coastal–open ocean exchange in the black sea: … · cuniversite libre de bruxelles, ecologie des...

Estuarine, Coastal and Shelf Science (2002) 54, 601–620doi:10.1006/ecss.2000.0668, available online at http://www.idealibrary.com on

Coastal–open Ocean Exchange in the Black Sea:Observations and Modelling

E. V. Staneva,f, J. M. Beckersb,g, C. Lancelotc, J. V. Stanevad,h, P. Y. Le Traone,E. L. Penevaa and M. Gregoireb,i

aDepartment of Meteorology and Geophysics, University of Sofia, 5 James Bourchier Street, 1126 Sofia, BulgariabUniversite de Liege, GHER, Sart-Tilman B5, B-4000 Liege, BelgiumcUniversite Libre de Bruxelles, Ecologie des Systemes Aquatiques, Campus de la Plaine, CP 221,Boulevard du Triomphe, B-1050 Bruxelles, BelgiumdNational Institute of Meteorology and Hydrology, Sofia, BulgariaeCLS–Space Oceanography Division, Toulouse, France

Received October 1998 and accepted in revised form February 2000

The interaction between physical and biological processes in the areas of continental margins governs the variability ofecosystems. The complexity of processes in these areas requires detailed studies combining modelling and surveyingefforts. One promising step in this direction was undertaken in the framework of the EROS 21 project, focusing on theshelf part of the north-western Black Sea. In the present paper, we focus on the results of physical studies aiming toimprove the understanding of the fundamental exchange processes in the ocean margins, as well as to quantify some ofthem in the Black Sea. We illustrate the capabilities of circulation models to reproduce physical processes with differenttime- and space-scales: coastal waves, internal waves, baroclinic Rossby and topographic waves. Another class ofimportant phenomena in the coastal zone is associated with convection. Sources at the sea surface and in the outflowareas give rise to plume dynamics that play a crucial role in the vertical mixing and provide the mechanism for water-massformation. Most of the results are illustrated for the shelf part of the Black Sea. The verification of simulations isperformed by comparison with survey data, altimeter data from the Topex/Poseidon mission and radiotracer observa-tions. The latter, in combination with simulations from circulation models, are used to trace the penetration of tracersinto the intermediate and deep layers. We show that although most 90Sr is introduced by river runoff, large amounts ofthis signal penetrate the halocline in the Bosphorus Straits area and along the southern coast. Another important fractionof the river water penetrates the intermediate layers at the shelf edge in the north-western Black Sea.

� 2002 Elsevier Science Ltd. All rights reserved.

Keywords: coastal waves; upwelling; circulation; internal mixing; water mass formation; ventilation of coastal zone

fCorresponding author. Present address: ICBM, University ofOldenburg, Postfach 2503, D26111 Oldenburg, Germany.Tel: +49-44-1798 4061; Fax: +49-44-1798 3404. E-mail:[email protected] Associate, National Fund for Scientific Research,Belgium.hPresent address: Alfred-Wegener-Institut for Polar and MarineResearch, PO Box 120101, 27515 Bremerhaven, Germany.iResearcher, National Fund for Scientific Research, Belgium.

Introduction

The coastal zone covers only 8% of the entire oceansurface, but its role is dominant for most processes,particularly those related to the exchange betweenland and ocean. This key area includes estuaries anddeltas and the entire region between the shorelineand the beginning of the deep ocean. It provides aboundary layer for physical, chemical and biologicalprocesses. Growing interest in better quantifying theexport from/into the coastal zone results from its

0272–7714/02/030601+20 $35.00/0

importance for the functioning of ecological and sedi-mentological systems. As is well known, an importantpart of nutrients available to coastal ecosystems issupplied from the land. This, along with thefavourable physical conditions, maintains a very highprimary production (about one quarter of the primaryproduction in the ocean is due to the coastal ocean).Part of this material settles on the bottom and anotherpart enters the deep ocean. The ratio between the twois not well known, but it is accepted that it greatlyvaries in different ocean margins. This motivatesinterdisciplinary studies on the ventilation of thecoastal zone, that is the exchange between coastalwaters and open-ocean, and the fate of organic matterproduced and imported.

