Phys. Oceanogr., Vol. 5, No. 4, pp. 295-308(1994) 9 VSP 1994.

Hydrological conditions of water mass subduction in the eastern North Atlantic Ocean (Cruise 43 of the R/V Akademik Vernadskg)* YU. V. ARTAMONOV, A. I. KUBRYAKOV and A. YU. KUFTARKOV Abstract -- This paper reports some results of a large-scalesurveyconductedin the eastern North Atlantic

Ocean on the WOCE project. The weather conditions, the structure of the upper mixed layer, the water circulation and thermohaline structure, as well as the potential vortieitydistribution in the test area, are described. INTRODUCTION Contemporary oceanographic investigations have allowed the major factors governing the structure and properties of the open ocean thermocline to be revealed, namely, the horizontal/vertical advection and the upper mixed layer (UML). The latter has a well-pronounced annual cycle. It is from the UML, where heat, impulse, and energy exchange between the atmosphere and ocean occurs, that water, being potentially vortical, penetrates in some areas into deep layers, thereby causing transformation of the isopycnic surface and of the velocity field there. This process, termed 'subduction', is crucial for describing the general ocean circulation. Besides the large-scale processes (geostrophic and drift currents, vertical convection, etc.), there also seems to be a significant contribution from the small-scale processes of vertical exchange which are likely to lead to diffusion of various characteristics at isopycnic surfaces and to the generation of these motions affecting the latter surfaces. With this in mind, large-scale measurements were carried out at the SUBDUCTION test area during Cruise 43 of the R / V Akademik Vernadsky. In the course of the investigations, the boundary conditions at the sea surface and at the lateral vertical fluid boundary of the test area were observed; in addition, a diagnosis was made for the conditions in the inner part of the test area. Also, small-scale measurements were conducted, designed to identify the mechanisms for the generation of the hydrophysical field's and turbulence fine structure under the test area background conditions and the role of fine structure in the transport of matter and energy. This paper discusses the data collected during a large-scale survey. The experiment included two stages: (1) a reconnoitring survey (from 30 June to 6 July 19991) and (2) a major survey (from 10 July to 3 August 1992). During the initial stage, sounding down to a depth of 1000m was performed by means of an MGI-1201 probe while the ship was moving. The major survey involved 75 soundings down to 2000m and sampling by means of an ISTOK-7 CI'D probe at drift stations, with the spacing between them *Translated by Vladimir A. Puchkin. UDK 551.465.

296

Yu. I4. Artamonov, A. L Kubtyakov and A. Yu. Kuflarkov

being 1~ zonally and meridionally. In addition, 74 CTD probings were accomplished between the stations when the ship was in progress. Current measurements at 25, 75, 125, 250, 650, and 900m depth levels were conducted by MGI-1301 current meters from 10 July to 2 August 1991 at a point with the coordinates 30058.8' N and 23*30.9' W. Moreover, for interpretation of observations, historical data were used. METEOROLOGICAL CONDITIONS IN THE TEST AREA As analysis of the climatic data on wind stress in the subtropical Atlantic has shown, the test area is subject to the impact of two wind systems, namely, north-east trade winds and westward transport from the mid-latitudes. During the year, the wind pattern shifts northwards (from winter to summer) and southwards (from summer to winter). The general meteorological situation in the test area is governed by north-east trade winds. The velocity field of the mean diurnal near-surface wind during the survey is consistent with the climatic distribution characteristic of this time of the year. Thick cloudiness over the test area, as indicated by direct observations and NOAA satellite-derived imagery, occurred only during early July. The acquired weather and actinometric data were used to calculate heat balance components at the sea surface [1]. Thermal fluxes made up 3% of the latent heat, LE. The following average values of the heat balance components were derived: P = 24.7MJm-Zday, H = 0.3MJm-2day, and LE = 10.8MJm-2day. Moreover, if the latent heat flux is positive throughout the area, then the manifest heat flux is positive in the north-western section and negative in the south-eastern section of the test area. Besides direct in situ measurements of the total solar radiation and from the determination of the radiation balance, NOAA's remotely-sensed data were invoked to reconstruct 10day and monthly averages of the total solar radiation at the sea surface and the latter's radiation balance; sea surface temperature maps were also constructed. Comparison of the observed radiation balance with the respective reconstructed values indicated that the radiation balance values, based on satellite-provided data, were 15-20% smaller, which is within the error limits of that parameter determination [2].

