Mean currents and current variability in the iceland basin

135

Netherlands Journal of Sea Research 33 (2): 135-145 (1995)

MEAN CURRENTS AND CURRENT VARIABILITY IN THE ICELAND BASIN

HENDRIK M. VAN AKEN Netherlands Institute for Sea Research, P.O. Box 59, 1790 AB Den Burg, Texel, The Netherlands

ABSTRACT Long-term (>6 months) current measurements from five moorings in the Iceland Basin have been analysed for the mean currents and the structure of the variable current components. The time-averaged flow at all five moorings had a strong baroclinic character. The mean circulation in the upper layers with relatively warm Sub-Polar Mode Water appears to have a general north-eastward direction with maximum mean velocities of 6 to 7 cm.s "1. In the bottom layer south of Iceland, where the cold IcelandScotland Overflow Water flows westwards along the topography in a Deep Northern Boundary Current, mean velocities of the order of 10 to 20 cm-s" have been observed. Over the deep slope of the Hatton Bank, water enters the Iceland Basin in a branch of the Deep Northern Boundary Current which has a cyclonic rotation sense in the Iceland Basin. The variable part of the current has been analysed by means of principal-component analysis. The current variations in the central Iceland Basin appear to have a mainly barotropic character while variations in the baroclinic flow of Iceland-Scotland Overflow Water contributed 10% or less to the total energy of the variable deep flow. Over the slope of the Hatton Bank the variable currents had a mainly baroclinic character with shear in both current speed and direction. Comparison of the geostrophic velocity with the mean Eulerian velocity has revealed that the ae=27.725 kg.m "3 surface can be used adequately as level of no-motion for the geostrophic modelling of the flow along the Icelandic and Hatton slopes. The mean westward geostrophic transport of ISOW south of Iceland relative to this reference surface amounted to 3.5 Sv, in agreement with existing independent estimates.

Key words: currents, variability, Iceland Basin

1. INTRODUCTION The Iceland Basin is a choking point for the global thermohaline circulation. Through this basin SubPolar Mode Water (SPMW) in the surface layer flows to the Faeroese waters where it enters the Norwegian Sea. This water type originates from the North Atlantic Current and is characterized by high temperature and salinity. In the opposite direction, cold waters, formed in the Nordic seas, enter the bottom layer of the Iceland Basin as Iceland Scotland Overflow Water (ISOW) via overflow across sills in the Faeroe Bank Channel and on the Iceland-Faeroe Ridge (Schmitz & McCartney, 1993). The circulation in the Iceland Basin is strongly influenced by the topography which guides the ISOW in a Deep Northern Boundary Current (DNBC) along the Icelandic slope toward the slope of the Reykjanes Ridge (McCartney, 1992; Van Aken & De Boer, 1994)o Along the slope of the Hatton Bank Lower Deep Water (LDW) is

assumed to flow into the Iceland Basin, where it joins the flow of ISOW in the DNBC (McCartney, 1992). In the Iceland Basin this LDW is entrained by the ISOW and contributes to the total transport of the DNBC along the Icelandic and Reykjanes slopes (Van Aken & De Boer, 1994). Because of tracer distributions the DNBC is assumed to have a cyclonic circulation in the Iceland Basin (McCartney, 1992; Tsuchiya et aL, 1992). Most of our present knowledge of the circulation in the Iceland Basin is obtained from hydrographic surveys (Harvey & Theodorou, 1986; McCartney, 1992; Van Aken & De Boer, 1994). Only few current meter observations have been reported from the Iceland Basin, concentrated on the flow of ISOW in the bottom layer (Shor et aL, 1977; Saunders, 1993). For the surface circulation some drifter measurements have become available recently (Otto & Van Aken, pers. comm.). In 1989, 1990, and 1991 hydrographic surveys (CTD) were carried out by RV 'Tyro' in the Iceland

136

H.M. VAN AKEN

64

63

62

61

60

59

58

57 -32

-30

-28

-26

-24

-22

-20

-18

-16

Fig. 1. Topography of the Iceland Basin (depth in m) with the positions of the moorings indicated by crosses (x). The dots indicate the positions of CTD stations along two hydrographic survey lines (AE and AR7) occupied in April 1991. Basin as part of the DUTCH-WARP programme. This programme studies the circulation and hydrography of the Iceland Basin (Van Aken & De Boer, 1994). One of the focal points of this programme is the transport and water mass modification of ISOW in the DNBC. In order to be able to carry out quantitative estimates of the circulation in the Iceland Basin by means of geostrophic calculations or inverse modelling, it is a prerequisite to have insight into the spatial structure and the temporal variability of the currents. For the study of the relative importance of barotropic and baroclinic modes for the circulation in the Iceland Basin current meter moorings were deployed in 1989 and 1990 at strategic points in the Iceland Basin (Fig. 1, Table 1). On the steep slope south of Iceland moorings IB89/ 1 and IB90/1 were deployed in two consecutive years at approximately the same position. Moorings IB90/3 and IB90/4 were situated on the slopes of the Hatton Bank and the Reykjanes Ridge, respectively, while IB90/2 had a position in the deeper and more level central part of the northern Iceland Basin. All moor-