The Black Sea is a basin where such exchanges areof prime importance and must be correctly under-stood. It is a deep basin (greatest depths of about

� 2002 Elsevier Science Ltd. All rights reserved.

602 E. V. Stanev et al.

2200 m), with a large shelf and continental slopecovering 30–40% of its surface. This land-lockedbasin is located in the temperate and subtropicclimatic zone, having a negative freshwater balance atits surface caused by excess evaporation (evaporationminus precipitation yields about 50 km3 per year, withprecipitation of 300 km3 per year; O} zsoy & U} nluata,1998. However, it is its wide drainage area, covering alarge part of Europe and Asia and providing a total offresh water supply of about 350 km3 per year thatmakes it very different from most other seas in tem-perate and subtropical areas. The excess fresh water atthe surface of the Black Sea, the restricted exchangewith the Mediterranean Sea through the Straits ofBosphorus and the basin shape and topography havefundamental consequences for its physical system,creating a unique chemical and biological environ-ment. The river runoff affects most of the physicalcharacteristics, which makes them strongly dependenton the hydrological cycle over large areas of Europe,as well as on coastal processes. This dependency iseven stronger for the biological processes, since thelatter are affected by the pollution of river watersoriginating from a vast area of Europe. This hasresulted in the well-known eutrophication, observedover the past 30 years, and in the complete deterior-ation of the Black Sea ecosystem. Eutrophication isknown as a process of intensive algal bloom, which hasa strong local dependency. The cause, characteristicsand, in particular, the quantification of the processesassociated with the recent changes are not easilyaddressed from the point of view of observations,since the data are too sparse. Much can be done usingnumerical simulations, and an illustration of this isgiven by Lancelot et al. (2002) in this issue. However,the physical and biological processes governing theaccumulation of biomass require us to address thisphenomenon as three-dimensional, and deal withmass fluxes due to currents and turbulence. Inthis paper, we will illustrate the progress made inthe understanding of physical issues that are directlyrelated to the functioning of the biological systems,and we will compare some simulations with in situ andsatellite data from surveys and remote sensing. Sincethe coastal–open ocean exchange is not a local pro-cess, we will analyse the circulation using basin-widedata and numerical simulations. Further details on thedynamics and biological transformations can be foundin the accompanying papers (in this issue) of Beckerset al. (2002) and Lancelot et al. (2002). Instead, thepresent paper details and validates the physical mod-elling results underlying these papers. We will showthat the output from numerical modelling presents animportant supplement to existing data.

Data and models description

Data

Several data sets have been used to compare modelsimulations with observations. The climatic data setdescribes monthly mean temperature and salinitybased on more than 25 000 stations over the last70 years (Altman et al., 1987).

The temperature and salinity database of theCo-operative Marine Science Black Sea (CoMSBlack)programme (Oguz, et al., 1993, 1994; O} zsoy &U} nluata, 1998) was used to describe synoptic featuresin the Black Sea circulation. In particular, the datacollected during three quasi-synoptic surveys (2–29September 1991, 4–26 July 1992 and 2–14 April1993) were interpolated onto a regular grid withresolution 1

12� in the meridional and 19� in the zonal

direction at 35 levels.The temperature and salinity profiles measured on

board the RV Professor Vodyanitsky, during the cruiseof April–May 1997 (Lancelot & Egorov, 1997), wereused to study the shelf–open sea exchange in thenorth-western Black Sea.

