Observations of the deepwater flow into the Baltic Sea

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 101, NO. C4, PAGES 8895-8911, APRIL 15, 1996

Observations of the deepwater flow into the Baltic Sea

Bengt Liljebladh and Anders Stigebrandt Department of Oceanography, G6teborg University, G6teborg, Sweden

Abstract. It has earlier been hypothesized that the deep water flowing into the Baltic Sea forms a bottom pool in the Arkona Sea, just inside the entrance sills. The flow from the pool farther into the Baltic Sea was assumed to be baroclinic geostrophic and controlled by the pool stratification. This was supported by a realistic statistical description of the deepwater flow into the Baltic Sea, computed using historical vertical density profiles from a hydrographical station in the Arkona Sea. However, basin-wide synoptic measurements of density and currents were not available for a direct verification of the pool model. A hydrographic survey was undertaken in the Arkona Sea shortly after a major inflow event in the beginning of 1993. The conductivity-temperature-depth sections reveal a thick bottom pool of deep water separated by a halocline from the surface water of Baltic origin. Approaching the northern coastal boundary from the central parts of the pool, the halocline sank by about 20 rn before hitting the bottom. The ship-mounted acoustic Doppler current profiler recorded a complicated current field. However, subtraction of the assumed barotropic part gives a current field quite similar to the baroclinic geostrophic current field computed from the density distribution. Thus a dense bottom pool including a baroclinic geostrophic boundary current along the northern flank of the pool is quite evident from our measurements.

1. Introduction

The Baltic Sea is a huge estuary (Figure la). The surface area is approximately 370,000 km 2, and the mean depth is about 60 m. The connections to the sea, through Fehmarn Belt (Strait) and the Danish Belts in the Belt Sea and the direct route to Kattegat through Oresund, are rather narrow and shallow, with sill depths of only 18 and 8 m, respectively. The Belt Sea and Kattegat are shallow, with mean depths of about 13 and 23 m, respectively [see Svansson, 1975]. Tides are ex- ceptionally small in Kattegat, with amplitudes of only a few centimeters. The variance of the daily mean sea level, mainly caused by meteorological forcing, is approximately 350 x 10 -4 m 2. About 60% of this is caused by fluctuations with periods of 2 months or shorter. Owing to the choking effect of the narrow and shallow straits in the mouth, cooscillating sea level fluctuations of periods less than about 1 month have reduced amplitudes within the Baltic Sea [cf. Stigebrandt, 1984].

In the Baltic proper (Figure la) a halocline, located approximately between 60- and 80-m depths, separates the upper layer of salinity of 7-8 practical salinity units (psu) from the strongly stratified deep water, where salinity increases with

variable, both with respect to salinity and volume flow. Instan- taneous outflows and inflows are in the range 0-300,000 m3/s, with salinities varying between 8 and 28 psu. Thus short-term inflows, feeding dense bottom currents in the Baltic Sea, may be an order of magnitude greater than the long-term mean.

The water exchange of the Baltic Sea is predominantly barotropic and due to the ever changing sea level difference between Kattegat and the Baltic Sea [e.g., Hela, 1944]. This was shown quantitatively for the (Sresund by Jacobsen [1980] and Omstedt [1987] and for the coupled channel system by Wyrtki [1954] and Stigebrandt [1980, 1992]. Slightly more than 70% of the volume transport goes through Fehmarn and the Danish Belts, while the remainder passes through (Sresund [Jacobsen, 1980]. The amplitude of the barotropic flow has a marked annual cycle, with maximum in early winter. The annually averaged barotropic water exchange, however, does not vary much from year to year [Stigebrandt, 1984].

The salinities of the water forced into the Baltic Sea are

determined by the previous history of outflows of low-saline Baltic surface water and by diapycnic mixing in the Belt Sea and Kattegat with underlying saltier water supplied by Skag-

depth from 10 to about 13 psu. Gravity-forced dense bottom errak. Large amounts of waters of usually low salinity (s) are currents, carrying the water of relatively high salinity from the ß stored in the Kattegat (-250 km 3, S - 15-25 psu) and in the Belt Sea and Kattegat that intermittently spill over the en- Belt Sea and Oresund (-100 km 3, S - 12-16 psu). About 37 trance sills, are important for the maintenance of the vertical stratification in the Baltic proper. The long-term mean water exchange of the Baltic Sea can be described approximately as consisting of an outflow of 30,000 m3/s of Baltic surface water with salinity of 8.5 psu and an inflow of 15,000 m3/s of water with salinity of 17 psu. The difference in volume flows, 15,000 m3/s, is due to the positive freshwater balance of the Baltic Sea. The short-term characteristics of the water exchange are quite

Copyright 1996 by the American Geophysical Union.

Paper number 95JC03303. 0148-0227/96/95JC-03303505.00

km 3 of water have to flow into the Baltic to raise the sea level

by 0.1 m. In periods with high inflow rates (15-20 km3/d) this contribution to the sea level rise in the Baltic Sea is much

greater than the contribution due to the positive freshwater balance (1-2 km3/d). Most inflow events comprise less than 100 km 3 of water. Thus water forced into the Baltic Sea usually has rather low salinity. However, a net baroclinic transport component from the lower layer in the Belt Sea into the Baltic Sea complicates the picture [e.g., Stigebrandt, 1983]. The inflowing waters attain high salinities only when inflow rates are large and persistent and preceded or accompanied by strong wind- driven vertical mixing in Kattegat and the Belt Sea, raising the

8895

8896 LILJEBLADH AND STIGEBRANDT: DEEPWATER FLOW INTO THE BALTIC SEA

I

12oE 16 ø

Kattegat'b,

SWEDEN

n

Oresund

Sea ('/•

BornhoLm

Figure la. Map over the Arkona Basin.

surface salinity in these seas. Owing to the small volume of the surface layer in the Oresund, inflows taking this route may rather quickly attain the salinity of the Kattegat surface layer. Major persistent inflows of high-salinity water, accompanied by almost continuous rises of the sea level in the Baltic by 0.5 m or more, occur irregularly and seldom, usually with intervals of several years. Such extreme events, which contribute to the high end of the distribution of inflows with respect to salinity, are important for the renewal of the deepest deep water in the Baltic proper. For a statistical analysis of major Baltic inflows, see Matthiius and Franck [1992].

Because of great short-term variability of flow rates and salinities, it is hard to assess the distribution of inflows to the Baltic Sea among different salinities from measurements in the Danish Sounds and 0resun& Petr•n and Walin [1976] therefore measured the inflow distribution in a vertical section (near section F, Figure lb) across the Bornholm Channel, where the dense inflow should be smoothed and rectified. They obtained detailed observations of the dense bottom current. The trans-

port measured on about 30 occasions showed great variability, but during most, some transport of dense water was found [Walin, 1981].