ings contained four current meters, of which two were mounted at 40 and 200 m from the bottom. In this paper the currents measured at the moorings are described in terms of mean Eulerian velocities, their directional stability and variance. The structure of the variable part of the flow is studied by means of principal component analysis. The mean Eulerian velocity is compared with estimates of the geostrophic velocity profiles, based on the density distribution along hydrographic sections, as observed during hydrographic surveys of the Iceland Basin. The main conclusions are summarized in the final section. 2. OBSERVATIONS The geographic position of the moorings is given in Fig. 1. The position of the current meters projected on the temperature field as observed along nearby CTD sections from 1991 (Van Aken & De Boer, 1994) is shown in Fig. 2. The uppermost current meters at moorings IB90/2, IB90/3 and IB90/4 were located

MEAN CURRENTS AND CURRENT VARIABILITY IN THE ICELAND BASIN

mooring

TABLE 1 Mooring information. water depth (m) depth CM (m)

latitude

longitude

IB89/1

61 °33.1' N

20°00.8' W

2120

IB90/1

61 °34.1' N

19°57.8' W

2118

IB90/2

60°59.6' N

19°59.3' W

2388

IB90/3

59°16.1' N

18°24.7' W

1906

IB90/4

58°45.1 ' N

28°39.6, W

1923

above the permanent thermocline in the SPMW. The lower two current meters at moorings IB89/1, IB90/1 and IB90/2 were located in the ISOW layer. The full section AE (Fig. 1) was surveyed in 1990 and 1991 with a station spacing of about 28 km (15 nautical miles). In 1989 this section was surveyed only north of 60°N. Moorings IB89/1, IB90/1, IB90/2, and IB90/3 were within a few km of section AE. Mooring IB90/4 was situated about 25 km north of WOCE Hydrographic Program section AR7, which had a station spacing of 56 km. In addition to the CTD data from RV 'Tyro', CTD data from RV 'Oceanus' from 1988 along the 20°W meridian were available for comparison. For this study NBA DNC-2M current meters were used. Data were recorded every 30 min. The data recovery was on average slightly over six months per current meter, while two out of the 20 current meters failed completely (Table 1). In order to eliminate tidal motion the velocity components were low-pass filtered with a 1/e cut off frequency of 1/(30 hr). These low-pass filtered data were used for further processing, described below. A detailed description of the data and data processing is given by Van Aken & Ober (1991; 1992). The mean current components from all moorings as well as their standard deviations are listed in Table 2. Added to this table is also the Neumann/Weyers directional stability parameter B, defined by: B = 100 c/c(%)

(eq. 1)

1320 1520 1920 2080 1318 1518 1918 2078 488 1388 2188 2348 306 1006 1706 1866 323 1023 1723 1883

137

first record

last record

11-Aug-89 11-Aug-89 11-Aug-89 no data 16-Jul-90 16-Jul-90 16-Jul-90 16-Jul-90 14-Jul-90 14-Jul-90 14-Jul-90 14-Jul-90 26-Jul-90 26-Jul-90 26-Jul-90 26-Jul-90 11-Jul-90 11-Jul-90 11-Jul-90 no data

04-Jan-90 02-Jan-90 28-Dec-89 07-Sep-90 05-Jan-91 30-Jan-91 13-Feb-91 12-Feb-91 13-Jan-91 12-Jan-91 09-Feb-91 13-Feb-91 28-Jan-91 09-Feb-91 03-Feb-91 01-Feb-91 19-Jan-91 28-Dec-90

j

L 0 c',

IOO0

2000

slope

3

L [3_

a i 5000

i

i

i

63

I

i i 62

I B 89,/,, 1 . IB90/1 IB90/2 i i i i i i i i 61

~.._____._.~ ~-'~--~ i i i i i i 60

/ IE i

i

i

i

59

LGfifude "Z" c~ "0

1000

2000 L [3_

3000

51

50

29

28

27

26

25

Longitude

Fig. 2. Position of the moored current meters, projected on the potential temperature field from the nearby CTD sections. (a) section AE; (b) section AR7.

138

H.M. VAN AKEN

TABLE 2 Mean velocity components (standard deviation in brackets), directional stability B and number of days recorded.

east (cm.s7)

north(cm.s1)

stability(%)

days recorded

IB89/1 IB89/1 IB89/1 IB89/1

mooring ident, depth (m) 1320 1520 1920 2080

distanceto bottom (m) 800 600 200 40

-1.2 (7.8) -2.3 (7.9) -12.8 (7.4)

1.6 (7.2) 0.7 (7.7) 1.4 (5.7) no data recorded

24 27 92

147 145 140

IB90/1 IB90/1 IB90/1 IB90/1

1318 1518 1918 2078

800 600 200 40

-3.1 (13.8) -3.9 (9.0) -15.2 (7.8) -17.7 (8.0)

(9.9) (8.3) (3.1) (5.0)

