ORIGINAL PAPER

Hans Burchard Æ Hans Ulrich Lass Æ Volker Mohrholz

Lars Umlauf Æ Jurgen Sellschopp Æ Volker FiekasKarsten Bolding Æ Lars Arneborg

Dynamics of medium-intensity dense water plumes in the Arkona Basin,Western Baltic Sea

Received: 13 January 2005 / Accepted: 28 June 2005 / Published online: 11 October 2005� Springer-Verlag 2005

Abstract In this study, the dynamics of medium-inten-sity inflow events over Drogden Sill into the Arkona Seaare investigated. Idealised model simulations carried outwith the General Estuarine Transport Model suggestthat most of the salt transport during such inflow eventsoccur north of Kriegers Flak, a shoal with less than20 m water depth surrounded by water depths of morethan 40 m. This assumption about the pathway is sup-ported by recent ship-based observations in the ArkonaSea during a medium-intensity inflow event. The prop-agation of a saline bottom plume could be observedduring several days after having passed Drogden Sill. Inthe area north of Kriegers Flak the plume was about10 m thick, and propagated with more than 0.5 m s�1

and a salinity of up to 20 psu (with ambient watersalinity being 8 psu) eastwards. Although the modelsimulations were idealised, the structural agreement be-tween the observation and model result was good. Thestructure and pathways of these medium-intensity inflowevents are of specific interest due to the plans forerecting extensive offshore wind farms in the Arkona Seawhich may under certain circumstances lead to increasedentrainment of ambient water into the bottom plumes.

Keywords Bottom plumes Æ Entrainment ÆTurbulent mixing Æ Baltic Sea

1 Introduction

The process of saltwater inflow from the Kattegat viathe Danish straits into the Arkona Sea is well investi-gated (Matthaus and Frank 1992). This water exchangewith the North Sea is governed by processes with twoseparate time scales. The exchange associated with theestuarine circulation is dominating on time scales longerthan a few months. The water exchange on time scalesshorter than a month is predominantly barotropic anddriven by the sea level differences between the Kattegatand the Baltic Sea. Barotropically driven inflow into theBaltic Sea advects thick layers of saline Kattegat watersthrough the Danish straits towards the Arkona Sea. Abarotropic inflow event transporting a large amount ofsaline water into the Baltic Sea is called a major inflow(Matthaus and Frank 1992). These events, which areoccurring at the inter-annual time scale and have thepotential to reach the bottom waters of the Gotland Sea,are important for replacing the stagnant bottom water inthe basins of the Baltic Sea. Recent baroclinic andbarotropic inflow events occurring during the years 2002and 2003 have been investigated and documented indetail by Feistel et al. (2003a, b, 2004).

Inflow events transporting smaller amounts of Katte-gat water into the Arkona Basin, the so-called medium-intensity inflow events, occur several times per year. Themixing of these water masses with the ambient water ofthe Arkona Basin often results in a density anomalyranging between 7 and 8 kg m�3 in the Bornholmgatt(Fig. 1). Water of this density has the capability to con-tinue to flow into the intermediate layers of the BornholmBasin and probably even the Eastern Gotland Basin.Both ventilation processes, the major inflows and themedium-intensity inflow events, are important for main-taining the haline stratification and thus play a key rolefor the ecological regime of the entire Baltic Sea.

Responsible Editor: Alejandro Souza

H. Burchard (&) Æ H. U. Lass Æ V. Mohrholz Æ L. UmlaufBaltic Sea Research Institute Warnemunde, Seestr. 15,18119 Rostock-Warnemunde, GermanyE-mail: hans.burchard@io-warnemuende.de