,,

51o~M'J._m;.2d'~y

,P

-

!e

\,,, I 51 ~

2.9~

Figure 1. Thermalbalance of the ocean surface.

27 ~

2,f*

23014/

Water mass subduction in the eastern North Atlantic Ocean

297

The derived values of the sea surface heat balance are positive (Fig. 1), as during this time of the year the basic contribution to the latter is provided by the radiation heat flux. The total heat flux from the atmosphere to the ocean in the test area was B = 13.6MJm-2day. THE UPPER MIXED LAYER AND EKMAN DYNAMICS IN THE TEST AREA

In describing subduction processes, it is vital to ascertain the UML's state, as well as to describe its variability and interplay with the underlying layers. One of the basic characteristics of the layer is its thickness, whose determination involves two operations: identifying the upper quasi-thermal layer (AT) and the upper quasi-haline layer (hs), whose thicknesses are usually different. In the case under consideration, the thicknesses of hT and hs were determined by a method based on identifying the local curvature of the vertical profile [3]. Note that the tendencies in the thickness distribution of the upper quasi-thermal layer and the quasi-isohaline layers in the test area coincided. In the northern section of the area, the thicknesses of hr and hs are small, with the seasonal thermocline nearly surfacing at the northern boundary (Fig. 2). To the south, the UML's thickness increases, attaining 20-30 m in the central area and 40-50 m in the southern area. The largest values of Ar and hs, 70 m, were observed at some stations in the south-western part of the area. Figure 3 depicts the difference in the thickness distributions, A h = hs - hT. Note that throughout the area the thickness of the upper quasi-isothermal layer did not exceed that of the upper quasi-isohaline layer, i.e. hs > hr. Over the larger part of the area north of 27 ~N, hs and h r virtually coincided, except for a minor site in the central part. However, in the southern section, the thickness of the isohaline layer is larger than that of the isothermal layer, amounting to 20-30 m. It should be noted that observations conducted in the southern part of the subtropical anticyclonic gyre, in the vicinity of the North Atlantic current, and in the north-eastern tropical Atlantic [3] reveal the reverse picture, i.e. the upper isothermal layer is consistently thicker than the isohaline layer. Obviously, the data collected in a single survey

/V ,rtr/-

,

, / ' - + 1

/I

, ,'31 o

_

_

_

-i

29~

_

J _

/

7

~

~,,. ~

-

29 ~

-2.-

-

~20.,~,--,,_z.~ -. I

I

J ~ / ~ _ _

7

51 ~

-~

- / -

I

, )

__ _

\ ~_~-15 / /;~ _t __ __ _ _ ( r

-

2.7 ~

Figure 2. Thicknessof the upper quasi-thermallayer.

25 ~

25 ~ W

298

Yu. 7. Artamonov, ,4. L Kubryakov and An Yu. Kuftarkov

N

A~,m

'

)/

S ~

,

A" ~

o-~,""-J

1

29 ~

27 ~

I

51~

2.r

27*

25' 61 ~

2S0

2-5'~W

Figure 3. Difference between the thicknesses of the quasi-haline layer and the quasi-thermal layer.