This led to the idea that the dense water entering through Oresund and Fehmarn Belt does not flow all the way to the Bornholm Channel in a coherent bottom current but is first

stored in a pool at the bottom of the Arkona Basin. This scenario was assumed by Stigebrandt [1987a], who also assumed that the flow of dense bottom water from the pool and farther into the Bornholm Channel is hydraulically controlled

in a rotational baroclinic sense by the pool itself, i.e., by the vertical stratification in the Arkona Sea. A standard formula

for the baroclinic geostrophic flow as a function of the stratification was applied and tuned. Using this formula, the inflow of dense water to the Baltic Sea may be computed from representative hydrographic observations in the Arkona Sea. Thus extensive current measurements in vertical cross sections may be replaced by deep conductivity-temperature-depth (CTD) observations from only one representative hydrographical station. Seemingly realistic statistics of flows into the Bornholm Channel, computed from 182 historical vertical density profiles obtained at the international hydrographic station BY1 in the Arkona Sea, were presented by Stigebrandt [1987a]. Recently, this approach was applied by K6uts and Omstedt [1993], who obtained quite similar statistics using 278 historical vertical profiles obtained at the hydrographical station BY2, located closer to the Bornholm Channel.

The inflow statistics, as computed from the stratification at BY1, were used in a model for the long-term vertical circulation of the Baltic proper [Stigebrandt, 1987b]. The model re- produces the main observed large-scale properties of the time- dependent, vertical stratification in the whole water column of the Baltic proper. This can be achieved only if the inflow is well described with respect to the flow distributions in the time and salinity domains, respectively. In spite of this, it may be hard to gain wide acceptance for the idea of a self-controlled rotational baroclinic flow of dense water in the Arkona Sea before simul-

taneous observations showing the expected structure of the density and flow fields have been presented.

A field study of the anatomy of the Arkona dense pool and the relationship between the outflow and the vertical stratification started in 1988, when CTD casts were obtained along some transects across the Arkona Sea. From these we found

evidence for the anticipated baroclinic bottom current along the northern boundary of the dense pool (Figure 2). However, currents were not measured. After having equipped our research vessel Svanic with a high-resolution acoustic Doppler current profiler (ADCP), we made two cruises. During the first one, in January 1991, there was no dense pool in the Arkona Sea, in spite of a preceding major inflow from Kattegat. The reason for this was that a severe storm a few days before our cruise had thoroughly mixed the whole water column from the sea surface to the bottom. During the second cruise, in Feb- ruary 1993, we were lucky to find a well-developed pool of dense water.

The aim of this paper is to present our measurements of the dense bottom pool in the Arkona Sea in February 1993 and compare our empirical results with geostrophic calculations. The outline of the paper is as follows. In the next section we present the general oceanographic conditions and ideas about the flow in the Arkona Sea. This is followed by a presentation of experimental results. The paper is concluded by a discussion and some remarks. In the appendix we present the precondi- tions for the field experiment in February 1993.

2. General Conditions in the Arkona Sea

The Arkona Basin is the first in a series of basins encoun-

tered by dense water entering the Baltic Sea. A deep plain, with water depths in the range 40-50 m, occupies about half of the basin. In the east there is a transition from the deep plain to the Bornholm Channel. The latter leads to the next basin, the Bornholm Basin. The maximum depth of the transition to

LILJEBLADH AND STIGEBRANDT: DEEPWATER FLOW INTO THE BALTIC SEA 8897

I•resund I•.!.. ":%.! .

......... ß ;i;'"'!!;"'"'J.. •DROfiDE :.!' ""'.".: .:.5" .... " '"•...:

Ystad , • ?..'.;': ......••... '...'..•..',:. :,'•/•; BornhoLm ß /.'.:..•....• ....... K u [ [ a '.•,...' .,.....,.'• • '"x 9.. '•" '"?"'"•.u.p.d.-'"••.r-,-----• "'• •.";'Channel•,

-_ :,..,--•, , • •,., .,•'

2o, m 3.•m- '"'"',•,, ',?--, • • Xt; d'g?•d..•'"'• ?f!i BornhoLm • \;..j .... •)'"'•"..? /-",.,'"' • • •/•. D'x•,' • •.'..."' •..R nne \ :' ............. "x '"' --" ,.,' '" t ""' :

.,• M,.•, / ./K½iegers.•"..{ . \ B\ X[- •,.t" ..... "2' r,l'••"• :""' r-•rt •...•._- :•; © • ¾ • ^\ - \ •,, X ,•o • o.•"'•--• •' ....

....' L•

::.:, 5 ............... • .... ".. I,,,., ..... '... I • ..."'

• _.....,• '" ........ . ,,._ .,"'"' --_, .,..,. • ... ........ /__. grund ½ • N

"'-•. ............. '- ..' 20 Fehmar .......

..... ''' "-'"'• :•/ "z. om N•ut•c• Figure lb. Map with sections A-F described in the text. The dashed line ending at Svartgrund is the old section (Figure 2) from 1988.

the Bornholm Channel appears to be about 45 m. From available sea charts based on quite a few depth soundings, it is ditticult to establish both the depth of the deepest connection (the sill depth) and the minimum width of the transition at different depths. Nevertheless, minimum width of the transition with water depths greater than 40 m appears to be about 20 km. In the southeast there is a subsurface ridge between the Bornholm and R/igen Islands with a sill depth of about 27 m (Figures la and lb).

A halocline separates Baltic surface water with salinity of about 8 psu from the dense waters forming a bottom pool. The Baltic surface water has practically free access to the Arkona Sea. In the central parts of the Arkona Sea the depth to the halocline usually varies in the interval 20 to 40 m. The salinity of the pool is in the range 10-25 psu. The volume and salinity of the pool vary in response to inflows and outflows, and the inflowing dense water entrains ambient water during the descent to the Arkona Sea bottom pool. According to rough estimates based on salt budgets, the volume flow thereby in-

SaLinity (psu)

Figure 2. The salinity distribution in January 1988 in a transect across the Arkona Basin (obtained along the dashed line in Figure lb).

creases by some 50-80% and the salinity decreases corre- spondingly [see Stigebrandt, 1987a; Kbuts and Omstedt, 1993].

Dynamics of the Dense Pool

It is assumed that a baroclinic geostrophic current, expected to drain the pool, should develop along the northern and eastern perimeter of the pool [Stigebrandt, 1987a]. For an ide- alized, steady, frictionless, and rotating two-layer system with a pool of dense water of thickness H on a horizontal bottom and a motionless less dense upper layer, the transport Q along the rim of the bottom pool can be estimated using the integrated Margueles equation

Q = g'H2/2f (1)

Heref is the Coriolis parameter, g' = g(P2 - P])/P2, g is the acceleration of gravity, and p• (P2) is the density of the light (dense) water. In (1) the transport is completely determined by the pool stratification, as described by H and g'. Thus the flow may be termed self-controlled. It should be noted that (1) also describes the frictionless transport through a wide (compared to the baroclinic Rossby radius) rotating channel if the upstream potential vorticity is f/H, implying that the upstream relative vorticity is zero [cf. Whitehead et al., 1974].