J. Sellschopp Æ V. FiekasFWG, Klausdorfer Weg 2-24, 24148 Kiel, Germany

K. BoldingBolding Burchard Hydrodynamics GbR, Strandgyden 25,5466 Asperup, Denmark

L. ArneborgDepartment of Oceanography,Goteborg University, Guldhedsg. 5A,Box 460, 40530 Goteborg, Sweden

Ocean Dynamics (2005) 55: 391–402DOI 10.1007/s10236-005-0025-2

In this paper we investigate the dynamics of suchmedium-intensity inflow events, since they may be sub-ject to additional mixing due to the foundations ofextensive offshore wind farms which are projected in thearea of the Arkona Sea. This problem is not relevant forthe case for major inflows which fill up most of theArkona Sea with high-salinity water in such a way thatadditional mixing has no significant effect. The pathwaysof saltwater spreading from the sills into the Arkona Seahave recently been investigated by Lass and Mohrholz(2003). The inflow events over Darss Sill in the west ofthe Arkona Sea typically occur a few days after inflowstarts at Drogden Sill, such that the overflow at DrogdenSill can be investigated separately. Saline water spillingover the Drogden Sill propagates southward, flowsaround the Kriegers Flak and follows the depth con-tours along the southern Arkona Basin towards theBornholmgatt (Fig. 1). Such an inflow event has recentlybeen observed in detail for the first time and some of theresults are presented here.

In this paper, we compare idealised simulations ofmedium-intensity inflow events over Drogden Sill withrecent observations. First the numerical model, GeneralEstuarine Transport Model (GETM), and its setup forthe Arkona Sea are briefly described and the resulting

propagation of the dense bottom plume is shown (Sect.2.1). Field observations of a medium-intensity inflowevent investigated in January and February 2004 aredescribed in Sect. 2.2. A detailed comparison betweenobservations and idealised model results is given in Sect.3. Finally some conclusions and a future outlook arepresented in Sect. 4.

2 Methods

2.1 Numerical modelling

For simulating the dynamics of medium-intensity inflowevents over Drogden Sill, the GETM (Burchard andBolding 2002; Burchard et al. 2004; Umlauf and Lem-min 2005) has been applied. GETM is a three-dimen-sional free-surface primitive equation model using theBoussinesq and boundary layer approximations. Verti-cal mixing is parameterised by means of a two-equationk-e turbulence model coupled to the algebraic second-moment closure by Canuto et al. (2001) (see Burchardand Bolding 2001) explicit horizontal mixing is ne-glected. For the discretisation, a high-resolutionbathymetry (0.5 nm resolution) has been used as well as

Fig. 1 Bathymetric map of the Arkona Sea. The western purple lineshows the track for the first (Feb 1 from 16:20 to 20:41 h,southwards) and the western red line shows the track for the second(Feb 2 from 10:24 to 14:47 h, southwards) observational sectionsouth of Drogden Sill. The eastern red line shows the track for theKriegers Flak observational section (Feb 5 from 10:45 to 15:31 h,

southwards). The two green lines indicate the sections south ofDrogden Sill and across Kriegers Flak, as they have been extractedfrom the model simulation. The diamond indicates the positionnorth of Kriegers Flak where the water column measurementshave been carried out

392

bottom- and surface-fitted vertical coordinates with 25vertical layers and a horizontally homogeneous bottomlayer thickness of 0.4 m [for details of this version ofgeneral vertical coordinates, see Burchard and Bolding(2002)], such that the flow can smoothly advect along thebed. For momentum and salinity, the third-order TotalVariation Diminishing (TVD) advection scheme byLeonard (1991) is applied by using a directional-splitmethod as suggested by Pietrzak (1998). Temperaturevariations have been neglected in the idealised simula-tions presented here. The model domain (Fig. 2) hasopen boundaries at the northern end of the Sound,towards the West across the Fehmarn Belt and towardsthe East along 14�46.5¢E.

The model is forced by a sea surface elevation slightlyincreased by 0.02 m at the boundary to the Kattegat in

the North and a prescribed salinity for inflowing waterof 25 psu, whilst the sea surface elevation is forced tozero elevation at the other open boundaries. As initialconditions, flow at rest with a salinity of 8 psu is chosen.At the sea surface a spatially and temporally homoge-neous wind stress of 0.22 N m�2 from south-west (240�)is applied to the momentum equations. This value hasbeen calculated from a wind speed of 12 m s�1 typicalfor early Feb 2004 (Fig. 5) according to the bulk for-mula by Large and Pond (1981). Compared to full-scalerealistic simulations, the idealisations are due to thehomogeneous initial and boundary conditions, the sta-tionary and homogeneous surface stresses, and theneglect of surface and boundary heat fluxes.