"

wE

/ I

.,310

29 ~

\ ',

I

270

I

25 ~

250W

Figure 4. Vertical velocity at the lower boundary of the Ekman friction layer. (Negative velocityvalues indicate downwetling waters.) do not allow us to determine whether the discrepancy is associated with specific water dynamics in different regions or with differences in the mode of air-sea interaction (the observations in the test area were conducted during the time of intense warming, whereas those discussed in ref. 3 were made during the period of water cooling and entrainment). Figure 4 shows the vertical velocity distribution at the lower boundary of the Ekman friction layer, 1 wE = - - - r o t , , p0 calculated from the wind speed data. Although, according to the climatic data compiled in this region, the vortical components of the tangential wind stress remain negative the

Water mass subduction in the eastern North Atlantic Ocean

299

whole year round, the picture observed at the test area is heterogeneous (it must be remembered, however, that observations were not conducted synchronously). Alongside the downwelling areas, located in the eastern, north-eastern, and south-western sections of the test site, upwelling occurred in the central section. Interestingly, the upwelling and downwelling areas do not necessarily coincide with the respective areas that have minimal and maximal thicknesses of the UML which is indicative of the greater influence of warming and salting processes, as compared with wind forcing. Horizontal Ekman transport over the entire test area was predominantly northwestward. Minimal meridional Ekman transport, integrated over the test area width for each latitudinal circle, was observed across the northern boundary of the area and amounted to 0.26 Sv; the maximal transport occurred of the 28~N circle (1.7 Sv). In both cases, transport was northward. Comparison with climatic data indicates that about 4060% of the Ekman transport occurs across the test area in the eastern margin of the subtropical gyre. WATER CIRCULATION Most of the geostrophic flows in the region under study have positive (anticyclonic) vorticity, which is accounted for by the adjustment of the ocean to the forcing of the large-scale wind field. The eastern subtropical Atlantic is known to accommodate four major elements of large-scale water circulation. Between approximately 30~ and 40* N, the North Atlantic current propagates eastward; in the Western literature a section of that current in the vicinity of the Azores is named the Azores current [4]. East of 35-33~ it turns southward. Close to the European shore, the Portuguese curre0t travels southward. To the south of 30~N, the two currents form the Canary current, which turns south-westward. To the west of the Cape Verde Islands, the currents, now called the North Equatorial current, have predominantly south-western and western components. The subtropical gyre, as a large dynamic feature, is readily traceable in all seasons, although the currents are known to change their position. From April to September, southward transport along Africa's coast is greater than that from October to March, when the subtropical gyre becomes more intensive to the west of the Canaries. A significant point is the following: in summer, the Azores current shifts in the south direction, i.e. below 35* N, whereas in winter it moves to the north [4]. Note that the geostrophic currents' climatic velocities in the study area did not exceed 10ms -1, with seasonal variations of 5-10cms -1. The mean annual transport in the subtropical gyre east of 35"W within the 0-1500m layer is orientated southward and amounts to 10 Sv, with the seasonal variations being 3-5 Sv. Geostrophic current computations, using the dynamics technique, were implemented versus the 700 reference surface. Basically, the currents proved to be weak, the velocity being 10cms -1. A maximum (13cms -1) was observed in the upper 150m thick layer between the eastmost mooring of the zonal transect. Geostrophic velocities decreased rapidly and below 250-300m did not exceed 2-3 cms -1. Schematic patterns of the dynamic topography showed that the geostrophic currents were mostly oriented south- and south-westward, and only at the north-eastern transect did they have an eastern component. The prevalence of the meridional circulation, as indicated by the reconnoitre survey, determined a tack orientation when the principal survey was carried out. Mooring sites were selected within the areas of the largest geostrophic currents (23~W, 30~N).