30 46 99 99

54 173 198 212

IB90/2 IB90/2 IB90/2 IB90/2

488 1388 2188 2348

1900 1000 200 40

6.1 6.2 4.6 3.1

(7.7) (8.0) (6.6) (6.3)

2.6 (8.7) 1.7 (11.2) -2.9 (7.7) -4.7 (6.4)

58 51 55 69

213 183 182 210

IB90/3 IB90/3 IB90/3 IB90/3

306 1006 1706 1866

1600 900 200 40

1.9 -0.4 2.4 3.9

(7.8) (3.7) (4.6) (4.8)

59 40 61 75

202 186 198 192

IB90/4 IB90/4 IB90/4 IB90/4

323 1023 1723 1883

1600 900 200 40

0.4 (5.2) 1.4 (4.5) -0.5 (2.2)

15 40 30

205 192 169

with c being the magnitude of the vector-averaged velocity and c the arithmetically averaged magnitude of the velocity. This parameter can vary between 0% 25 cm/s

2.1 1.0 -1.5 -7.2

5.9 1.4 2.2 3.0

(9.2) (3.2) (4.6) (3.9)

0.5 (5.1) 1.3 (2.9) 0.4 (1.5) no data recorded

and 100% and is assumed to be a good measure of the directional stability of the current (Neumann, 1968). For a uni-directional flow B will amount 100%. 25 cm/s

IB90/2

I B90/3

2188m

a I

I

I

I

I

I

I

I

I

1-Jul-90 1-Aug-90 1-Sep-90 1-Oct-90 1-Nov-90 1-Dec-90 l-Jan-91 1-Feb-91 l-Mar-91

I'''"1 180

..... I ..... I ..... I'''"1 210

240

..... I ..... I ..... I

270 300 330 Day Number(1990)

360

390

420

b I

I

I

I

I

I

I

I

I

1-Jul-90 1-Aug-90 1-Sep-90 1-Oct-90 1-Nov-90 1-Dec-90 1-Jan-91 1-Feb-91 1-Mar-91

I"'"1 180

..... I""'1 210

240

. . . . . I' .... I ' ' ' " 1 270 300 330 Day Number(1990)

..... I''"'1 360

390

Fig. 3. Stick plots of the low-pass velocity vectors from mooring IB90/2 (a), and mooring IB90/3 (b).

420

MEAN CURRENTS AND CURRENT VARIABILITY IN THE ICELAND BASIN

62-X

61--

59-/ a 58

' -30

631,1

.

I

'

-28

~

I

'

-26

I

I

I

'

-24

I

I

I

'

-22

I

I

I -20

I

I

-18

I

I

-16

L.

I

o,, -30

28

-26

24

-22

-20

-18

-16

Fig. 4. Mean current vectors with the surrounding topography. (a) current vectors in the upper 1520 m; the current vectors in the SPMW are marked with a cross (x); (b) current vectors from the lowest 200 m above the bottom. The currents at 40 m from the bottom are marked with a cross (x). Stick plots of the low-pass velocity vectors of moorings IB90/2 and IB90/3 which cover most of the water column are shown in Fig. 3. These moorings are typical of the different vertical velocity structures as observed in the Iceland Basin. Whereas the current variations at IB90/2 in the relatively smooth central Iceland Basin are quite coherent over the whole water column (Fig. 3a), the currents at IB90/3 over the steep Hatton slope vary considerably across the main thermocline (Fig. 3b). The typical time scale of the current variations amounts 5 to 10 days for moorings IB89/1 and IB90/1 to 10 to 15 days for IB90/2, IB90/3 and IB90/4. 3. THE MEAN CURRENTS In Fig. 4a the mean velocity vectors in the upper 1520 m are depicted, while Fig. 4b shows the velocity vectors in the lowest 200 m. It is clear from Table 2 and Fig. 4 that the mean current is strongly baroclinic at