An inspection of Fig. 2 shows that after about 5 daysof simulation, the Sound is filled with high-salinity water

Fig. 2 Near-bed (upper two panels and lower left) and near-surface(lower right panel) salt distribution after 1 month of simulation in aquasi-steady state. Note the different scales for the surface salinity

plot, which emphasise the sinking of Sound water at the DrogdenSill and a small amount of mixing of saline waters into the surfacebrackish waters

393

which then starts flowing southwards over Drogden Sill.After 15 days, Kriegers Flak is fully surrounded byeastward flowing saline bottom water, the northern andsouthern branch of which join at the south-eastern endof Kriegers Flak. After about 30 days of simulation, aquasi-steady state is reached where the saline bottomwater covers most of the Arkona Sea and is flowing overBornholmgatt in the Bornholm Sea. A plot of the seasurface salinity shows that saline water is detaching fromthe sea surface just south of Drogden Sill and only at afew positions, sea surface salinities above the back-ground value of 8 psu are found. It is noteworthy thatmost of the saline water flowing through the Arkona Seapasses Kriegers Flak on its northern side. Kriegers Flakitself is not covered with saline bottom water.

This result is counter-intuitive in the sense that amainly geostrophic propagation of the plume would resultin flow along the isobaths and thus in this case a propa-gation southwards along the Danish coast and west ofKriegers Flak, with a turn to the east at Darss Sill and afurther propagation along the southern boundary of theArkona Sea towards Bornholmgatt (Lass and Mohrholz2003). Thus the geostrophic balance of the near-bottomflow must be significantly modulated by bottom friction,which draws the plume down the slope. In addition, theremay be an internal hydraulic control west of Kriegers Flakthat limits the amount of dense water flowing into thesouthern Arkona Sea. It may, however, also be the casethat the stationary model forcing is too far from realisticconditions, model physics (specifically the turbulenceclosure model) is unrealistic or the numerical accuracy islow (e.g. resulting in numerical diffusion). It is thereforeuseful and challenging to compare the model results tofield observations of a medium-intensity inflow event overDrogden Sill, see Sect. 2.2.

2.2 Field observations

From Jan 26 to Feb 13, 2004 the FWG Kiel (Germany)organised a field survey in the Arkona Sea during whichmedium-intensity saltwater inflow over Drogden Silloccurred, starting on Feb 1. This inflow event could bereconstructed on the basis of observed sea level data.Several studies [e.g. Jakobsen et al. (1997), Green andStigebrandt (2002)] show that the flow through theSound is well described by a quadratic friction law oftype

gN � gS ¼ KQjQj; ð1Þ

where gN � gS is the total head difference from southto north, which is taken to be equal to the water leveldifference, Q is the southward volume flux, and K is thespecific resistance. Jakobsen et al. (1997) estimated thevolume fluxes from ADCP (Acoustic Doppler CurrentProfiler) measurements from the Drogden and Flintenchannels, and used sea level data from Hornbæk andViken to the North and Rødvig and Skanør to thesouth. They found the average value for the specificresistance to be K=2.26·10�10 s2 m�5, which is thevalue used in the present work, to calculate the volumefluxes from the observed sea level differences betweenViken and Skanør (Fig. 3). The accumulated volumeflux over Drogden Sill (Fig. 3) indicates that a signifi-cant inflow event must have started around Jan 31. Theflux Q deduced from Eq. 1 and southward currentvelocity data observed at 4.1 m depth on Drogden Sill(Fig. 4) indeed correlated (r=0.57). Significant south-ward velocity peaks of more than 0.5 m s�1 occurredon Jan 30 at 7 a.m., Feb 1 at 9 a.m. and between Feb4 at 8 a.m. and Feb 5 at 1 p.m. Specifically, the stronginflow event on Feb 1, with current velocities of more

Fig. 3 Water mass flux (upperpanel) and accumulated watermass flux (lower panel) throughthe Sound in Jan and Feb 2004calculated by means of Eq. 1.Positive fluxes are directedsouthwards into the ArkonaSea

394

than 1 m s�1, ceased around 6 p.m. of that day. Theseobservations are associated with southward fluxestimates of more than 30,000 m3 s�1 (Fig. 3). The lackof proportionality between the velocity observationand flux estimate may be mostly explained by flowdivergence along Drogden Sill. This also explains whythe time-integrated flow velocity does not reach zero-level after these inflow events (Fig. 4), in contrast tothe accumulated flux estimate (Fig. 3). The above de-scribed inflow events are obviously the result of wind

from south-west, increasing from 2 m s�1 on Jan 27 to17 m s�1 on Feb 1, with a few north-westerly bursts(Fig. 5). The strongest southward flow on Feb 1 isassociated with the maximum wind velocity on Feb 1, 9a.m., blowing from a north-westerly direction.