300

Yu. v. Artamonov, A. L Kubryakov and A. Yu. Kuftarkov

In situ current observations, by and large, confirmed the data compiled in the course of the reconnoitre survey. At all depth levels down to 650 m, the average velocity vectors were 10cms -1 at most, and had a prevalent south component. The highest velocity (10cms -1) was documented at the 25m depth level. At 650m, the average diurnal current velocity vectors were observed to turn south-eastward. At the 900m depth, the currents were orientated eastward and south-eastward, and some average diurnal vectors swung east-north-eastward. During the last days of observations at the 900 m depth level, the current was turned sharply towards north-north-west, its velocity rising up to 20cms -1 (the maximum average diurnal vector being 21 cms-]). Apparently, the enhanced variability of currents at 900 m depicts complex water dynamics in the area of interaction between the North Atlantic and Mediterranean intermediate water masses. Calculation of the time spectra for the current's zonal and meridional components revealed three basic energetic periods: 22h (inertial oscillations), 12.5 h (semi-diurnal tidal fluctuations), and 6 h. In the temperature spectra inertial oscillations were missing. Analysis of the calculated geostrophic current data compiled in the course of the major survey, with respect to the reference zero surface at 1200 m, indicated that current velocities in the area of mooring deployment did not exceed 5-7 cms -1, being mainly southward. Modest velocities were observed throughout the surveyed area. Exceptions were individual stations located in the north-western part of the area, between which geostrophic velocities in the upper 250 m layer reached 10 cm s-1, with the largest values being 14 cm s-~ at 75-200m depth levels. Horizontally, currents in the upper 200 m layer are mainly orientated south-westward (Fig. 5). According to current views, the acquired structure of large-scale flows constitutes part of the south-eastern periphery of the subtropical anticyclonic gyre, i.e. the area where the Azores current and the Canary current turn to the south-west, and where the North Equatorial current is generated. With depth, the current velocities decrease and the currents proper become markedly zonally orientated. Below 500 m, the water circulation comprises several vortices with different signs, amongst which the largest is a cyclonic gyre in the western part of the test area, readily traceable from the sea surface down to a depth of 1000 m.

25"

50" iV i

i

|

1

!

85

I

I

!

I

I

I

I

Figure $. Dynamictopographyat a depth of 100m versus 1200m.

1

!

Water mass subduction in the eastern North Atlantic Ocean

301

Figure 6. Bottom relief in the test area. Figure 6 shows the bottom topography in the test area. It can be seen that in the north-west the test area is bounded by a nearly vertical wall. The ocean depth there is merely 300-500m. Parallel to that elevation, a relatively deep trough with depths exceeding 5000 m passes from the south-east to the south-west. Specific features of the bottom topography may influence the pattern of flows, and a cyclonic vortex observed in the north-western section of the area seems to result from such an influence. Meridional volume transport is concentrated in the upper 500 m layer and is directed southward. The largest meridional transports of heat and volume across the entire test area were observed at 29~ and amounted to 35.7x 1013Js-1 and 5Sv, respectively; that is, over the survey period, approximately half of the transport, retrieved in ref. 5 from the climate density data, occurs across the zonal section of the test area along 29 ~ N. The overall (geostrophic/drift) transport in the test area is orientated southward and amounts to 3-4 Sv. LARGE-SCALE THERMOHALINE STRUCTURE OF WATERS The vertical distribution of the thermohaline characteristics in the test area may be attributed to the monotonic thermal/non-monotonic haline type [6]. The upper part of the water column accommodates the mixed layer, whose structure has been described in detail above. Below the UML, the seasonal thermocline is centred, with vertical gradients exceeding 0.3~ per m, whose depth reflects the tendency in the UML thickness variations. Beneath the seasonal thermocline, at the majority of sections, a layer with relatively small vertical temperature gradients occurs, apparently coupled with the water upwelled from various ocean areas at the end of the cooling season. This layer overlies the main thermocline, at the base of which is centred a layer with relatively low vertical gradients, ostensibly linked with the Mediterranean intermediate