139

all mooring positions since the shear vector between the lowest and uppermost current meters has a magnitude of the order of the mean current velocity. In the upper part of the water column the mean Eulerian velocity vectors had a direction in the north to east quadrant at moorings IB90/2, IB90/3 and IB90/4 (Fig. 4a). This indicates a general north-eastward flow of the thermocline water and the Sub-Polar Mode Water (SPMW) over a large part of the deep Iceland Basin. Over the steep south Icelandic slope (moorings IB89/1 and IB90/1) the current had a mean west to north-west direction between 1318 and 1520 m. At this position and depth interval the water mainly consists of a mixture of SPMW, Labrador Sea Water (LSW) and ISOW (Van Aken & De Boer, 1994). Probably part of the north-eastward flowing surface water re-circulates in a cyclonic way close to the Icelandic shelf, in agreement with the hydrographic evidence which indicates a westward flow of this slope water over the steep Icelandic slope (Van Aken & De Boer, 1994). This circulation scheme agrees with the Lagrangean surface drifter tracks, reported by Otto & Van Aken (pers. comm.), who observed a general north-eastward surface drift along the Hatton slope, as well as some cyclonic re-circulation along the Icelandic and Reykjanes slopes. The directional stability in the upper 1500 m was of the order of 50% or less (Table 2), indicating considerable eddy activity relative to the mean motion in the Iceland Basin as was also reported by Otto & Van Aken (pers. comm.). Estimated from the standard deviation of the current velocity (Table 2) the typical eddy velocity in the SPMW is of the order of 10 cm's -1. The flow in the bottom layer (Fig. 4b) shows large mean velocities, O (10 to 20 cm.s-1), over the steep Icelandic slope where ISOW flows westwards in the DNBC. At IB90/1 a maximum velocity of over 40 cm.s -1 , persisting for two days, has been observed 40 m from the bottom. The order of magnitude of the mean currents in the bottom layer at moorings IB89/1 and IB90/1 is comparable with the order of magnitude of the short-term mean currents (6 to 20 cm.sq), reported by Shor et aL (1977) for positions over the Icelandic slope 100 to 150 km east of our moorings. The current direction more or less agrees with the direction of the isobaths, indicative of topographic steering of this strong baroclinic flow. The difference in direction between the mean currents 200 m from the bottom at moorings IB89/1 and IB90/1, only a few miles apart, but from different years, is probably due to small-scale local topographic features. A considerable mean shear in south-south-westerly direction is observed between 200 and 40 m from the bottom at mooring IB90/1 (Fig. 4b). In order to check whether the southward veering of the flow direction is due to friction effects, the thickness of the bottom Ekman layer was estimated according to Weatherly & Martin (1978) to be of the order of 20 to 30 m. This agrees with observations by Williams & Armi (1991) in the

140

H.M. VAN AKEN

4. THE STRUCTURE OF LOW-FREQUENCY deep boundary current near Newfoundland and rules out frictional effects as cause of the observed direcVARIATIONS tional shear between 40 and 200 m above the bottom. Most probably local density gradients in the flow The standard deviation of the velocity components direction, caused by the containment of the baroclinic was everywhere, with the exception of the lowest 200 flow by the irregular topography of the Icelandic m at moorings IB89/1 and IB90/1, larger than the slope, are responsible for the observed veering of the magnitude of the mean velocity (Table 2). This is remean current direction at mooring IB90/1. flected by the directional stability B which is nearly The near-bottom velocity vector from mooring IB90/ 100% in the bottom layer at moorings IB89/1 and 2, about 65 km south of IB89/1 & IB90/1, showed a IB90/1, due to the topographic confinement of the mean south-eastward flow (Fig. 4b). The lateral veer- near-bottom flow. Also for moorings IB90/2 and IB90/ ing of the deep flow direction between the two moor- 3 the B maxima were found at the lowest current ing sites is attributed to topographic steering by an meters. The topographic steering of the near-bottom extension of the West Katla Ridge (Fig. 1; Malmberg, flow at IB90/3 can be seen clearly in Fig. 3b. The low1975), which rises up to 200 m above the neighbour- est B values (20-40%) were found over the Reykjaing Icelandic slope. A qualitatively similar, although nes Ridge (mooring IB90/4) and between 1318 and less extensive lateral veering was observed by Shor 1520 m over the steep Icelandic slope (moorings et aL (1977) near the East Katla Ridge. Tracer distri- IB89/1 and IB90/1). There the flow was irregular with butions in the ISOW (Van Aken & De Boer, 1994) no clearly prevailing flow direction. The B values from confirm that the flow of ISOW is mainly around the the other current meters were of the order of 50 to extension of the West Katla Ridge rather than across 60%. it. For the determination of the structure of the variaAt mooring IB90/3, positioned on the north-western ble flow, the low-pass filtered velocity components slope of the Hatton Bank, a mean north-eastward cur- were de-trended by subtracting a least squares linear rent with the maximum velocity at 40 m from the bot- approximation. The resulting de-trended components tom was observed with high directional stability present the variable part of the current with time (B=75%), closely following the isobaths (Fig. 4b). The scales of one day to about six months. Not any signifmean directional shear between 40 and 200 m from icant correlation was observed between the variations the bottom was small, compared to the moorings over of the current components of current meters mounted the Icelandic slope discussed above. The flow along in different moorings, not even between moorings the more regular topography of the north-western IB90/1 and IB90/2, only 65 km apart. This indicates Hatton Bank appears to be less well able to generate that the variability of the currents is mainly caused by density gradients in the flow direction compared to transient eddies with horizontal scales of less than 65 the south Icelandic slope with its many ridges and km. canyons. The north-eastward flow along the Hatton 100 Bank probably contributes to the DNBC and transports water of a more southern origin, among which LDW, into the Iceland Basin. A similar deep boundary / current which transports deep water in northern direc80 tions has also been observed along the edges of the // Porcupine Abyssal Plain (Dickson et al., 1985). LDW transported in this eastern boundary current is 60 assumed to re-circulate in the Iceland Basin and to modify ISOW properties (McCartney, 1992; Van Aken E & De Boer, 1994). With an inflow along the Hatton 0 Bank and a westward flow along the Icelandic slope the deep circulation of the Iceland Basin has a e ~ cyclonic character. i! The deep flow observed at mooring IB90/4 is ~ 20 extremely small (Table 2) and did not show any 3 southward flow as expected by Van Aken & De Boer (1994). Probably the south-flowing bottom waters originating from ISOW mixed with SPMW, which folI I I I I I I I low the topographic contours, was diverted here, due 1 2 3 4 5 6 7 8 to the locally rough topography of the Reykjanes EOF number slope, with numerous isolated peaks and ridges. Because of the low velocities encountered at IB90/4, Fig. 5. Cumulative contribution (%) of the EOFs to the total this mooring is left out of the further analysis dis- energy of the variable motion. IB90/2 full line, IB90/3 dashed cussed below. line.