No salinity time series for the Drogden Sill areavailable during this period. However, a CTD section onFeb 1 (16:20–20:41 h) shows that a saline bottom plumehas already reached a latitude of 55.2�N, approximately45 km south of Drogden Sill (Fig. 6). Given the

Fig. 5 Wind speed (upper panel)and direction (lower panel)observed by the stations DarssSill (DS) and Arkona Sea (AS)operated by the Baltic SeaResearch InstituteWarnemunde (Germany) in Janand Feb 2004

Fig. 4 Southward currentvelocity (upper panel) andaccumulated surface southwardcurrent velocity (lower panel)observed by an ADCP atDrogden Sill in Jan and Feb2004 at a depth of 4.1 m. Thedata have kindly been providedby The Royal DanishAdministration of Navigationand Hydrography inCopenhagen, Denmark

395

propagation speed of the plume deduced from the ob-served southward current velocity within the plume(approximately 0.5 m s�1, see also Fig. 6), it may bereconstructed that the plume must have passed DrogdenSill on the afternoon of Jan 31.

This is consistent with the estimate of southward fluxand the observation of southward velocity over DrogdenSill discussed above and shown in Figs. 3 and 4, whichpredict significant southward transport at that time. InSect. 3, three combined CTD–ADCP sections (two fromDrogden Sill southwards and one across Kriegers Flak)and the instantaneous CTD and ADCP profiles ob-tained during a 24 h station are compared in detail tothe results of the idealised model simulation.

3 Observation–simulation comparison

In this section, observations and model results arecompared for three different cross-sections at different

times and for vertical profiles at a station north ofKriegers Flak.

The inflow event itself is documented by the CTD andADCP section taken on Feb 1 (Fig. 6). Salinity valuesare up to 21 psu near Drogden Sill, which poses a sig-nificant gradient with respect to the ambient salinity of8 psu. The along-plume (southward) velocity compo-nent reaches 0.5 m s�1 over a large portion of the plume.The plume thickness amounts to up to 10 m, with a well-defined halocline on top and a fairly well-mixed core ofthe plume. The high current velocity and the well-mixedcore of the plume indicate high levels of turbulencewithin the plume. An eastward velocity component of upto 0.3 m s�1 at 55.35�N shows that the plume is headingsouth-eastwards, probably following the talweg towardsKriegers Flak. This section had been observed duringthe inflow with peak velocities of more than 1 m s�1 atDrogden Sill. When the inflow ceased, 18 h later, theplume is still present, having now extended southwards

Fig. 6 Observed salinity on a north–south transect south from Drogden Sill, deduced from CTD profiles. Left panel: Feb 1 from 16:20 to20:41 h; right panel: Feb 2 from 10:24 to 14:47 h. The red lines show the positions of the single CTD profiles

396

to 55.1�N (Fig. 6). Maximum salinity is still higher than20 psu. However, at its northern end towards DrogdenSill, the plumes have already collapsed due to the ceasingof supply from the Sound. Current velocities in theplume were down to 0.2 m s�1.

The observed propagation of the plume on Feb 1 and2 is compared to results from the idealised simulationdescribed in Sect. 2.1 for days 8 and 9 (Fig. 7). Also themodel results show high salinities of above 20 psu, withwell-mixed conditions in the core of the plume and astrong halocline on top. The current velocity inside theplume is, however, significantly lower than in the ob-served plume, with southward velocities of up to0.3 m s�1 only. The reason for this may be that theplume in reality has a substantial barotropic componentwith a significant transport over Drogden Sill (see dis-cussion in Sect. 2.2) which is strongly underestimated bythe idealised simulation with stationary forcing. In de-tail, the simulated quasi-steady state flux over DrogdenSill is approximately 26,000 m3 s�1, whereas the ob-served peak flux during Feb 1 was about 60,000 m3 s�1

(Fig. 3). Due to this stationary forcing, the plume supplyfrom Drogden Sill never ceases, such that also on modelday 8 the plume is highly dynamic, having reached itsmaximum salinity of 25 psu at the northern end andreached a maximum extent of down to 55.1�N. Thevelocity structure in the simulated plume on day 8 showsthat the plume is propagating further to the east fromthis model section.