The recent US/French mission Topex/Poseidon(T/P) provided the scientific community with highquality altimeter data. Their errors are below 3 cmrms, thus the accuracy of estimates on the variationsof Black Sea mean sea-level (MSL) is quite good.T/P data from October 1992 to July 1997 of thelatest version of Topex/Poseidon (T/P) M-GDRsdistributed by AVISO (MGC-B, version 2) wereused (AVISO, 1996). Standard altimetric correctionsare applied, except for ocean tides and atmosphericpressure, which are very small and are not correctedfor the Black Sea. Sea-level anomaly (SLA) relativeto a 4-year mean (1993–1996) is then obtainedusing a conventional repeat-track analysis and asuboptimal space/time interpolation method (LeTraon et al., 1998) onto a 0.2��0.2� spatial grid,with 10-day averages from the 1 to 3 days repeattrack data.

Atmospheric forcing was derived from the climatichandbook edited by Sorkina (1974) and the meteoro-logical data of the Hadley Centre, United KingdomMeteorological Office (UKMO). This data setincludes twice-daily temperature, relative humidityand wind velocity for the period June 1993–May1995. The resolution is 0.44� in latitude and longi-tude. The climatic data set of Sorkina (1974) origi-nates from coastal and ship measurements (67 000measurements in total). The procedure for the calcu-lation of wind stress is described by Staneva andStanev (1998).

Coastal–open ocean exchange in the Black Sea 603

T 1. Models used in this study, and their resolution

Models

Resolution

horizontal(km)

vertical(levels)

DMG-MOMcoarse 28 24fine 9 24

GHERcoarse 15 25fine 5 25

Numerical models of the Black Sea circulationWe will here briefly review two basin-wide models: theDMG and GHER model. They are described in detailin the references cited below. The GHER model ispresented in this issue by Beckers et al. (2000), andwill, therefore, only be described schematically,focusing on its major differences from the DMGmodel. Both models use a set of primitive equationsfor velocity, temperature and salinity in hydrostaticapproximation. The DMG model is based on theModular Ocean Model (MOM) code (Pacanowskiet al., 1991), widely used in ocean modelling, whilethe second one is based on the GHER mathematicalmodel code (e.g. Nihoul et al., 1989; Beckers, 1991).The application of these models to the Black Sea isdocumented in the papers of Stanev et al. (1997)and Stanev and Beckers (1999a, b). As can be seenfrom these works, many differences exist betweenthe model formulations, numerical schemes andparameterizations. However, initial conditions areessentially the same and the forcing functions havealmost the same climatic characteristics. The forcingfunctions used in the DMG-MOM model are basedon atmospheric analysis data (atmospheric tempera-ture, humidity and wind). Aerodynamic bulk formu-lae are used to compute heat and momentum fluxes,using stability-dependent exchange coefficients. Therealistic forcing of DMG-MOM ensures correct simu-lations of events associated with water-mass formationand the ventilation of coastal regions. As shown in anumber of papers addressing parameterizations, sen-sitivity studies and intercomparisons with observa-tions (Stanev et al., 1997, 1998; Staneva & Stanev,1997, 1998; Staneva et al., 1999), the model is welltuned to the Black Sea conditions and realisticallyreplicates the major circulation and thermohalineproperties of the Black Sea system. The forcing of theGHER model is more simple and includes only theseasonal variability of atmospheric forcing (the seasurface temperature and salinity are relaxed tomonthly mean climatological data). River dischargesfrom the three main rivers in the Black Sea (i.e.Danube, Dnepr and Dnestr) are also prescribed. Inaddition, momentum fluxes are computed from theclimatological monthly data and are interpolated ateach time-step. The data used to force both modelsare described in detail in the paper of Staneva andStanev (1998).