/,

117

MEAN CURRENTS AND CURRENT VARIABILITY IN THE ICELAND BASIN

~

IB90/2

G) tO O.

E 0 o

..C 0

o

Z

,, j-

o-

141

•

•

•,

..~ ~

#o

306m\.'/,./

o~°

1006 m ~ , , 1 "

488 m 1388 m 2188 m "%2348m

o o

IB90/3

East component

East component

Fig. 6. Vectors at four levels, representing EOF1 (full line) and EOF2 (dashed lines) for IB90/2 (a) and IB90/3 (b). We assume that the observed current variations are caused by a superposition of barotropic and baroclinic waves or eddies. While barotropic waves will have no vertical shear at all, the baroclinic waves are characterized by a definite vertical velocity structure, depending on the baroclinic mode number (LeBIond & Mysak, 1978). In the deep Iceland Basin the first baroclinic mode has its level of no horizontal motion at a typical level of 1000 m. In order to study the vertical structure of the current variations empirically, principal component analysis (Preisendorfer, 1988) was performed on the variable current components from the current meters within a single mooring. Different methods exist for the principal component analysis of vector series (Kundu et aL, 1975; Mercier & Colin de Verdiere, 1985; Owens, 1985; Lippert & Briscoe, 1990). Here we will follow Owens (1985) and Lippert & Briscoe (1990) and present the most general (i.e. least constrained) approach by considering the velocity components as individual real valued scalar quantities, qi(t) (i=1 . . . . . . 2N), where N is the number of current meters in a mooring. With principal component analysis an empirical orthogonal presentation of these data is generated, where the original data are represented by: 2N

qi (t) = ~, eijX j (t)

(eq. 2)

j=l

Here eij are the i-components of the j-th empirical orthogonal function (EOFj) and X,j(t) are the corresponding time (t) dependent amphtudes. The amplitudes are not correlated, that is

(XiXj) = IjSij

(eq. 3)

where the angle brackets represent the ensemble average over t, Ii is the variance of amplitude Xi, and 5ii is the Kronec'ker delta. This implies that the varialice of the scalar parameter qi is formed by the linear combination of the contributions of the different EOFs according to 2N

( qiq; = ~, I] e2

(eq. 4)

j=l If there is some structure in the relation between the variables qi usually a few EOFs account for most of the variance of q. Therefore the EOFs are ordered for decreasing variance. For this study the EOFs and their amplitudes have been determined according to the methods described by Preisendorfer (1988). An EOF, based on a series of N current vectors, can be represented as a vector in a 2N-dimensional space, but also as a series of N 2-dimensional vectors. Each of these 2-D vectors are related to one of the current meters. We will use this property to present the structure of each EOF as a series of N 2D vectors. Since we only have data from three or four current meters per mooring, only the barotropic and lowest baroclinic mode can be investigated from the EOFs. Effects of higher baroclinic modes will be aliased into the low order EOFs. Moorings IB90/2 and IB90/3 (both with N=4) cover a large part (>80%) of the water column. The current components from these moorings have been developed into EOFs (Figs 5 and 6).

142

H.M. VAN AKEN

order of magnitude over distances of 700 km (Lippert & Briscoe, 1990). As will be shown below, the dynamic character of the eddy motion also varies in the Iceland Basin. For mooring IB90/3 it appears from the cumulative contributions to the total energy (Fig. 5, dashed line) that the first two EOFs contributed 66% to the total energy. But for IB90/3 both EOF1 and EOF2 showed a strong shear in direction as well as in velocity, concentrated in the main thermocline between 300 and 1000 m (Fig 6b). The directional shear of about 120 ° between 306 and 1006 m in these EOFs indicates that the variable currents are neither purely barotropic nor purely baroclinic. This implies that over the slope of the Hatton Bank the variable part of the current has a mixed barotropic-baroclinic character. The baroclinic part may be connected with eddies in the per-