302

Yu. V. Artamonov, A. L Kubryakov and A. Yu. Kuftarkov

water present at those depths [7]. Then the temperature declines with depth, with the gradients becoming larger. The vertical salinity distribution seems to be more complicated than the temperature distribution. In the central and south-western part of the test area, where the UML is well developed, highly-saline waters are observed. In the marginal areas (in the north and east), a structure where the vertical salinity profile in the upper 200 m layer displays two maxima is typical. One of the maxima occurs in the top layer during summer as a result of salinity rising. The other one (subsurface) probably occurs due tothe subduction of saline waters from the halistatics centre. Below the highly-saline waters the main halocline is located, whose salinity decreases from 36.5-37.0ppm (100-200m) to 35.5ppm (700-800m). The absolute salinity maximum occurred at depths of 1000-1100m in the central and south-western sections of the test area; there the salinity declined in the southern direction from 25.5 to 35.1ppm. Between 700 and 1400 m, the salinity increases again, which is rationalized by the influence of the Mediterranean intermediate water. The depth of the Mediterranean salinity maximum vacillates between 850 and 1400 m at various parts of the test area. Starting from the 1400 m depth, the salinity decreases anew throughout the area, attaining a minimum value of 35.5 ppm at 2000 m. The thermohaline structure of the area under study evolves owing to the interaction of the following water masses: (1) surface waters constituted by highly-saline, warm waters in the western and southwestern margins of the test area, waters from the northern periphery of the subtropical gyre transported by the Azores current, and coastal upwelling waters brought by the Canary current; (2) Madeira mode waters (or the upper modification of the central North Atlantic waters), which are an analogue of the 18"C Sargasso Sea water, forming as a result of winter/spring convection north of Madeira and propagating southward and southwestward [8]; (3) central North Atlantic water (or the water mass of the main thermocline) forming as a result of winter-time convection in the vicinity of the subpolar front between approximately 40 ~ and 50 ~ N; (4) Mediterranean intermediate waters entering the test area from the north-east; and (5) North Atlantic deep water masses formed in the vicinity of the subpolar front and centred at depths greater than 1000 m [9].

STRUCTURE OF THE POTENTIAL VORTICITY FIELD In studying the ocean circulation, involving physical interpretation and model verification, of relevance is such a parameter as potential vorticity (PV) [10, 11], which in the large-scale approximation is equivalent to the Coriolis parameter multiplied by the potential density's vertical gradient: I .f q = Po

0a0 Oz

Traditional observations of the seawater parameters provide the data necessary to compile PV charts. One of the major goals of the SUBDUCTION/VENTILATION project,

Water mass subduction in the eastern Noah Atlantic Ocean

5oOw

303

25*

ot,~i'iho" ~0 Q

N

)

~-

2~

(a) 30*W

r*,c

25"

-~ (~OJ/

~d N

2ff

(b) 50*W

s,o~.

2a*

J l~jJ c

50* a

(c) Figure 7. Potential vorticity (a), temperature Co), and salinity (c) at the isopycnic surface r conventional units.