Bottom topography is taken from the UNESCObathymetric map and discretized with the model reso-lution. One important difference between the twomodels is the vertical discretization. MOM uses geo-metric depth as vertical co-ordinates, with variablethickness of model layers, and the Arakawa B-grid in

the horizontal. To avoid possible artefacts associatedwith the amplification of basin waves over the abyssalplain (Stanev & Rachev, 1999), we increase the dis-cretization there. Thus, it varies from 5 m in thesurface 20 m layer, 10 m until a depth of 90 m,decreasing to 400 m in the deep homogeneous layersincreasing again to 60 m in the deepest levels. TheGHER model uses the double � co-ordinate system(which allows us to represent the abyssal plain pre-cisely) in the vertical and the Arakawa C-grid in thehorizontal. An important physical difference betweenthe two models is related to the use of a 2D prognosticvariable: total stream function in MOM and free seasurface in the GHER model. The subgrid parameter-izations are also different, and are documented in theabove papers. Both models have been developed suchthat they can be run with two different resolutions inthe horizontal (Table 1). This allows good efficiencywhen studying large-scale circulation and relativelyslow processes of water-mass formation with coarseresolution. Fine resolution is used to study the impactof eddies on the circulation (e.g. Staneva & Stanev,1997; Stanev & Staneva, 2000; Gregoire, 1998).Since the Rossby radius of deformation in the BlackSea is of the order of 20–30 km, the two fine-resolution models (with horizontal resolutions of 9and 5 km for the DMG-MOM and GHER models,respectively) can be seen as eddy-resolving models. Aswill be demonstrated when discussing the results ofthe simulations, the eddy resolution is of utmostimportance if we want to adequately replicate theexchange processes in the frontal area, as well as onthe shelf.

This paper addresses the simulation of the basin-wide circulation, with a particular focus on thedescription of the north-western shelf circulationand on the estimation/quantification of the exchangesbetween shelf waters and open-sea waters.

We will illustrate different results produced by thetwo models, since they are complementary. It is worth


noting that the evolution of the circulation andthermohaline fields simulated by the two models obeythe same type of behaviour. This proves that bothmodels are calibrated in such a way as to give closeresults, in particular with respect to the seasonalvariability of circulation.

The Black Sea circulation: evidence fromobservations

Black Sea surface elevation and currents

The circulation of the Black Sea has been widelyillustrated, and further details and references can befound in Blatov et al. (1984), Stanev et al. (1988),Simonov and Altman (1991) and O} zsoy and U} nluata(1998). We restrict the analyses in this study to issuesdirectly linked to the coastal–open sea exchange,presenting estimates based largely on new data andmodelling.

Currents in the Black Sea are mainly quasi-geostrophic, but strong deviations from this balanceexist in the surface and bottom boundary layers, aswell as in the jet-like current (encompassing the entirebasin) where inertial force is substantial. Wind isthe main driving force, tending to create a cyclonicgeneral circulation (Stanev, 1990; Stanev & Beckers,1999a). The buoyancy anomalies due to river runoff,precipitation and evaporation enhance the cycloniccirculation, since most of the fresh water enters the seain the coastal area (Stanev, 1990; Oguz et al., 1995;Bulgakov et al., 1996). This forcing exerts an indirect,but very strong, impact on the circulation, forming(together with the exchange through the Straits ofBosphorus) the unique vertical stratification.

It is accepted that the Black Sea can be divided intotwo major circulation areas: the cyclonic (in the basininterior) and the anticyclonic (between the jet currentand the coast). The anticyclonic area is narrow, sincethe continental slope is very close to the coast overmost of the sea, and the circulation in this area isdominated by a number of small coastal eddies. Ascan be shown from the dynamical analyses of therecent basin-wide quasi-synoptic surveys (Oguz et al.,1993, 1994; Korotaev et al., 1998), the dynamicheight correlates with the general climatic pattern.What is less well known, and is very impressivelyillustrated by the recent T/P data, is the spatialvariability of the sea surface. In order to compareit with the existing observations, we subtract theclimatic signal from the dynamic heights of threeComsBlack surveys. The reference level is taken as500 dbar, since the density anomalies below thisdepth are very small, so that the results remain almost