100

¢1

,'- 80

60 c

~

40

E

20

o

I

I

1

2

I

I

3 4 EOF number

I

5

EOFI&2

EOF3&4

Fig. 7. Cumulative contribution (%) of the EOFs to the total energy of the variable motion of the currents below 1000 m. IB89/1 full line, IB90/1 short dashes, and IB90/2 long dashes. The cumulative contribution of the EOFs to the total variance or energy at the levels of the current meters at IB90/2 (Fig. 5, full line) shows that the first two EOFs accounted for 81% (EOF1 56%, EOF2 25%) of the total energy while the remaining EOFs each contributed only a few percent to the total energy. EOF1 for IB90/2, showed nearly parallel vectors with only a small shear in magnitude and direction as is expected for barotropic currents (Fig. 6a). Also EOF2 was presented by almost parallel vectors with low shear, nearly perpendicular to the vectors representing EOF1 at all four current meter levels (Fig. 6a, dashed lines). This suggests that the current variance at the site of mooring IB90/2 is formed mainly by barotropic eddies or waves, generating a NW-SE oriented variance ellipse. No predominant rotation sense of the barotropic current vector could be established by means of a rotary-component spectral analysis (Gonella, 1972). The typical time scales for the variations in the amplitudes of EOF1 and 2 appears to be 12 to 14 days, of the same order as found by Saunders (1993) south of Iceland. Because of the limited vertical extension of his moorings, the latter was not able to discriminate between barotropic and baroclinic variability. These results agree with the findings of Owens (1985) for the low-frequency variability observed during the POLYMODE experiment, and can probably be ascribed to topographic Rossby waves. However, the predominance of barotropic waves may be a local feature of the deep Iceland Basin. The ratio of barotropic and baroclinic eddy energy in the Atlantic Ocean is known to vary an

i

,'el

b

a

East component

East component

c

~.)

o d

C

East component

East component

& 8 o East component

East corn )onent

Fig. 8. Vectors at three levels, representing the EOF structure for the currents below 1300 m at IB89/1 (a, b), IB90/1 (c, d) and IB90/2 (e, f). EOF1 and EOF3 are drawn in full lines, EOF2 and EOF4 are dashed.

MEAN CURRENTS AND CURRENT VARIABILITY IN THE ICELAND BASIN

143

I

500 --

I-

~'~1000 "1~ v ~

-

~

-

I

,

13. 15o0 - -

, I

I

2000 -

i

I'

i I

---

2500

'

W

I'

I

In II

-

I i

IB90/2 • i ,

IB89/1 • IB90/1 o

-

I

a I'

I'

-20 -10 0 Velocity (cm/s)

,I

, ~I

10 E

,

IB90/3 • b

c

~

W

-10 0 10 Velocity (cm/s)

E

SW

-10 0 10 Velocity (cm/s)

NE

Fig. 9. Mean Eulerian velocity components (circles) perpendicular to section AE for moorings IB89/1, IB90/1 (a), IB90/2 (b), and IB90/3 (c), plotted with the geostrophic velocity profiles relative to the Oe=27.725 kg.m-3 surface (lines) based on the repeated CTD surveys by RV 'Tyro'. The error bars indicate the standard deviation of the Eulerian velocity (thick bars) and of the geostrophic velocity (thin bars). manent thermocline below the core of warm SPMW (Fig. 2a). In order to study the structure and relative importance of the variations in the flow of ISOW, the data from the deepest three current meters from moorings IB89/1, IB90/1 and IB90/2 were analysed by means of principal component analysis. Since here only data from below the permanent thermocline are involved (Fig. 2), it is expected that the analysis will combine the barotropic mode and the first baroclinic mode into single EOFs, while the EOFs with a baroclinic character are expected to be caused by variations in the flow of ISOW. The resulting EOF1 and EOF2 in all three cases represented well over 80% of the total energy (Fig. 7), while their structure was mainly 'barotropic' with low shear in direction and velocity (Figs 8a, c, e). EOF3 and EOF4 had a clear baroclinic character with the shear concentrated above 200 m from the bottom at moorings IB89/1 and IB90/1 and at about 200 m at mooring IB90/2 with EOF3 more or less perpendicular to EOF4 (Figs 8b, d, f). These EOFs are clearly connected with the baroclinic variations in the flow of ISOW relative to the overlying water. However, the

total energy involved in these baroclinic modes contributed only of the order of 10% or less to the total energy (Fig. 7), as measured with the lowest three current meters. 5. COMPARISON WITH GEOSTROPHIC CURRENTS Geostrophic currents relative to the oe=27.725 kg.m -3 level have been calculated for the station pairs on the AE section which surrounded moorings IB89/ 1, IB90/1, IB90/2 and IB90/3. This density level is situated between the permanent thermocline and the core of LSW at a typical depth of approximately 1200 m. Over the Icelandic and Hatton slopes this level coincided with a mid-water column extreme of the geostrophic velocity (Figs 9a, c). From comparison of geostrophic velocity with short-term current observations, Worthington & Volkmann (1965) also found that a similar extreme in the geostrophic velocity is the best choice for a reference level to be used when estimating the transport of ISOW over the slope of the Reykjanes Ridge near the Charlie-Gibbs Fracture

144

H.M. VAN AKEN

TABLE 3 Westward geostrophic transport of the ISOW (ce>27.8 kg.m-3) across 20 ° W. vessel