= 26.5

304

Yu. v. Artamonov, A. I. Kubryakov and A. Yu. Kuftarkov

conducted in the framework of the WOCE programme, is to describe the PV field evolution [12]. The charts of the PV field constructed by climatic data [12, 14] show that PV distributions tend to be uniform at the base of the thermocline in the North Atlantic, as distinct from the seasonal thermocline, as well as from the upper and middle parts of the main thermocline,' where a specific distribution of the potential vorticity is observed. In the upper part of the thermocline, the PV values are fairly large near the eastern and western boundaries of the northern subtropical gyre. The latter's central part houses extensive areas of low PV values, which is probably explained by the input of winter-time convection waters to the thermocline. To analyse the PV structure, for each identified parcel of water, we chose the characteristic isopycnic surface where potential vorticity, temperature, and salinity distributions were considered. In the northern part of the test area, between 75 and 200 m, zones of minimal vertical temperature gradients are located approximately between the 17.5 and 19~ isotherms. As ref. 8 claims, that minimum may be indicative of the Madeira mode waters, as the depth and range of temperatures are close in those of ref. 8. Figure 8 shows the potential vorticity, temperature, and salinity distributions at the appropriate isopycnic surface, a0 = 26.5. It is seen that the water under consideration may flow in from the north near 25* W. Its temperature and salinity are somewhat higher compared with the surrounding waters. Apparently, this water is formed during the autumn/winter-time convection and is then injected into the thermocline through subduction north of Madeira, where the isopycnic surface ~r0 = 26.5 is outcropping. Hydrochemical data have revealed that in the vicinity of 25 ~W at the north of the test area, the oxygen concentration in seawater is high, which is also an indirect indication that the water mass is of surface origin. From the optical characteristics, this layer stands out because of the tongue of highly turbid waters there. From the genesis area, the Madeira mode waters propagate southward and south-westward. As exhibited in Fig.7, the PV, temperature, and salinity distributions match one another, implying that motion in the area is essentially isopycnic. However, from the survey data Madeira mode waters can be observed merely up to 30* N. Siedler eta/. [8] also point out that upon their formation, these waters are traceable for about 6 months, covering a range equal to nearly 500km. The fact that subduction has an interstitial nature may be behind this. The point is that over that time, part of the water mass leaving the UML as a result of subduction, as the latter becomes thinner, cannot sink to a sufficiently large depth before the UML starts to deepen again and may be entrained by that deepening layer and contribute anew to the mixing process. This phase of subddction is called temporal subduction [15], as distinct from the effective subduction, when water entrained by the thermocline's geostrophic flow does not return to the UML during the annual cycle. Figure 8 displays the depth of the isopycnic surface t r o = 26.5 and the UML thickness recovered from the climatic data for February [16]. It is seen that, at least in the northern part of the test area, these parameter values are similar to one another. This may imply that the Madeira mode waters, having travelled from the site of their formation to approximately 30~ N, may be again entrained by the deepening UML; thereby contributing to mixing and transforming. Against the background of monotonically decreasing temperature and salinity with depth, temperature-uniform parcels of water evolve in the underlying layers of the North

Water mass subduction in the eastern North Atlantic Ocean

305

Atlantic water masses. The most pronounced amongst these is centred at the depth of the 15~ isotherm (300-400 m). Calculation of the potential vorticity at the sections has shown that both its minimal value and the minimum of its vertical gradient `occur there. According to ref. 17, central North Atlantic waters evolve between 35 ~ and 40 ~ N, as in some seasons the horizontal T, S curves for the sea surface are similar to the vertical T, S curves for the central water mass. Today, we presume that North Atlantic waters propagate from the area of their formation southward, mainly isopycnically. Figure 9 shows the temperature, salinity, and potential vorticity distributions on the isopycnic surface of potential vorticity equal to 26.8 conventional units, and the depth of this surface. The characteristics are distributed most uniformly in the northern part of the test area receiving central North Atlantic waters. The dynamic topography at the isopycnal of 26.8 conventional units displays flows which are directed predominantly westward. South of 30~ N, horizontal heterogeneity of the temperature field is better pronounced. In the potential vorticity and salinity fields, uniformity persists. The average value of the potential vorticity is about 10 x 10-]1 m -1 s -1. Similar values were obtained in ref. 13 at the meridional section along 27 ~ W. Thus, independent calculations indicate the presence of a homogenized layer in the vicinity of the isopycnal of 26.8 conventional units. Note that the depth of the layer at issue changes widely over the area. It is largest (400 m) in the north-western part of the test area. Near the eastern and southern boundaries, it decreases to 240-360m. The same depth alteration is coupled with the general anticyclonic motion of central North Atlantic waters within the main thermocline. Vertical Sections of the potential vorticity reveal another peculiarity. At depths corresponding roughly to the 26.8 conventional unit isopycnal, cores with low potential vorticity (8x 10-11 m -1 s -1) are observed locally. It is argued in ref. 18 that these are the traces of central North Atlantic waters evolving north of 50~ N. As indicated by the q, T, and E distributions on the isopycnic surface tro = 27.2, these waters enter the test area from the north-east. Rhines and Young [19],seem to be the first to have scrutinized the origin of homogenized PV areas in the lower part of the thermocline. They assumed that the relatively weak lateral diffusion of PV along the isopycnic surfaces in the areas bounded

50" 9

9

25* W' 9

9

9

9

w

,

i\o) 50c N

'L ..5'

Figure 8. The depth of the isopycnicsurface tr0 = 26.5 conventionalunits. The dashed line indicates the climatic value of the upper quasi-isothermal layer's thickness for February.