The next CTD and ADCP section is available for Feb5 over Kriegers Flak with a water depth of less than20 m, covering also the deeper channels to the north andto the south with water depths down to more than 35 m.It can be clearly seen that the plume has reached thenorthern channel and forms a bottom layer of about10 m thickness with salinities up to more than 18 psu(see Fig. 8). The halocline separating the plume from theambient water with salinities of about 8 psu is signifi-cantly sloping up towards Kriegers Flak in the south,while the core of highest salinity is displaced towards thenorth (Fig. 10). Although the noise to signal ratio forthe ADCP measurements from the cruising ship was

Fig. 7 Simulated salinity and current velocity after 8 and 9 days, respectively, on a north–south transect at 12�30¢E south of Drogden Sill

397

fairly high due to the low concentration of particles inthe water, an enhanced eastward current velocity of upto 0.5 m s�1 in the region of the halocline is observed.This is accompanied by a small southward velocitycomponent of up to 0.2 m s�1 in the same region.

The model result basically reproduces these obser-vations of salinity and current velocity north of KriegersFlak (Fig. 9). Salinities north of Kriegers Flak reach upto 21 psu whereas they only reach 18 psu south ofKriegers Flak. As in the observations, the halocline is

Fig. 8 Observed salinity andcurrent velocity on a north–south transect across KriegersFlak deduced from CTDprofiles and ADCPobservations, respectively. Theobservations were taken on Feb5 from 10:45 to 15:31 h. The redlines show the positions of thesingle CTD profiles

398

sloping towards Kriegers Flak in the northern channel,and the core of the salinity is displaced towardsthe north (Fig. 10). Furthermore, the halocline issignificantly wider towards Kriegers Flak than towardsthe north, a feature which can partially be seen in theobservations as well. In further agreement with the

observations, maximum eastward current velocities ofup to 0.5 m s�1 are observed in the halocline, with asouthward component in the halocline and a northwardcomponent near the bed. The latter forms a secondaryflow structure in the transverse direction, consistent withthe classical Ekman theory. The displacement of the

Fig. 9 Simulated salinity andcurrent velocity in quasi-steadystate on a north–south transectat 13�E across Kriegers Flak

399

salinity core towards the north may be explained by thismechanism.

On Feb 7, during a 24 h station north of KriegersFlak (Fig. 1), velocity and salinity profiles were observedby means of a ship-mounted ADCP and a free-fallingmicro-structure profiler, respectively. The results showthat the 10 m thick bottom plume is still present.Salinities reach up to 20 psu flowing eastwards withcurrent velocities up to 0.6 m s�1 (Fig. 11). The plume isoverlaid by brackish water with salinities of 8–9 psuwith a sharp interface between the plume and ambientwater.

Also here, the idealised simulation reproduces mostof the observed features quite well (Fig. 11). Majordifferences between observations and the simulation arethe weaker salinity gradient at the interface between theplume and ambient water, the smaller peak salinity andthe smaller propagation speed of the plume, which maybe explained by idealisations due to the stationary modelforcing, see above. A predominant feature of the velocitystructure is the sharp eastward velocity peak visible inthe observations as well as in the model results. Thispeak is associated with the lower limit of the halocline. Itcan be explained by the fact that at this position theinternal pressure gradient still has its maximum valuewhile the bed friction decreases due to the stratificationin the halocline. It may be assumed that under theseconditions significant entrainment of ambient water intothe plume is occurring, increasing the volume of theplume and decreasing the plume salinity. Whether this isthe main mixing process subject to the plume needs to beinvestigated in further studies.