unchanged if we consider a deeper reference level.The results are shown in Figure 1(a–c). A number ofsynoptic features dominates the anomaly pattern,indicating strong mesoscale/sub-basinscale variability.Horizontal scales vary between tents to 100 km. Thelargest anomalies are observed in the areas of theBatumi eddy and in the north-western Black Sea. Ascan be seen from the comparisons between theanomalies of dynamic heights obtained from hydro-graphic and T/P data, the agreement is satisfactory[cf. Figure 1(c,d)]. This proves that the two types ofdata are consistent and can be used as complementary(a large number of observations of SLE in T/P dataand a complete description of the thermohaline fieldsin the survey data). It is worth noting that the datafrom the survey resolve eddies smaller than those fromthe T/P, which is due to the relatively large distancesbetween tracks (Korotaev et al., 1998). This couldexplain the higher slope in SLE estimated from thesurvey data. We will give some further intercompari-sons between our model estimates and both types ofdata.

Seasonal variability

The sea-level elevation (SLE) oscillates with ampli-tudes of 10–20 cm. The maxima are associated withlarge river runoff in spring and early summer. Theminima are observed in late autumn and are due tosmall freshwater fluxes in summer and autumn. Thisevidence, well known from measurements in coastallocations, is nowadays supported by satellite data(Figure 2). Along with the strong signature of seasonalsignal, these data give a well-resolved trend of about3 cm yr�1 in the last 5 years (a similar trend in theCaspian Sea is much higher), indicating possible long-term changes associated with the freshwater balance.

The altimeter data clearly allows us to identifyregions with higher or lower variability (e.g. energeticBattumi and Sevastopol eddies, Figure 3).

The seasonal variability of the Black Sea circulationis externally forced and carries a substantial part of thespectral energy. In winter, not only does the windmagnitude increase, but also its curl, which con-tributes to the intensification of circulation (Stanev,1990; Staneva & Stanev, 1998). The decreasingintensity of circulation in summer is also a directconsequence of the change in the mechanical forcing.The corresponding seasonal transitions are illustratedby the shallowing of the halocline in the central(cyclonic) part of the basin and its deepening alongthe coast (anticyclonic part of the basin) in winter. Asa result, the slope of the halocline increases. The


inverse process takes place in summer, leading to adecrease in the intensity of circulation.

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

F 1. (a, b, c) Anomalies of dynamic heights (mm) from hydrographic surveys. Dynamic heights calculated from annualmean climatological data are subtracted in order to obtain the anomalies. The reference level is taken to be 500 m, whichexplains why the plot does not cover the whole area. (a) September 1991; (b) July 1992; (c) April 1993. (d) Anomaly ofsea-surface elevation from the T/P data (April 1993).

Penetration of the signals from the sea surface into thepycnocline

The Black Sea stratification (surface salinity of about17·8 and salinity at 150 m of about 21) tends to shielddeep layers from the processes occurring in the surfacelayer. The consequences of this ‘ decoupling ’ (studiedby Stanev, 1990) are impressively demonstrated bythe anoxic conditions below 150 m. The depthreached by winter convection is governed by thestability of the stratification and, unlike the oceanbasins at the same latitudes, is very small. Thus, theupper layer is ventilated down to about 50–150 m.The newly formed cold intermediate water (CIW) isoverlaid by the seasonal thermocline. The reducedvertical exchange caused by the strong stability ofstratification shields the CIW from mixing with sur-face and deep waters (Stanev, 1990), and the coldintermediate layer (CIL) is observed as a perennial

thermic, characteristic at depths ranging from 50 m(the central basin) to 150 m (the easternmost BlackSea).

Radiotracers give valuable information for estimat-ing the speed of penetration of signals from the seasurface into the deep ocean layers. In the case of theBlack Sea, the Chernobyl accident created such asignal and made possible the evaluation of the rate ofmixing between the Mediterranean and Black Seawaters, which contributes to internal mixing in theBlack Sea (Buesseler et al., 1991). Numerical modelscan be used to test the contribution of differentmechanisms of mixing in the coastal–open seaexchange. Since the limited amount of observa-tions could not give reliable information about thebasin-wide exchange, we give some results of themodelling, though a more detailed presentation ofthe simulations will be given in the next section.