306

Yu. V. Artamonov, A. L Kubryakov and A. Yu. Kuftarkov

50 * W .

.

.

.

25* .

,

9

,

.

9

-t -1 ._'ti (~ jm " JG . IO

)

Ir

25*

5

d

(12-~,

0

(a) 50" W

25 ~

T,~

/.

z# N

250

(b) 50~

25 ~

5,~.o IV

C> 2~

(c) Figure 9. Potential vorticity (a), temperature (b), and salinity (c) at the isopycnic surface as = 26.8 conventional units.

Water mass subduction in the eastern North Atlantic Ocean

307

by closed streamlines must invariably lead to PV homogenization. Another possible rationale was offered in ref. 20 by the authors of the so-called LPS model (a ventilated thermodine model). Lyxten eta/. hold that as the waters injected into the main thermodine evolve through convection, originally they have a rather uniform potential vorticity. Therefore, a variety of low-intensive small-scale processes must occur in the thermodine in order to rationalize the homogenization. A third mechanism was suggested in refs 21 and 22. It relies on the fact that mixing by mesoscale eddies penetrates deeper than PV advection; hence, diffusion may dominate the PV advection at the base of the thermocline, smoothing out all sorts of gradients induced by the subduction. The data set compiled in the test area does not allow us to identify or give preference to one or other mechanism in order to explain the uniform distribution of PV in the layer discussed above. The mean isopycnal, on which the core of the intermediate Mediterranean water is centred, is 27.7 conventional units. The temperature/salinity distribution at that isopyenal indicates that the least transformed Mediterranean waters enter the test area from the north-east. A tongue of relatively warm (7-8~ and highly-saline (35.5 ppm) waters stretches across the entire northern part of the test area in the west direction. The maximum temperature and salinity values there are 8.3~ and 35.6 ppm, respectively. South of 31 ~ N, Mediterranean waters are less pronounced. Notwithstanding this, on the T, S curve, these appear throughout the area by virtue of the deep salinity maximum, which declines to 35.2 ppm at the south-western stations. The depth of the salinity maximum increases in the south-west direction (by the 27.7 conventional unit isopycnal's position) from 1000 m at the northern boundary of the test area to 1300 m at the southern one. The geostrophic circulation at the 27.7 conventional units isopycnal indicates the occurrence of flows, predominantly west-orientated, in the area of Mediterranean water spreading. Calculations of the potential vorticity in the layer between the isopycnals of 27.65 and 27,75 conventional units have shown that the Mediterranean intermediate waters possess a high potential vorticity (5 x 10-11 m - i s -1 in the north-east and 3x 10- n m -1 s-1 in the south-west of the area). The layer containing deep North Atlantic water is strongly homogenized. We calculated the temperature, salinity, and potential vorticity distributions at the 27.8 conventional unit potential density isopycnal situated close to the lower boundary of observations. The ranges of the respective measurements at that isopycnal are 5.5-5.5~ 35.15-35.20ppm, and 1.0-1.5x10 - u m -1 s-1. Note that the values decrease in the south direction. The depth of the 27.8 conventional unit isopycnal changes from 1500 m at the northern margin of the test area to 1800-1900m at the southern one. CONCLUSIONS As a result of surveying the test area and of preliminary analysis of the acquired data, the following conclusions can be made: (1) Two water masses resulting from subduction were present in the test area: (i) Madeira mode waters located in the vicinity of the isopycnic surface ~0 = 26.5 conventional units and (ii) central North Atlantic waters located in the vicinity of the isopycnic surfaces tre = 26.8 - 27.2 conventional units (2) Madeira mode water seems to be affected by temporal subduction and to contribute to thermocline ventilation for a limited period of the annual cycle.