4 Summary and discussion

This combined observational and model study suggeststhat the major pathway of medium-intensity saline

plumes over Drogden Sill passes Kriegers Flak north-wards, whereas only a smaller amount of salt is trans-ported around its southern edge. In the observations noyoung Drogden Sill water could be found south ofKriegers Flak, and the simulation with its steady-stateflux over Drogden Sill showed a much weaker plumesouth than north of Kriegers Flak. More evidence thanthis would be needed for proving that this represents thegeneral pattern of salt propagation over Drogden Sill,but it is obvious that the strong plume north of KriegersFlak is a typical feature. The observational study byLass and Mohrholz (2003) in the same region had beencarried out some days after an inflow event such that itwas not possible to directly reproduce the pathways ofsaline bottom plumes. It may be necessary to revise theconclusion of Lass and Mohrholz (2003) that the majorbalance in gravity-driven plumes in the Arkona Sea isbetween pressure gradients and the Coriolis acceleration,since at least the present study suggests that also frictionplays a substantial role. This hypothesis will be investi-gated in more detail during later studies.

The observed general structure of the plume is ingood agreement with the idealised model simulation.Main features are strong shear and stratification at theinterface between plume and ambient water, with avelocity jump of more than 0.5 m s�1 and a salinity jumpof more than 10 psu over 3 m. Inside the plume, salinityis homogeneous and velocity follows the law of the wall.Cross-sections through the plume show that the interfacebetween the plume and ambient water is sloping to theright, and the salinity core is displaced towards the left.This is probably related to secondary circulation due toEkman dynamics, which needs further investigation.

A deeper insight into the dynamics of medium-intensity inflow events requires more refined modelling,with realistic forcing at the surface and the lateralboundaries and a larger model domain. Detailed obser-vations like those presented here are difficult to obtain

Fig. 10 Bottom salinity fromobservations and an idealisedmodel simulation together withbottom topography. It can beseen from both, observationsand model results, that the coreof saline water is shiftedtowards the north

400

due to the intermittency of these events. However, thedeployment of moorings over several seasons moored atsensitive locations such as north of Kriegers Flak willhelp to quantify the occurrence of inflow events.

For further numerical experiments, it will also benecessary to analyse and, if needed, to improve thenumerical mixing properties of the model. The 25model layers refined towards the bottom are probablynot sufficient to resolve the sharp gradients at the

interface between the plume and ambient water(Fig. 11). It may also be that the advection schemesused here are too diffusive or the horizontal resolutionof the model too coarse. These potential problems willbe analysed later. In contrast to this, the physicalmixing prediction of the model should be fairly realisticdue to the use of a well-tested turbulence closuremodel. In a later study, we will compare micro-struc-ture observations carried out on Feb 5–6 with model

Fig. 11 Observed salinity and current velocity (upper panels) and simulated salinity and current velocity (lower panels) at a position northof Kriegers Flak during a medium-intensity inflow situation on Feb 7 at 4 a.m. Shown for the observations are hourly averages

401

results for the turbulent dissipation rate. Furthermore,empirical models for plume dynamics (Bo Pedersen1980 and Stigebrandt 1987) will be compared to ourobservations in detail.

It also remains unclear how the medium-intensityinflow event propagates further, after having passedKriegers Flak. It will probably spread out onto the flatbottom of the Arkona Sea and add to the thin layer ofdense bottom water typically found there. It will beinteresting to understand how much of this water ismixed locally in the Arkona Sea and how much isflowing over Bornholmgatt into the Bornholm Sea forventilating the halocline there.

The understanding of these highly intermittent eventsin the Arkona Sea is of specific interest, since extensiveoffshore wind farms are planned in this area. An accu-mulation of such constructions in the pathway of densebottom water could lead to a dilution of dense bottomwater and decreased ventilation of the halocline in ba-sins further east such as the Bornholm Sea. For largebridge projects across the Great Belt and the Sound, theso-called zero blocking solution has been proposed(Hansen and Møller 1989 and Stigebrandt 1992) mean-ing that no changes of the transport of water and saltbetween the Baltic Sea and the North Sea should occur.Concerning offshore wind farms such an internationallycoordinated environmental impact strategy has not yetbeen developed.