Oceanus Tyro Tyro Tyro mean transport standard deviation

time

transport

July 1998 July 1989 July 1990 April 1991

3.8 Sv 4.0 Sv 3.2 Sv 3.1 Sv 3.5 Sv 0.4 Sv

Zone. For the first three moorings hydrographic section data of RV 'Tyro' cruises are used from 1989, 1990 and 1991, while for the last mooring only data from 1990 and 1991 can be used. From these calculations mean geostrophic velocity profiles have been constructed. Although the standard deviation of the geostrophic velocity relative to the mean profile is based on only a few individual profiles, it is assumed to be roughly indicative of the baroclinic variability. With the ce=27.725 kg.m -3 reference level the mean geostrophic velocity component perpendicular to section AE agreed well with the mean Eulerian velocity components from moorings IB89/1, IB90/1 and IB90/ 3 (Fi.~s 9a, c) with an RMS difference of only 1.2 cm.s-'. The standard deviation of the geostrophic flow in the lowest 800 dbar near IB89/1 and IB90/1 was definitely less than the standard deviation of the east components of the Eulerian velocity (Fig. 9a). This supports our earlier findings that baroclinic variations of the east component of the Eulerian velocity of the overflow are considerably less important th.an the low-shear 'barotropic' variations. The east component of the mean velocities at mooring IB90/2 which, contrary to the mean current vectors, have a mainly barotropic character differed considerably from the calculated mean geostrophic velocity (Fig. 9b). The mean deviation here amounted 4.3 cm's -1. These results indicate that over the steep parts of the topography the mean circulation can be modelled quite well with a geostrophic flow relative to a reference level which more or less coincides with the geostrophic velocity minimum. The standard deviation of the geostrophic flow near IB90/3 (Fig. 9c) confirms the results from the principal component analysis that over the Hatton slope a considerable baroclinic variability exists in the upper 1000 dbar of the water column. Also over the steep Icelandic slope such a geostrophic variability, although slightly less (Fig. 9a), is observed in the upper layers. Over the deeper and more level parts of the Iceland Basin, considerable time-dependent as well as permanent barotropic flows may exist with an unknown horizontal scale. The standard deviation of the geostrophic velocity near IB90/2 (Fig. 9b) confirms that the variability of the baroclinic flow was small compared with

the observed mainly barotropic variability, attributed to topographic Rossby waves. But because of the conservation of potential vorticity and the limited depths of the sills near the Faeroer (-450 and -850 m) compared with the central Iceland Basin (-2000 to -2700 m), only a limited part of this barotropic transport will be directly connected with the Norwegian Sea. Therefore most of the barotropic flow must be involved in recirculation of the water masses in the central Iceland Basin rather than with the mean through-flow. The results presented above indicate that geostrophic calculations relative to the ce=27.725 kg'm -3 reference level may well be used to calculate the mean transport of ISOW in the Iceland Basin. We have calculated this transport across the 20°W meridian north of 60°N, that is over the Icelandic slope, for the four available hydrographic sections (Table 3). The resulting mean transport of 3.5 Sv (1 Sv=106 m3.s -1) nearly equals the existing most reliable transport estimate of 3.4 Sv ISOW south of Iceland, based on a section of long-term current meter moorings (Saunders, 1993; Dickson & Brown, 1994). The relatively low standard deviation of the geostrophic transports (0.4 Sv) is a clear indication of the minor importance of baroclinic variations of the overflow, in agreement with the results from the principal-component analysis of the Eulerian velocity data. 6. SUMMARY AND CONCLUSIONS Long-term current observations from five moorings in the Iceland Basin have been analysed to determine the structure of the mean circulation as well as the structure of the variable or eddy part of the motion. The mean flow, as observed with the current meter moorings in the Iceland Basin had a strong baroclinic character with shear in direction as well as in velocity. The mean currents in the SPMW above the permanent thermocline had a general north-eastward direction. At deeper levels, in the DNBC, the flow was cyclonic, with a north-eastward inflow along the slope of the Hatton Bank and a westward outflow over the steep Icelandic slope. Especially in the lowest 200 m over the steep part of the Icelandic slope at moorings IB89/1 and IB90/1 the mean flow transports cold ISOW with mean velocities of the order of 10 to 20 cm.s -1 and with very high (>90%) directional stability. Irregularities in the topography such as ridges appear to guide the baroclinic near-bottom flow by influencing the density field. Topographic effects are assumed also to be responsible for the south-eastward flow of ISOW as observed at IB90/2, in the centre of the northern Iceland Basin. Over the slope of the Reykjanes Ridge, with a water depth of 1923 m, the mean currents were very weak and had a low directional stability. The variable part of the flow had no significant correlation between the different moorings. The vertical