For an application of (1) to the real dense pool in the Arkona Sea the dynamical thickness H of the pool should be taken as H = hf - hp, where z = hp is the pycnocline depth in the pool, well away from the baroclinic boundary current, and z = hf is the depth where the pycnocline intersects the bottom, in this case at the northern and eastern boundary (see Figure 2). If the pycnocline level in the pool hp varies, one has to figure out how a representative value should be defined.

8898 LILJEBLADII AND STIGEBRANDT: DEEPWATER FLOW INTO THE BALTIC SEA

St•rs indicate the sections

ø

t

0 '-

_eveIs - 1.0 -&O -5.0 -7.0 -9.0

-11.0 -15.0

-1 -19.0 -21.0 -25.0 -25.0 -27.0 -29.0 -31.0 -33.0 -35.0 -3 7.0 -59.0 -• 1.o 43.0 -•5.0

Plate 1. (top) The anatomy of the dense pool at the % = 10 surface based on 55 stations taken during the survey. (bottom) The topography of the Arkona Basin gridded from a database with 5.3 x 9.3 km 2 resolution.

The apparent front depth hœ for a certain location in the pool may be found if a long enough hydrographic time series (as from BY1 and BY2) exists there. The flows calculated by (1) (with a certain front depth) from this data should satisfy salt conservation, assuming a long-term mean steady state in the Baltic Sea. Using observations from BY1 to estimate hp, p:, and/92, together with (1), this requirement gave an apparent front depth of 41 m in the Arkona Sea [Stigebrandt, 1987a]. This happens to be close to the observed depth of the bottom intersection of the front along the northern boundary, as described later in this paper. It should be noted that KDuts and Omstedt [1993], who used hydrographic data from station BY2, found an apparent front depth h;. equal to about 48 m which is clearly below the observed hf and also below the mean depth of this area. One explanation for the difference in estimates may be that station BY2 possibly is situated within the region of the boundary current. If this is the case, some potential energy of the pool has been transferred to kinetic energy in this

region, with a corresponding descent of isopycnal surfaces and the local hp.

One expects that (1) may underestimate the baroclinic transport of the boundary current if the mean water depth in the area occupied by the current is greater than the depth h;. at the bottom front. Bottom friction, on the other hand, should decrease the transport compared to the estimate given by (1) and, in addition, cause an ageostrophic transport close to the bottom toward the bottom front. Barotropic currents and wind- driven circulation would also influence both the horizontal

distribution and the amplitude of the transport of pool water. All these factors complicate an empirical verification of (1) using synoptic current and density measurements in vertical cross sections of the pool.

Superposed Wind-Driven Flows

Csanady [1982] shows that the wind-driven circulation in a stratified, elongated basin is described by coastal jets in the


2OO

150- .

•oo-

50-

o-

-5O

3O

25

2O

15

-10

5

45

Days from Jan. 1 199.3

Figure 3a. Salinity (s) and accumulated inflow (v) through Great Belt and Oresund in the period January 1 to February 14, 1993.

wind direction and countercurrents in the central region. If this general result is applied to the Arkona Sea, one expects that westerly winds create an opposing (westward) return flow in the central parts, i.e., against the eastward, density-driven flow toward the Bornholm Channel. Easterly winds, on the other hand, would reinforce the density-driven outflow. Thus one expects that the outflow from the Arkona bottom pool may be modulated by local east-west winds. Model computations by Krauss and Brtigge [1991] showed wind-forced modulations of the deepwater flow in the Baltic Sea east of the Bornholm Island. However, wind-generated barotropic oscillations of the Baltic Sea induce large inflows and outflows of surface water through the Bornholm Strait and over Adlergrund between Bornholm and Rtigen. As can be seen in Figure 3b (Klag- shamn, dashed line), the sea level changes about 50 cm during

the survey, giving large differences in the net flow through the sections. We try to deal with this in section 4.

The general response of a stratified, elongated basin to wind forcing described above was largely verified by Lass and Talpsepp [1993] for the western part of the southern coastal boundary of the Arkona Sea. Their observations of stratification and currents showed that the offshore scale of the coastal

boundary exceeded the baroclinic Rossby radius by at least a factor of 2. To first approximation, the observed coastal jet may be regarded as barotropic. The coastal jet was found to be asymmetric, with eastward flow stronger than westward flow. Lass and Talpsepp explained that this was due to an eastward pressure gradient force along this coast, which was caused by highly saline bottom water at the Darss Sill in connection with westerly winds. Thus they found that besides the expected wind

lOO

50-

-5o

..... Klagshamn ½ I Stockholm

t .• • ---Viken

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

5 10 15 20 25 30 35 40 45

Days from Jan. 1 1993

Figure 3b. Recorded sea levels in Klagshamn, Stockholm, and Viken in the period January 4 to February 14, 1993.


Velocity projected ot 75 Degrees ond Sigmotheto 09:18- 16:40

-15

-20'

-25- -3O

-35

-4O

-45

-10"

-25

-30

-4O

-45

(m) 10000 20000

i'-1

UcdcI•- Ustice

(m) 10000 20000 50000 4OO00

Ugeost•o:pA!j

5O000

5O000

60OOO 7OOO0

7OOOO

-15

-20-J -25

-30

-35

-40

-45

(m) 10000 20000 30000 40000 50000 60000 70000

Plate 2. Cross-sectional currents in section A, (top) measured currents, (middle) measured baroclinic currents, and (bottom) geostrophic currents, calculated from the conductivity-temperature-depth (CTD) profiles with zero velocity in the surface. The section was taken February 6 from 0918 to 1640 LT. It starts at Kullagrund and ends in shallow waters northeast of Rfigen. The measurements consist of 16 CTD profiles and 106 velocity profiles. Plate 2 (top) shows a barotropic eddy-like structure with westward flow along the Swedish coast and eastward flow in the deepest portion of the basin. To the south the flows are weaker and more variable Apart from the horizontal velocity shear and a very weak horizontal density gradient, the upper layer is vertically homogeneous. Comparing Plate 2 (middle and bottom), it can be seen that the calculated and measured baroclinic velocity fields are rather similar.


-10-

-25

-5O

-55

-45

-15

-2O

-25

-35-

-4O

-45

-20-

-30

-35

-4O

--45

Velocity projected et 75 Degrees end S•grnntheto •8:12 - 22:40

(m) lOOO0 2O0OO 5OO0O 4OOOO 50OOO 6O000

2O00O

r•Scp- r•sJ•ce

40OOO 5OO00 60000

(m) 10000 20000 ,30000 40000 50000 60000

Plate 3. Same as Plate 2, but for section B, obtained February 6 between 1812 and 2240 LT, starting close to Adlergrund and ending south of Ystad. The measurements consist of eight CDT profiles and 89 velocity profiles. This section has somewhat weaker westerly flow on the Swedish side than the previous A section, but large horizontal shears still exist. The eastward going surface water in the central parts of the basin has now moved somewhat up on the southern slope. The dense bottom water has an eastward flowing boundary current on the northern slope and a weak one on the southern slope (Plate 3, top). The halocline is sharpest above the boundary currents and especially at the edge of the northern perimeter current, where bottom stress apparently homogenizes the water. The thickest homogeneous bottom layer is seen in the deepest part of the section(s).