Acknowledgements The work for this study was carried out in theframework of the international QuantAS Consortium (Quantifi-cation of water mass transformation processes in the Arkona Sea),which is partially funded by the QuantAS-Off project (Quan-tAS—Impact of Offshore Wind Farms) by the German FederalMinistry of Environment, Nature Conservation and Nuclear Safety(BMU). We are further grateful for the support and flexibility ofthe crew on board of the R/V Helmsand without which theobservations presented here could not have been obtained. TheRoyal Danish Administration of Navigation and Hydrographymade the current velocity observations on Drogden Sill available.The sea level data were obtained from the Swedish Meteorologicaland Hydrological Institute. All two-dimensional graphics in thispaper have been generated with the Generic Mapping Tool (GMT),and we are grateful to the specific support given by Paul Wessel,Hawaii. We are further grateful for the constructive comments bytwo anonymous reviewers.

References

Bo Pedersen F (1980) A monograph on turbulent entrainment andfriction in two-layer stratified flow. Ph.D. thesis, DanishTechnical University, Lyngby, Denmark, published as Seriespaper 25, IHHE, DTU, Lyngby, Denmark, 397 pp

Burchard H, Bolding K (2001) Comparative analysis of four sec-ond-moment turbulence closure models for the oceanic mixedlayer. J Phys Oceanogr 31:1943–1968

Burchard H, Bolding K (2002) GETM—a general estuarinetransport model. Scientific documentation. Tech Rep EUR20253 EN, European Commission

Burchard H, Bolding K, Villarreal MR (2004) Three-dimensionalmodelling of estuarine turbidity maxima in a tidal estuary.Ocean Dyn 54:250–265

Canuto VM, Howard A, Cheng Y, Dubovikov MS (2001) Oceanturbulence. Part I: one-point closure model. Momentum andheat vertical diffusivities. J Phys Oceanogr 31:1413–1426

Feistel R, Nausch G, Matthaus W, Hagen E (2003a) Temporal andspatial evolution of the Baltic deep water renewal in spring2003. Oceanologia 45:623–642

Feistel R, Nausch G, Mohrholz V, Lysiak-Pastuszak E, Seifert T,Matthaus W, Kruger S, Hansen IS (2003b) Warm waters ofsummer 2002 in the deep Baltic Proper. Oceanologia 45:571–592

Feistel R, Nausch G, Heene T, Piechura J, Hagen E (2004) Evi-dence for a warm water inflow into the Baltic Proper in summer2003. Oceanologia 46:581–598

Green M, Stigebrandt A (2002) Instrument-induced linear flowresistance in Oresund. Cont Shelf Res 22:435–444

Hansen N-EO, Møller JS (1989) Zero blocking solution for theGreat Belt Link. In: Pratt L (ed) Physical Oceanography of SeaStraits. NATO ASI series C, vol 318, pp 153–170

Jakobsen F, Lintrup MJ, Møller JS (1997) Observations of thespecific resistance in Oresund. Nordic Hydrol 28:217–232

Large WG, Pond S (1981) Open ocean momentum flux measure-ments in moderate to strong winds. J Phys Oceanogr 11:324–336

Lass HU, Mohrholz V (2003) On the dynamics and mixing ofinflowing salt-water in the Arkona Sea. J Geophys Res108:3042, doi: 10.1029/2002JC001,465

Leonard BP (1991) The ULTIMATE conservative differencescheme applied to unsteady one-dimensional advection. Com-put Methods Appl Mech Eng 88:17–74

Matthaus W, Frank H (1992) Characteristics of major Baltic in-flows— a statistical analysis. Cont Shelf Res 12:1375–1400

Pietrzak J (1998) The use of TVD limiters for forward-in-timeupstream-biased advection schemes in ocean modeling.Monthly Weather Rev 126:812–830

Stigebrandt A (1987) A model for the vertical circulation in theBaltic deep water. J Phys Oceanogr 17:1772–1785

Stigebrandt A (1992) Bridge-induced flow reduction in sea straitswith reference to effects of a planned bridge across Oresund.Ambio 21:130–134

Umlauf L, Lemmin U (2005) Inter-basin exchange and mixing inthe hypolimnion of a large lake: the role of long internal waves.Limnol Oceanogr 2005, in press

402