Ridge at about 38oW (Figure 9B). Situated more or less on top of the Cape Verde Ridge, the IGY 16oN sections of temperature, oxygen (Figures 13A and B) , and phosphate (not shown due to low quality) show a western-intensified low temperature, high oxygen, and low phosphate bubble of Deep and Bottom Water at and west of 38oW consistent with northward flow of Vema-influenced water there.

Data is lacking to determine with any precision the extent of cold water penetration into the South Canary Basin and the temperature of the coldest flow across the Cape Verde Ridge. The coldest bottom temperatures at the Cape Verde Ridge are probably found in a > 5500 m deep and 70 km wide gap at 38oW. Unfortunately no observations are available from the depths of this feature; our best guess is QB£  1.75 C, based on a station just to the southwest (QB = 1.75 C at 5124 m and 15o39'N, 38o35'W; Reid, pers. comm., 1988). North of the Ridge, there is no data in the immediate neighborhood of the 38oW gap, and three of the four stations defining the QB contouring are very old: two Rambler stations listed by Wst (1933) from 1895, and Meteor 283 from 1927.

South Canary Basin Bottom Water Circulation

The Mid-Atlantic Ridge axis on the western side of the South Canary Basin curves from a north-south orientation at 24oN to east-west at 30oN. The Basin's deep western boundary current is crossed nearly at right angles by both a 24oN zonal section and also by a 35oW meridional section (respectively, 1981 Atlantis II 109 shown in Figure 12 and Knorr 104 shown in Figure 7). These sections cross the Basin interior and intersect in the southern basin near Knorr station 56 and Atlantis II station 176.

A Gambia Abyssal Plain origin for the cold (Q  < 1.9 C) water of the western South Canary Basin (Knorr stations 54 to 59 and Atlantis II stations 174 to 179) is suggested by its high oxygen (> 5.8 ml/l) and high stratification (see isopycnal spacing, Figures 7B and 12B); both are higher than levels anywhere else in the basin at these temperatures. A flow path along the Mid-Atlantic Ridge flank connects the high oxygen, high stratification pool and the Vema water (Knorr stations 70 to 80) of the 35oW section; their apparent isolation from each other is an artifact of the section orientation.

The South Canary Basin deep western boundary current transport at 24oN (Figure 14A) is 1.8 to 1.9 x 106m3s-1 after adjusting for topographic blocking effects, for cold (< 2.0 C) water and our preferred reference level range, 2.0  £ Q £   2.6 C. The analogous transport ranges at 35oW (Figure 14A) are remarkably similar, overlapping to within about 0.5 Sv. All these transports are, however, rather uncertain due to the very rough topography.

The deep western boundary current transport is smaller by roughly 2 to 3 x 106m3s-1 in the South Canary than in the Gambia Abyssal Plain and crossing Cape Verde Ridge (Figure 11A and B). One would expect boundary current transport streamlines to turn east into the interior circulation, and it is also possible that water leaks into the western trough at the Kane Fracture Zone at 24oN as Metcalf (1969) first suggested on the basis of high silicate levels in the Bottom Water west of the Ridge.

Bottom Water Flow into the North Canary Basin

Poleward flow of Bottom Water along the Ridge flank is even more evident in the southward rising of deep isotherms at about 25oN, 30oW, on two occupations of the beta-triangle eastern side (Figure 15; Armi and Stommel, 1983, and Behringer et al., 1983). Further continuation of Bottom Water transport up into the North Canary Basin is suggested by the western-intensified northward intrusion of the 1.94 and 1.96 C QB contours (Figure 9B).

Deep isotherms rise eastward across the North Canary Basin at 36oN [Figure 16, showing a segment of the Roemmich and Wunsch (1985) 36oN section]. The deep flow evidently northward continues here, but as noted by Saunders (1987), with eastern intensification. The coldest water, now with Q ³  2.0 C continues to have elevated oxygens, near 5.8 ml/l indicative of its immediate origin in the western South Canary Basin.

The North Canary Basin is bounded to the north by the East Azores Fracture Zone scarps. Two narrow passages with depths > 4500 m cut through to the north at 19o 30'W and 16oW (Discovery Gap), judging from the Searle et al. chart. According to Saunders (1987), geostrophic northward cold transport through the passages is 0.55 x 106m3s-1, relative a reference level of 2.3 C based on his analysis of the regional circulation. The transport consists of water with temperatures between slightly above 2.0 C (the coldest water here) and 2.05 C; about two-thirds of the reported transport comes though Discovery Gap.