Table 1. Summary of the Flows Calculated in the Different Ways Described in Text

ADCP

CTD #'(hf- hp)2/2f, hp, Ro, Sect ion Q net Q east Q, ( ADCP ) Q, (CTD) m 3/s m Acr 0 km

A -36,700 28,400 135,600 119,800 102,000 24 8.5 8.9 B -82,000 43,900 116,500 84,100 70,000 27 7.5 8 C 31,000 73,400 112,600 72,400 110,000 30 8 8.5 D 30,700 47,300 61,100 72,400 85,000 31 7.5 8 E -75,900 30,000 86,100 90,600 ...* ......... F -23,900 22,100 42,500 64,300 '"* .........

All Q values are in cubic meters per second. Qnet is the net flow shown in Plates 2-7 (top). Qi is east transport of water, while Qi is measured baroclinic transport, both of which are accumulated from the highest densities (up to) the pycnocline (or 0 < 10). Variables used in the two-layer calculations (from the deepest stations in the sections) are !7, acceleration of gravity; hf, front depth (based on the findings of Stigebrandt [1987a], where hf = 41 m); hp, depth of the pool; f, Coriolis parameter; and Ro, Rossby radius, calculated here using local depth.

*The flows vary widely, depending on the location of the chosen stations. For the deepest stations in these sections the flows are in the range 80-115,000 m3/s.

forcing of the coastal jet, there is additional forcing along the German coast associated with density perturbations in the region of the Darss Sill.

3. The Field Experiment The ultimate reason for our survey at this particular time

was to find a well-filled dense pool due to a major inflow event that occurred in January 1993. Strong and persistent westerly winds raised the sea level in Kattegat (see the record from Viken in Figure 3b). This caused a strong barotropic inflow (---300 km 3 in 3 weeks) to the Baltic Sea, where the sea level rose by 0.8 m in the period January 6 to January 27 (see the records from Klagshamn and Stockholm in Figure 3b). Ac- cording to the measurements in the Great Belt and Oresund, the salinity of much of the inflowing water was in the range 22-27 psu, with the highest salinity in Oresund (Figure 3a). After the inflow followed a period, outlasting our survey, with generally falling sea level and rather strong outflow from the Baltic Sea. However, in 3 days preceding the survey, i.e., Feb- ruary 3-5, the wind was westerly, with speeds around 10 m/s. According to Figure 3b, there was inflow through the Belts and Oresund on these days (about 100,000 m3/s). During the survey the wind was variable, blowing from the sector from SSW over west to north and with speeds in the range 4-7 m/s, and there was outflow from the Baltic Sea (about 100,000 m3/s).

A tentative volume budget for the 1993 major inflow is presented in the appendix. This suggests that during the period January 6 to February 6, 1993, the inflow of dense water to the Baltic Sea from the Arkona pool was in the interval 56,000- 75,000 m3/s, with water of mean salinity in the range 17-20 psu.

To investigate the dynamics of the dense pool, a survey was undertaken with R/V Svanic from February 6 to 8, 1993. Mea- surements of salinity, temperature, and current were obtained along six sections. The area of interest and the locations of the six sections are shown in Figure lb.

For salinity and temperature measurements we used a CTD (Neil Brown, marklII), and for the current measurements we used a vessel-mounted ADCP (RD Instruments). A total of 60 CTD stations were taken during the cruise. The density Po (kilograms per cubic meter) is computed from in situ values of temperature, salinity, and pressure. Here o- 0 is defined by o- 0 = Po - 1000. According to water samples measured with a Guild- line AUTOSAL, the CTD had an accuracy of <0.006 mmho/cm for conductivity and 0.004øC for temperature. The ADCP is a standard 600-kHz instrument but equipped with a

20 ø transducer which reduces the sidelobe interference above

the bottom by 50% compared with the 30 ø standard transducer. With a proper blanking of postping transducer ringing, this means for the Arkona Sea (45 m depth) that good data are, in general, available from about 6-7 m depth and down to 2-4 m above the bottom. Of course, physical factors like ship rolling may occasionally increase these shadow zones. It should be noted that close to the bottom, the ADCP may give a velocity component in the direction of the ship heading caused by influence of bottom reflections. The cross component will thus tend to zero, and it may look like the expected effect of bottom friction. The validity of data points close to the bottom may be examined by the use of quality information like echo intensity. Owing to differing sea states during the survey, the erroneous bottom zones differ somewhat between the sections, and the fact that the velocity tends to zero at the bottom cannot exclu- sively be regarded as due to bottom friction. However, a ten- dency for the velocity vector to twist to the left can be found in the bin closest to the bottom on CTD stations where ship drift is small.

The ADCP was configured with 2-m depth cells, and ensem- bles were obtained every third minute with maximum ping rate (2-Hz with bottom track enabled), giving a horizontal distance between the velocity profiles of 500 to 1500 m, depending on ship speed. The theoretical ensemble standard deviation is then around the precision limit of the instrument (1 cm/s), although the variance may increase due to "environmental factors." An integration in time of the bottom track velocity along the sections gives a decreased section length of 1-2% in comparison with navigation (DGPS). A small misalignment of the transducer gives very small errors in horizontal velocity when good bottom track is achieved.

4. Experimental Results

For the analysis of our current measurements we assume that the measured flow field is composed of steady geostrophic barotropic and baroclinic flows, and in a standard way we split the velocity components,

1 0 f(v, + v,) = (P, + P,) (2)

ß Po Ox

where v is the horizontal velocity perpendicular to the x axis, Po is a reference density, P is pressure, and subscripts s and i denote the surface (barotropic) and internal (baroclinic) components respectively. Ps and Pi are defined by


-2O

-3O

-35

.-45

-2O

-35

45

Velocity projected • 70 Degrees and Sigmatheta .

, .

ß

.

0•.•)0- 1350

(m)

I 1ooo0 20o00 3•x)o 400OO 5O000

rJc•dc•- /Js/,•ce ß

•- ':,•i -

- ; •,•.. .

,

,

•i":- . .••"•/•., , .. - - .::. h.: ........ ;, ••. . %•:,.• .... "' '. 5 • .., ' . .... ':':" Z! •J,v3':,, • . ' ....

--- • .•_ ••' • • • .. - ....... :........•

I

Ugeostroph•

ß "-•- 5 '

.

.

- .. .

. ½'• .

.. .(:• •'... $-'•

,

..._:_..

• 2• • 4•

30• i •s..o I 2e.o I 24,0 I 2zo I 20.0 I

•.o I •4.o I •zo I

I !

!

-•2.o I -• '1 -•o .! -m.o !

-24.o I

Plate 4. Same as Plate 2, but for section C, obtained February 7 between 0900 and 1350 LT. It is quite close to section B. The measurements consist of 10 CTD profiles and 113 velocity profiles (the very shallow ones are excluded). In the surface waters there are weak westerly transports both at the northern and southern flanks (Plate 4, top). The east going surface flow is here rather weak and centered around the deepest portion of the section. The surface flow on the northern side is less vivid than in previous sections and does not penetrate as deep. Both the measured and the computed baroclinic flow fields have two boundary currents. The boundary currents are somewhat narrower and more unidirectional than in section B.