308

Yu. V. Artamonov, A. I. Kubryakov and A. Yu. Kuftarkov

(3) Central North Atlantic waters (near the 26.8-27.2 conventional unit isopycnals) contribute to the continual thermocline ventilation. The most likely area of formation of these waters is the frontal zone situated between the subpolar cyclonic gyre and the northern subtropical one. Only more complete, experimentally derived data and model simulations of the ocean circulation and subduction can give a final answer to the question.

REFERENCES

1. Report from Cruise 43 of the R / V Akademik Vemadsky. Sevastopol: MHI (1993). 2. Timofeev, N.A. Radiation Regime of Oceans. I~ew. Naukova Dumka (1%3). 3. Baev, S.A., Korotaev, G.K. and Kubryakov, A.I. Peculiarities in the distribution of the upper isohaline/isothermal layers' thicknesses in various parts of the North Atlantic. Deposit. Manuscript No. 4496 B-90, Moscow: VINITI (1990). 4. Stramma, L. and Isemer, H.L. Seasonal variability of the meridional temperature fluxes in the North Atlantic Ocean. Z Mar. Res. (1988) 46, 281-299. 5. Artamonov, Yu. V., Polonsky, A.B. and Pereyaslavsky, M.G. Studies of the large-scale water circulation in the north-eastern tropical Atlantic. Deposit. Manuscript No. 992 B-87, Moscow: VINITI (1987). 6. Bulgakov, N.P. Convection in the Ocean. Moscow: Nauka (1976). 7. Kuksa, V.L. Intermediate Waters in the World's Ocean. Leningrad: Gidrometeoizdat (1983). 8. Siedler, G., Kyhl, A. and Zenk, W. The Madeira mode waters. Z Phys. Oceanogr. (1987) 17, 1561-1570. 9. Khlystov, N.Z. Structure and Dynamics of Tropical Atlantic Waters. Kiev: Naukova Dumka (1976). 10. Rhines, P.B. Vorticity dynamics of the ocean general circulation. Anna. Rev. Fluid Mech. (1986)18, 433-497. 11. Bryan, IC Potential vorticity in models of the ocean circulation. Q. J. 1~ Met. Soc. (1987) 113, 713-734. 12. World Ocean Circulation Erpetirnent. Implementation Plan. Vol.2. Scientific Background. Wormley (1988). 13. Keffer, T. The ventilation of the worlds's ocean: maps of the potential vorticity field. J. Phys. Oceanogr. (1985) 15, 509-523. 14. McDowell, S., Rhines, P. B and Keffer, T. North Atlantic potential vorticity and its relation to the general circulation. Z Phys. Oceanogr. (1982) 12, 1417-1436. 15. Cushman-Roisin, B. Subduction. Dynamics of the oceanic surface mixed layer. Proceedings "Aha Hulicda" Hawaiian Winter Workshop. Univ. Hawaii at Manoa (1987), pp. 181-1%. 16. Kuznetsov, A.L. The Upper Mixed Layer in the North Atlantic. Obninsk (1982). 17. Mamaev, O.I. About North Atlantic water masses and their interaction. Trudy MGI (1959) 19, 57-68. 18. Stramma, L. Potential vorticity and volume transport in the eastern North Atlantic from two long CrD sections. Dt. Hydrogr. (1984) 37, 147-155. 19. Rhines, P.B. and Young, W. A theory of wind-driven ocean circulation. 1. Mid-ocean gyres. Z Mar. Res. (1982) 40, 559-596. 20. Luyten, J.R., Pedlosky, J. and Stommel, H. The ventilated thermocline. J. Phys. Oceanogr. (1983) 13, 292-309. 21. Cox, M.D. An eddy-resolving numerical model of the ventilated thermocline. J. Phys. Oceanogr.

(1985) 15, 1312-1334. 22. Boning, C. and Cox, M.D. Particle dispersion and mixing of the conservative properties in an eddyresolving model (1988). (Available from GFDL/NOAA Ocean Group Technical Report).