The northward transport through the passages agrees well with the net transport at 36oN (Figure 16, Atlantis II stations 76 to 86) of 0.58 x 106m3s-1, computed using the same temperature reference level. At 36oN, the northward transport has shifted to the eastern side of the basin: the transport field consists of about 0.83 x 106m3s-1 flowing north in the east (stations 81 to 86) and about 0.37 x 106m3s-1 flowing south in the west (stations 76 to 81). A near-bottom low nitrate pocket (stations 76-77) indicates that the southward flow carries northern source characteristics.

Saunders suggested that the eastern-intensified current is the upstream boundary current supplying a dense overflow. We offer an alternative rationalization. A Stommel-Arons-type (northern hemisphere) basin can exhibit an eastern intensification of interior streamlines, if the flow is dominated by a low-latitude source. The flow field in this case consists of:

  • a northward-flowing boundary current north of the source up to the latitude where the last source streamline peels off into the interior;
  • a southward-flowing boundary current poleward of this latitude, fed by interior streamlines that strike the poleward basin boundary;
  • eastern-intensified northward-current-fed interior streamlines, the eastern intensification increasing to the north; and
  • southward-current-fed interior streamlines in the western interior that carry water cycled through the poleward basin boundary regime.

4. CONCLUSIONS We have confirmed (with new high-quality CTD hydrographic transects, Figure 2B), Mantyla and Reid's inference that eastward flow through the Vema determines the abyssal water mass characteristics of the northeastern Atlantic basins (Gambia Abyssal Plain, South Canary Basin, and North Canary Basin), and laid to rest Warren's idea that Romanche-derived waters play a significant role there. The near-bottom temperature field shows water emanating from Vema and spreading along the southern boundary of the Plain as far as Kane Gap, though none flows south there, and along the eastern flank of the Mid-Atlantic Ridge as far north as the North Canary Basin. Neither northward nor southward flow colder than 2.0 C occurs at Kane Gap.

We have estimated geostrophic transport colder than 2.0 C at the Vema exit, as seen in our 1983 high-resolution station line, to be 1.4 to 2.3 x 106m3s-1 relative to a 2.1 to 2.5 C reference level. This range of reference levels was chosen because it produces Vema flows consistent with nitrate extrema and sensible northeastern basin western boundary current transports. These new transport values are much higher than all previous Vema transport estimates, but these older values were either limited to colder waters (Vangriesheim, Eittreim et al.) or result from model runs that significantly misrepresent the mean tracer characteristics of the Vema cold water flow (Schlitzer). It appears that the transport through the Vema of cold water would be significantly less if the Bottom Water of the western trough were not `transposed' onto the western flank of the Mid-Atlantic Ridge; the dynamics of this interesting relationship are explored elsewhere (McCartney and Speer, 1989).

Vema influence spreads through the eastern Atlantic basins by way of a system of deep boundary currents; the existence of these previously unknown currents was hypothesized but not quite predicted by Warren. Just outside the Vema exit, we found the deep flow to turn from almost northward to about eastward, then to bifurcate into a poleward flow along the flank of the Mid-Atlantic Ridge and an eastward flow along the southern boundary of the Gambia Abyssal Plain. Our favored transport estimates for these flows are summarized in Figure XX. We have argued that the dynamics of the south Plain eastward flow could be those of a topographically channelled jet or of southern boundary current of the Plain's abyssal circulation, or a combination of the two (in principal the eastward flow could feed southward Kane Gap throughflow, but the observations indicate that the throughflow is negligible).

Below 2.0 C, the eastern Plain abyss is a cul-de-sac occupied by cyclonically turning flow. The coldest flow evidently returns westward relatively directly, incurring little change in its nitrate and oxygen values, while the warmer waters seems to swing farther east before turning, in the process gaining nitrate and losing oxygen in significant amounts; these tracer changes may be due to greater aging along a longer flow path, or to contact with organic sediments on the continental rise beneath the west African coastal upwelling regime.

| TOP | HOME |

This web page (previously international environment & development professional's home page) created by water environment international.  All site pages (c) wei. Last modified 26 Aug 2002 .  Comments? Problems? Email wei@bicn.com

You are visitor  since 25 Aug 98