Ps(x) = gporl(x) (3)

o Pi(x, z) = g p(x, z) dz (4)

Here z is the vertical coordinate andz = r/(x) is the deviation of the surface from the geoid. For each section we present currents in three panels of each plate. In Plates 2-7, north is to the left and current velocities are positive when directed east-


ward. Plates 2-7 (top) show the measured fields of density and velocity (ADCP) perpendicular to each section. Plates 2-7 (middle) show the measured baroclinic flows, as described below. Using the measured density field p(x, z), the baroclinic velocity field is computed by discrete integration of the baroclinic part of (2). The results are presented in Plates 2-7 (bottom). According to the definition above, these velocity fields vanish at the sea surface. If only barotropic and baroclinic currents were present, then the measured surface current should be the barotropic one. The baroclinic current field may then be obtained by subtraction of the surface current from the measured currents in the whole water column. How-

ever, the ship-mounted ADCP does not measure close to the surface and there may also be horizontal velocities balanced by stress and/or inertia (steady and unsteady Ekman currents). These should be strongest close to the sea surface and decline downward with a possible increase close to the bottom. For this reason the assumption about steady geostrophic flows should hold better in the interior of the water column. To separate the barotropic current from the measured current, we have done as follows. Mean velocities in the measured field are calculated

in a 2-m-thick slice centered at 30 m depth. To obtain the barotropic velocity, we thus subtract the baroclinic velocity (as determined from the density field) from the measured velocity in the slice. Finally, the barotropic velocity is subtracted from the whole measured field. The resultant velocity field, which we call the measured baroclinic, is presented in Plates 2-7 (middle).

A dynamically equivalent, two-layer stratification can be computed using the method described by Stigebrandt [1987a]. Then, using (1), the transport in the boundary current can be calculated. It should be noted that both methods give identical results for a particular station if the pycnocline intersects the bottom at the reference depth hœ. However, the transport calculated in this way from a particular station (preferably outside the boundary currents) shall be compared with the "measured baroclinic" transport in order to validate the gen- erality of (1). The results are listed in Table 1.

Summary Discussion of the Observations

During our survey the upper part of the halocline was at about 20 m depth in the central Arkona Sea. However, it deepened toward the northern and eastern perimeter, where it intersected the bottom at water depths of about 40 m (see Plates 2-7). In Plate 1 (top) we have plotted the anatomy of the dense pool for a certain density surface (o- o = 10 based on 55 stations, denoted by stars). It can be seen that the pool is sloping toward section D, where the flow is most narrow and then climbs Bornholm as the flow bends northward. The gray thick isoline shows where the pool intersects (hf) the bottom. The very high pool level westward of section A in Plate 1 is, of course, unverified and probably an artifact of the extrapolation routine. In the right corner the slope should, instead, go down and then up, but there is a lack of data here. Plate 1 (bottom) shows the topography of the basin (based on mean depths calculated in squares of 5.3 x 9.3 km). The dense water in the pool was strongly stratified, with the weakest stratification in the western and central parts. For a two-layer approximation of the pool stratification the density difference is about 9 kg/ m 3, giving an internal wave speed of Ci '-• 1 m/s and an internal Rossby radius of Ci/f--- 8 km.

Net Flows Through the Sections

In all sections there is a net westward flow in the layers of low salinity. The westward flow is relatively weak and uniform in the Bornholm Channel. In the Arkona Sea the westward

flow is quite strong along the Swedish coast and, to some degree, also along the German coast. In the middle of sections A-D the surface flow is directed eastward. With winds essen-

tially from west before and during the survey one would expect, as discussed in section 2, a reversed flow system compared with what we actually observed; that is, one expects eastward flow along the coasts and westward flow in the middle of the sections. The departure from the purely wind-forced theoretical circulation may be due to the sea level oscillations of the Baltic and the westward throughflow of the Arkona Sea during the survey, caused by outflow through the Belt Sea and •resund. Our observations may suggest that the outflow to Kattegat is mainly drawn from the coastal boundary layers in the Arkona Sea. Accordingly, the flow is quite variable during our survey. The net transport through the sections varies from 82,000 m3/s westward to 31,000 m3/s eastward (Qnet in Table 1). The mean flow through the Belt Sea and {3resund during our cruise should cause a net westward flow through the Arkona Sea of '-'100,000 m3/s. However, since the areas of the measured sections vary widely and cover neither the uppermost 7 m of the water column nor the shallow coastal regions, we do not expect the integrated transports through these sections to equal the barotropic flow through the Belt Sea and {3resund.

Baroclinic Flows

The measured baroclinic transport Qi(ADCP) (obtained after subtraction of the measured barotropic flow, as explained in the beginning of this section) is eastward for the dense water and varies between sections B-F in the range of 60,000 to 120,000 m3/s, with an average of ---80,000 m3/s. For the same sections the baroclinic transport Qi(CTD), computed from the density field, varies in the interval of 70,000 to 100,000 m3/s, with an average of about 80,000 m3/s. Thus the average measured and computed baroclinic transports are rather close to each other and are equal to about 80,000 m3/s during our survey. This is also close to the upper range of estimates, mentioned earlier, derived from a simple volume budget of the pool for the period January 6 to February 6, 1993. Equation (1) gives transports based on the pool stratification in the A and B sections and hf = 41 m in the range of 70,000 to 100,000 m3/s. On the basis of the stratification in the C and D sections and

h•,. = 48 m, as suggested by Kouts and Omstedt [1993], one obtains transports of about 100,000 km3/s. All sections have a measured eastward transport of dense water (o¾ _> 10; see Qeast in Table 1), although this varies in the range of 22,000 to 74,000 m3/s, with an average of about 50,000 m3/s. The fact that measured flow of dense water differs from the baroclinic flow

is apparently, for the most part, due to the modulating effect of the barotropic flow components.

5. Discussion

Vorticity

The distribution of large-scale horizontal vorticity in the Arkona bottom pool is, so far, not calculated. If there is no relative vorticity in the interior of the pool, then the outflow from the pool will create anticyclonic relative vorticity concen- trated to the region of the northern baroclinic rim current.


Ur(x), In a slice Sigma(14-18) X(O) is about where the rim-current intersects the bottom

50 It i

ß H '• ..... / ,' 30-1' t ,•i• .......... ', / !

/' 14' "i'". ' ; I • ¾'.."3 • ,,"•. ! •.,!!! •-, , t ..... !

s . : ! : i .. ............ . , ,. ; .... ... 2 • *' • ' ' ' O• I : • .• • i • '.. . t : '. I:'\", ' • '. '"" .'

l/ \"i I/', ', ..... ;'.. ..'

": '"' .... ' i ,' '.' ,• •s --A

-10 • • I 0 (m) 20000 40000 150000

Figure 4. The horizontal distribution of east-west currents in the density interval 14 _< tr 0 -< 18 for the different sections. Currents toward east are positive. As a reference for the horizontal velocity shear, we have drawn a straight, vertical line with du/dx = -f (i.e., u = -fx), and x is the distance from the front where the rim current intersects the bottom.