MEAN CURRENTS AND CURRENT VARIABILITY IN THE ICELAND BASIN

structure of the current variations, analysed by means of a principal-component analysis, appeared to be mainly barotropic in the deep central Iceland Basin while over the slopes of the Hatton Bank these variations had a considerable baroclinic contribution with maximum shear in the permanent thermocline. Over the Icelandic slope, where ISOW is found near the bottom, the main variations of the currents below the permanent thermocline presented by the first 2 EOFs revealed a low shear in direction and velocity. There EOF3 and 4 represented variations in the baroclinic flow of ISOW along the bottom. However, these baroclinic modes only contributed of the order of 10% to the total eddy energy. Comparison of the mean Eulerian velocities with mean geostrophic velocity profiles alon,g section has AE revealed that the oe=27.725 kg.m -° surface can be used successfully as reference level for the geostrophic modelling of the flow over the Icelandic and Hatton slopes. In the Central Iceland Basin, the Eulerian velocity cannot be modelled adequately using a level of no-motion because of the relatively large barotropic contribution to its east component. However, because of conservation of potential vorticity, the barotropic mean flow is assumed to be mainly involved in recirculation of water in the deep parts of the Iceland Basin and to have less effect for the net through-flow of water masses. Therefore the transport of ISOW over the slope south of Iceland has been calculated with geostrophy relative to the ae=27.725 kg-m -3 surface, for four hydrographic sections from the years 1988 to 1991. The resulting mean transport of ISOW amounted to 3.5 Sv, in agreement with the most reliable independent estimate, based on long-term current meter observations. Acknowledgements.--This research was carried out as part of the DUTCH-WARP programme. This programme was supported by the Netherlands Marine Research Foundation (SOZ) of the Netherlands Foundation for Scientific Research (NWO). The hydrographic data from RV 'Oceanus' were made available by Lynne Talley. 7. REFERENCES Dickson, R.R. & J. Brown, 1994. The production of North Atlantic Deep Water: sources, rates, and pathways.-J. Geophys. Res. 99: 12319-12341. Dickson, R.R., W.J. Gould, T.J. M(Jller& C. Maillard, 1985. Estimates of the mean circulation in the deep (>2000 m) layer of the Eastern North Atlantic.--Deep-Sea Res. 14: 103-127. Gonella, J., 1972. A rotary-component spectral method for analysing meteorological and oceanographic vector time series.--Deep-Sea Res. 19: 833-846.

145

Harvey, J.G. & A. Theodorou, 1986. The circulation of Norwegian Sea overflow water in the eastern North Atlantic.--Oceanologica Acta 9: 393-402. Kundu, P.K., J.S. Allen & R.L. Smith, 1975. Modal decomposition of the velocity field near the Oregon coast.--& Phys. Oceanogr. 5: 683-704. LeBIond, P.H. & L.A. Mysak, 1978. Waves in the ocean. Elsevier, Amsterdam: 1-602. Lippert, A. & M.G. Briscoe, 1990. Observations and EOF analysis of low-frequency variability in the western part of the Gulf Stream re-circulation.---J. Phys. Oceanogr. 20: 646-656. Malmberg, Sv.-A., 1975. A note on the deep water south of Iceland - Overflow '73. ICES C.M. 1975/C:32. McCartney, M.S., 1992. Recirculating components to the deep boundary current of the northern North Atlantic.-Progr. Oceanogr. 29: 283-383. Mercier, H. & A. Colin de Verdiere, 1985. Space and time scales of meso-scale motions in the eastern North Atlantic.-~.l. Phys. Oceanogr. 15: 171-183. Neumann, G., 1968. Ocean currents. Elsevier Scientific Publishing Company, Amsterdam: 1-352. Owens, W.B., 1985. A statistical description of the vertical and horizontal structure of eddy variability on the edge of the Gulf Stream re-circulation.---J. Phys. Oceanogr. 15:195-205. Preisendorfer, R.W., 1988. Principal component analysis in meteorology and oceanography. Elsevier, Amsterdam: 1- 425. Saunders, P.M., 1993. The flux in the dense overflow south of Iceland.--Ann. Geophys. 11: suppl. 2, C142. Schmitz, W.J. & M.S. McCartney, 1993. On the North Atlantic circulation.--Rev. Geophys. 31: 29-49. Shor, A., D. Muller & D. Johnson, 1977. Transport of Norwegian Sea Overflow. ICES C.M. 1977/C:44. Tsuchiya, M., L.D. Talley & M.S. McCartney, 1992. An eastern Atlantic section from Iceland southward across the equator.--Deep-Sea Res. 39:1885-1917. Van Aken, HM. & C.J. De Boer, 1994. On the synoptic hydrography of intermediate and deep water masses in the Iceland Basin.--Deep-Sea Res. (in press). Van Aken, H.M. & S. Ober, 1991. A compilation of moored current meter data from the Iceland Basin (DUTCHWARP 89, mooring IB89/1). NIOZ Data-Report 1991-2, NIOZ, Texel: 1-8. --, 1992. A compilation of moored current meter data from the Iceland Basin (DUTCH-WARP 90, moorings IB90/1, 2, 3 & 4). NIOZ Data-Report 1992-1, NIOZ, Texel: 1-16. Weatherly, G.L. & P.J. Martin, 1978. On the structure and dynamics of the oceanic bottom boundary layer.--& Phys. Oceanogr. 8: 557-570. Williams, R., & L. Armi, 1991. The planetary boundary layer of the deep western boundary undercurrent.--DeepSea Res. 38: 393-399. Worthington, L.V. & G.H. Volkmann, 1965. The volume transport of Norwegian Sea overflow in the north Atlantic.--Deep-Sea Res. 12: 667-676. (accepted 4 January 1995)