However, the filling of the pool may induce cyclonal vorticity since the inflowing dense bottom currents, from the Fehmarn Belt and 0resund, should have an eastward velocity component when flowing in along the southeastern boundary of the pool.

The horizontal velocity shear in the dense pool can be seen from Figure 4, where we have drawn for each section the measured baroclinic velocity u averaged over the density interval (14 _< tr 0 _< 18). Here u is the velocity component perpendicular to the section. The horizontal axis is along the section and starts at the bottom front. The baroclinic horizon-

tal shear is greatest in the northern boundary current, but its absolute value is everywhere appreciably smaller than f. A weak velocity maximum on the southern side is also evident in sections C and D (Figure 4).

Friction

The density structure is closest to being two-layered on the northern side of the pool, where the pycnocline is very sharp, and hits the bottom at depths of about 40 m. The thickness of the well-mixed bottom layer ranges from 1 to 10 m (greatest thicknesses are found in the deepest part of the sections outside the velocity maximums) but is typically 5 m, and velocities are usually in the range 0.2-0.5 m/s. Thus we conclude that bottom friction does not appreciably violate the geostrophic assumption. The wind-mixed surface layer is usually homogeneous down to 15-20 m depth. The thickness of the zone of stratified water between the upper and lower well-mixed layers generally decreases from south to north. Maybe this is due to interleaving of inflowing brackish water in the south during the inflow event the days before our cruise. The stratified layer is also increasing in thickness going downstream from section A, especially in sections D to F. This stratification may possibly be explained as an advected reminiscence of interleaving of inflowing water of different densities or, possibly, as a result of

mixing produced by interfacial Ekman layers [Johnson and Ohlsen, 1994].

Variability

Several factors may introduce variability in the flow of deep water through the Arkona Sea. One important factor is the large-scale variability of the barotropic flow, forced by the sea level difference between Kattegatt and the Baltic Sea. During inflow to the Baltic Sea this flow also implies filling of the deepwater pool in the Arkona Sea. The variability due to the filling-emptying mode should be much smoother than the barotropic flow through the belts and Oresund, primarily because of the relatively long timescale for emptying the pool (a few weeks). Our cruise took place shortly after a major inflow event, and the deepwater flow was therefore anomalously large, about 4-5 times larger than the mean (see section 1). We observed a decreasing baroclinic flow and a decreasing pool height with increasing distance from the belts and Ore- sund. It seems probable that the filling of the pool the days right before our cruise may be of importance. Another explanation may be that there is a baroclinic anticyclonic circulation within the pool drained by the boundary currents entering the Bornholm Strait.

A varying local wind over the Arkona Sea is another factor that may impose variability in the deepwater flow. The local wind causes both advection, that, possibly through barotropic currents, moves the deepwater within the sea and diapycnal mixing and entrainment that may withdraw deep water to the surface layer. We have already mentioned an event (winter 1990-1991) when gale winds thoroughly mixed the waters in the Arkona Sea from the sea surface to the bottom, whereby the pool of deep water became practically exhausted. In the preceding section we showed that barotropic components of the flow may be removed from the measurements and the


-15-

-20 -

-25-

-30 -

-35

-4O

-45

-20-

-25-

-30-

-35

-4O

-45

-10

-2O

-25

-35

-41

-45

Velocity projected at 75 Degrees end Sigmatheta 14:40 - 16:15

(m)

(m)

5O0O 1O00O 15OOO 2O000

•(•d;c;p- rJs[•,ce

5000 10000 15000

rJgeostroph, lj

200OO

I 24.0 I 22.0 12o.o I I I •4.o I •2 -ø

'• 2.0

! -2.o , -4.0

-•.0

-lO.O

-14.0

-

-2zo -24.0

-26.0 -•.0 -•0.0

(rn) 5000 10000 15000 20000

Plate 5. Same as Plate 2, but for section D from the inlet to the Bornholm Channel, obtained February 7 between 1440 and 1615 LT, and located just to the west of the narrowest section at the entrance to the Bornholm Channel. The measurements consist of five CTD profiles and 34 velocity profiles. There are also in this section rather weak westerly currents at the sides and easterly currents in the middle of the surface layers (Plate 5, top). The two boundary currents visible in section C seem to have merged. The rather steep southern bottom slope is quite close to the right but outside the plot.

resulting baroclinic flow was found to be astonishingly stable. This may indicate that the response to a varying local wind is mainly barotropic.

One may also imagine that the baroclinic rim current may become dynamically unstable which probably should show up

in meandering and eddy production, but so far, we have seen no indications for such instabilities. It is, however, a little premature to draw conclusions regarding the variability of the deepwater flow in the Arkona Sea based upon our measurements. To investigate the variability, one should repeat the


sections a number of times and/or use stationary recording instruments to obtain time series. We hope to do such measurements in the future.

Upstream Influence

It is well known that bottom currents running in channels with subcritical longitudinal slopes may influence the upstream dense pool [e.g., Turner, 1973]. For a stationary case it is obvious that the transport of dense fluid entering a channel cannot be greater than the transport capacity of the channel. If during a certain period the transport toward the channel is greater than the transport capacity of the channel, then there will be feedback (upstream influence) upon the pool, leading to adjustment of the transport toward the channel such that this flow matches the transport capacity of the channel. The flow in the boundary current of the pool has to change through adjustments in H, i.e., through adjustments of h•,. and/or h p. Adjustments are probably most easily obtained by changes in hœ. The reason for this is that changes in hœ do not involve large volume flows; a relatively modest frontal movement up or down the sloping bottom is sufficient to adjust hœ. Dynamically, such an adjustment would occur through internal boundary waves, which are excited at the pool-channel junction and propagate against the pool boundary current. This requires that the speed of the boundary current is less than the speed of the waves, a condition that seems to be fulfilled in the Arkona Sea, where the fastest internal waves should have phase speeds around 1 m/s and the boundary current has only half of that speed. Adjustments in hp, however, involve large volume flows and probably only occur on longer timescales by changes in the balance between inflows and outflows. Thus it seems likely that adjustments of the outflow from the pool, possibly needed to match the baroclinic transport capacity of the channel, should take place through changes of the bottom intersection depth hf of the front. If the transport capacity of the channel controls the flow from the pool (as suggested by Gidhagen and H•kans- son [1992]), then we thus expect that the depth of the pool boundary hf should vary with the transport through the channel. The fact that we only have observed hœ --- 40-41 m under different flow conditions may be evidence (albeit weak because the number of observations is still rather small) against the hypothesis that there is upstream influence on the Arkona pool of dense water from the flow in the Bornholm Channel. Thus

the latter seems to have greater transport capacity than the baroclinic boundary current in the Arkona dense pool.

6. Concluding Remarks Our measurements demonstrate that the instantaneous cur-

rent field in the Arkona Sea may be quite complicated due to effects of wind-forced circulation, throughflow, and the dynamics of the dense bottom pool. However, after subtraction of the barotropic flow component from the measured currents, the resulting flow structure looks quite similar to the baroclinic flow computed from the density field. Thus, although the barotropic flow has quite strong horizontal shear, it is possible to separate barotropic and baroclinic velocity components since there are current measurements from the whole water column

with sufficiently high accuracy and spatial resolution. This was not feasible before the introduction of the ADCP.

Observations from the cruise presented in this paper and from earlier cruises show that the pool front in the northern Arkona Sea persistently is found to be located at water depths

of about 40 m. We have no good explanation for this. However, we believe that it is an indication of the absence of upstream control from the flow through the Bornholm Channel since we expect that upstream influence under different flow conditions should manifest itself through a varying depth of the pool front.

The observations of the density and flow fields presented in this paper support the idea of a self-controlled rotational baroclinic flow of dense water in the Arkona Sea. Realistic ba-

roclinic transports may be computed from the density field alone. This lends strong support to the method to compute the deepwater flow into the Baltic proper using hydrographic data from BY1, as suggested by Stigebrandt [1987a].

Being located quite close to the inflow area, it may well be that BY1 is not the perfect location for obtaining hydrographic profiles for flow estimates. However, by necessity, it takes many years before the number of measurements from a mea- suring site is great enough for computations of reliable inflow statistics. Since there already are so many profiles from BY1, covering several decades in time, it would not be wise to dis- miss this station for another. However, it would be wise to increase the vertical resolution of the observations.

The observed baroclinic flow field shows that the absolute

value of the horizontal shear of this flow is rather small in the

interior of the pool. It is greater but still appreciably less than f in the baroclinic geostrophic boundary current along the northern flank of the pool.

The dynamical properties of the dense Arkona bottom pool certainly deserve to be studied more in detail, not least because this sea seems to be well suited for tests of models for rotating dense pools and channel flow. One of the interesting properties to investigate is if the location of the northern pool front always is at a depth of about 40 m and, if so, the dynamical reasons for it. We suspect that the topography of the transition area between the Arkona Basin and the Bornholm Channel

may be of importance. However, existing sea charts are not good enough for a detailed topographical description of this area. A detailed topographical mapping, with horizontal resolution a small fraction of the internal Rossby radius, is badly needed. We are also curious about the upstream (western) coupling between the rim current and the dense pool, i.e., how the boundary current is fed by the pool, which leads to the question about the eastward decrease of the pool height being possibly due to an internal anticyclonic circulation. Finally, we would like to know, in detail, how the pool adjusts to filling and emptying, including the question of possible upstream influence upon the pool from the flow in the Bornholm Channel.

Appendix: A tentative volume budget for the 1993 inflow

In the period January 6-27, 1993, the sea level in the Baltic rose by 0.8 m (Figure 3b). Since the horizontal surface area of the Baltic Sea is about 370,000 km 2, a sea level rise in the Baltic Sea of this magnitude implies a volume increase of almost 300 km 3. The Danish Hydraulic Institute has permanent stations for hydrographic and current measurements on each side of Sprog6 in the Great Belt. The accumulated volume of inflowing water is calculated from ADCP measurements according to M6ller and Pedersen [1993]. In 0resund (between Sj•elland and Sweden), permanent measurements are made by the Swedish Meteorological and Hydrological Institute at a site at Drogden. These show that the inflows through the Great Belt and C}re-


-20

-25

-40

-45

-15

-2O

-25

-30

-35

-40

-45

-20

-25

-35

-40

-45

Velocity projected ot 40 Degrees end Signmtheta 0g._95 - 12:10

5OO0 •oooo •sooo 2000o 2sooo

Uadcp- Ustice

(r•) 50oo •oooo 15O00

UgeostroI•hlt

2000O 2500O

cm, 30.0 28.0 26.0 24.0 22.0 20.0 18.0 16.0 14.0 12.0 10.0

8.0 4.0 2.0 0.0

-2.0 -4.0

-8.0

-10.0

-•2.o -•4.o

/-•6.o

-•o.o --•zo -•.o -•.o

-30.0

(m) 5OOO 10000 t5000 2O0OO 25O00

Plate 6, Same as Plate 2, but for section E, obtained February $ between 0925 and ]210 LT, which goes from R0nne (Bornholm) to Svartgrund. The measurements consist of eight CTD profiles and 52 velocity profiles. The horizontal shear in the upper layer is now almost absent, but there is a rather strong westward transport in the surface layers (Plate 6, top). Again, there is rather good agreement between the measured and computed baroclinic velocity fields (Plate 6, middle and bottom). It may be noted in all pictures that dense water in the northern boundary current seems to be drawn from the middle of the pool. This must be provided by the action of bottom friction.

sund were 200 and 70 km 3, respectively (Figure 3a). During the same period there should have been an additional inflow through Little Belt (between Fyn and Jutland) of about 20 km -• (not measured) and an unknown (probably about 15 km 3) net supply of freshwater to the Baltic Sea by runoff and precipita-

tion minus evaporation. The measurements in the Great Belt and Oresund show that much of the inflowing water had a salinity of about 22-27 psu (Figure 3a). There are large volumes of water of usually low salinity stored in the Great Belt. This is clearly demonstrated in Figure 3a, which shows that the


-50-

-45

-2O

-25

-3O

-35-

-45

--15

-2O

-25-

-30 -

-55

-4O

-45

Velocity projected at 10 Degrees and Sigmatheta 16'25 - 19.00

(r,',) 10oo0 15000

rJ•d,c•p -

2OO0O 25OOO

sooo oooo =oooo

29.5 27.0 24.5 22.0 19.5 17.0 14.,5 12.0

9.5 7.0 4.5 2.0

-0.5 -3.0 -5.5 -8.0

-10.5 -13.0 -15.5 -1E[O -20.5 -25.0 -25.5 -28.0

(m) 5000 10000 15000 20000 25000

Plate 7. Same as Plate 2, but for the last section, F, taken February 7 between 1625 and 1900 LT. It goes from Svartgrund to Bornholm somewhat downstream of section E. The measurements consist of seven CTD profiles and 52 velocity profiles (the density profile at the southern slope was not measured). The flow looks similar to that in section E but with weaker westward transport in the surface layers (Plate 7, top). The flow has two topographically separated branches, a feature always present also in the measurements by Petr•n and Walin [1976]. Between the well-mixed bottom layer (2-7 m thick) and the surface layer there is a linearly stratified zone (which is generally thicker on the southern side), possibly suggesting interfacial mixing with increasing influence downstream (i.e., downstream of section A). The measured and computed baroclinic velocities again look quite similar, with two branches of strong currents (Plate 7, middle and bottom).

salinity in the Great Belt responds only slowly to large, unidirectional volume flows. Thus it should be noted that the water

entering the Arkona Sea during the first part of a major inflow event may be quite fresh (see Figure 3a). Some of the water

entering at the end of an inflow event, which typically is about 30 km 3 according to model computations by Stigebrandt [ 1983], may be swept back to the Belt Sea when the flow changes direction. Figure 3a also shows that the salinity in Oresund, as


Fehmarn Belt Arkona Basin •resund

220 •1 •30 S - 8-10 20 • .... '" ........... '"':'""':'"" '•x//, 80 100 ,.•;-/•.:" :!!L.•.. 2•0 V-80 .•:::'

Votume budget for dense water 930106-930206

Figure A1. Volume budget for the Arkona dense pool for the period January 6 to February 6, 1993. Volumes are expressed in cubic kilometers.

expected, responds faster to changes in the flow direction. The volume sloshing back and forth is about 10 km 3. Taking the effect of these buffer volumes into account, we estimate that the net flow of water of salinity 22-27 psu to the Arkona Sea during this inflow event was about 200 km 3. Owing to entrainment of ambient water, the volume of the dense water actually delivered to the Arkona bottom pool might be 50-80% greater. Thus some 300-350 km 3 of water of mean salinity 20-17 psu was probably supplied to the Arkona Sea bottom pool during the inflow event.

A volume budget for the major inflow event this year (1993) is presented in Figure A1. The volume of water in the pool of salinity greater than 10 psu was about 80 km 3 during our cruise. Thus about a quarter of the volume of dense water supplied during the inflow event was apparently still left in the pool during our survey. Accordingly, the loss of dense water from the pool during the period January 6 to February 6 should have been at least 220-270 km 3. Actually, the loss must have been even greater since there certainly was some water in the dense pool at the beginning of this period. Let us assume that the initial pool volume was 30 km 3. The loss from the pool then is 250-300 km 3. However, some of the lost water was certainly withdrawn from the pool by entrainment into the Arkona Sea surface layer due to strong winds in the period. Evidence for this may be found from the salinity record from Oresund, which shows that the outflowing water had a salinity of about 11 psu during the outflow in the beginning of February. Since the normal salinity of the Arkona surface water is 8-9 psu and the volume is ---300 km 3, we estimate that some 100 km 3 of water of salinity 18 psu was removed from the dense pool by wind-driven entrainment, an estimate which admittedly is quite uncertain. Thus we estimate that during the period January 6 to February 6, 1993, the inflow of dense water to the Baltic Sea from the Arkona pool was 150-200 km 3, with corresponding mean flow rates in the interval 56,000-75,000 m3/s.

Acknowledgments. This work was supported by the Swedish Nat- ural Science Research Council (NFR) and the Swedish Environmental Protection Board (SNV). The Danish Hydraulic Institute and the Swedish Meteorological and Hydrological Institute supplied data on the flow and salinity in the Great Belt and Oresund, respectively. The

latter institute also supplied data on sea level and meteorological conditions.

References

Csanady, G. T., Circulation in the Coastal Ocean, D. Riedel, Norwell, Mass., 1982.

Gidhagen, L., and B. Hfikansson, A model for the deep water flow into the Baltic Sea, Tellus Ser. A, 44A, 414-424, 1992.

Hfikansson, B., B. Broman, and H. Dahlin, The flow of salt and water in the sound during the Baltic major inflow event in January 1993, Publ. C.M. 1993/C:57(V), Int. Counc. for Explor. of the Sea, Copen- hagen, 1993.

Hela, I., Uber die schwankungen des wasserstandes in der Ostsee,Ann. Acad. Sci. Fenn., Ser. A, 28, 108 pp., 1944.

Jacobsen, T. S., Sea water exchange of the Baltic, measurements and methods, Belt Project report, 106 pp., Natl. Agency of Environ. Prot., Copenhagen, 1980.

Johnson, G., and D. Ohlsen, Frictionally modified rotating hydraulic channel exchange and ocean outflows, J. Phys. Oceanogr., 24, 66-78, 1994.

K6uts, T., and A. Omstedt, Deep water exchange in the Baltic proper, Tellus Ser. A, 45A, 311-324, 1993.

Krauss, W., and B. Brtigge, Wind-produced water exchange between the deep basins of the Baltic Sea, J. Phys. Oceanogr., 21, 373-384, 1991.

Lass, H. U., and L. Talpsepp, Observations of coastal jets in the Southern Baltic, Cont. Shelf Res., 13, 189-203, 1993.

Matthfius, W., and H. Franck, Characteristics of major Baltic inflows--A statistical analysis, Cont. Shelf Res., 12, 1375-1400, 1992.

M611er, S., and C. B. Pedersen, Analyse av hydrografiske data fra det sydlige Kattegat, Havforskning Milj6styrelsen 20, Minist. of Environ. and Energy, Copenhagen, 1993.

Omstedt, A., Water cooling in the entrance of the Baltic Sea, Tellus Ser. A, 39A, 254-265, 1987.

Petr•n, O., and G. Walin, Some observations of the deep water flow in the Bornholm Strait during the period time 1973-December 1974, Tellus, 28, 74-87, 1976.

Stigebrandt, A., Barotropic and baroclinic response of a semi-enclosed basin to barotropic forcing from the sea, in Fjord Oceanography, edited by H. J. Freeland, D. M. Farmer, and C. D. Levings, pp. 151-164, Plenum, New York, 1980.

Stigebrandt, A., A model for the exchange of water and salt between the Baltic and the Skagerrak, J. Phys. Oceanogr., 13, 411-427, 1983.

Stigebrandt, A., Analysis of an 89-year-long sea level record from the Kattegat with special reference to the barotropically driven water exchange between the Baltic and the sea, Tellus Ser. A, 36A, 401- 408, 1984.

Stigebrandt, A., Computations of the flow of dense water into the


Baltic from hydrographical measurements in the Arkona Basin, Tel- lus Ser. A, 39A, 170-177, 1987a.

Stigebrandt, A., A model for the vertical circulation of the Baltic deep water, J. Phys. Oceanogr., 17, 1772-1785, 1987b.

Stigebrandt, A., Bridge-induced flow reduction in sea straits with reference to effects of a planned bridge across Oresund, Ambio, 21. 130-134, 1992.

Svansson, A., Physical and chemical oceanography of the Skagerrak and the Kattegat, Rep. 1, 88 pp., Fish. Bd., Inst. of Mar. Res., G6teborg, Sweden, 1975.

Turner, J. S., Buoyancy Effects in Fluids, 367 pp., Cambridge Univ. Press, New York, 1973.

Walin, G., On the deep water flow into the Baltic, Geophysica, 17, 75-93, 1981.

Whitehead, J. A., A. Leetmaa, and P. A. Knox, Rotating hydraulics of straits and sills, Geophys. Astrophys. Fluid Dyn., 6, 101-125, 1974.

Wyrtki, K., Schwankungen im wasserhaushalt der Ostsee, Dtsch. Hy- drogr. Z., 7, 91-129, 1954.

B. Liijebladh and A. Stigebrandt, Department of Oceanography, G6teborg University, Earth Science Centre, S-41381 G6teborg, Swe- den. (e-mail: [email protected])

(Received January 12, 1995; revised August 31, 1995; accepted September 6, 1995.)

Date post:	30-Nov-2023
Category:	Documents
Upload:	gu-se
View:	0 times
Download:	0 times

Observations of the deepwater flow into the Baltic Sea

Documents