To study the exchange between surface anddeep waters, we add a new tracer—90Sr—to theDMG-MOM model with 1

4� horizontal resolution. Theparameterization of vertical mixing in the model isstability-dependent and tuned against chemical data


–1801998

180

Year

SL

A (

mm

)

1992

120

60

0

–60

–120

1993 1994 1995 1996 1997

F 2. Variability of the basin mean sea-level anomaly(SLA, mm) from the T/P data.

��

��

��

��

��

��

��

��

��

��

��

��

��

��

F 3. Amplitude of the variations in sea-level elevation(SLE, mm) from the T/P data.

250

90Sr (Bq m–3)(a)

Dep

th (

m)

40°N42°E

46°N

Longitude

(b)L

atit

ude

100

200

300

5 10 15 20

28°E 30°E 32°E 34°E 36°E 38°E 40°E

45°N

44°N

43°N

42°N

41°NMean = 17.7

Cl = 0.1

17.4

F 4. The penetration of 90Sr into the Black Seapycnocline. The results have been simulated with theDMG-MOM model with 1

4� resolution. Detailed descriptionof the model setup is given in Stanev et al. (1998). (a) 90Srvs depth for 1992. Dashed line indicates 1� deviation ofsimulated values from the basin mean. Squares correspondto observations for the same period. (b) Horizontal mixingpattern as seen in the distribution of 90S at �t=14·4 (30 May1991).

analysed by Lewis and Landing (1991). The forcingincludes fluxes of 90Sr located in the river mouths inthe north-western Black Sea, which are calculated as aproduct of river discharge times the measured con-centrations. As seen from Figure 4(a), the simulateddistribution of 90Sr in the vertical correlates wellwith the measurements. What is very important andrelevant to the present study is that it is not onlythe entrainment of Black Sea water by the sinkingMediterranean plume (Buesseler et al., 1991) thatgoverns the penetration of signals from the sea surfaceinto the deeper layers. One substantial part of thediapycnal transport occurs along the jet stream, which

is associated with the time variability and synopticoscillations. In the context of the ventilation of theshelf area, the Sevastopol eddies are of utmost impor-tance. As seen from the analyses of observations byIvanov et al. (1997), and as shown in the theoreticalstudy of Staneva and Stanev (1997), the volume ofCIW expands in the area of anticyclones. This makesthem potentially important stock elements for sub-stances transported from the shelf. The slope currentspropagate these pollutants rapidly along the basinperiphery, so that they can penetrate into the open sea


due to the diapycnal exchange. This is the case withthe Chernobyl 90Sr, which was discharged into the seaby the rivers.

The strong stratification in the Black Sea and thedifferent depth of pycnocline in the cyclonic andanticyclonic areas make the vertical profile of tracersquite noisy when plotted against depth [Figure 4(a),the squares correspond to observations in differentlocations]. Plotting the data in �t-co-ordinates usuallyreduces the dispersion created by dynamical reasons(e.g. Turgul et al., 1992) and makes possible theappearance of some fundamental features associatedwith the diapycnal mixing (Staneva et al., 1999).However, the horizontal gradients of transient tracersare usually small when plotted on �t-surfaces and arenot easily detected from observations. In this case, thesimulations can give useful supplementary informa-tion about the mixing paths or the diapycnal penetra-tion into the pycnocline. This is illustrated by thesimulated 90Sr at �t=14·4 [Figure 4(b)]. It is clearlyseen that 90Sr penetrates the ispycnal surface fromits periphery, where the highest concentrations areobserved. Equally important, and at the same timevery peculiar, is the fact that the Chernobyl 90Srpenetrates the pycnocline far from the region of itsorigin (i.e. the rivers in the north-western part of thesea) along the whole basin periphery. This demon-strates the importance of coastal circulation for thevertical/diapycnic spreading of signals from the seasurface into the interior.

Circulation on the north-western shelf

The Black Sea shelf consists of two distinct regions: avery flat area in the north, lying approximatelybetween the Cape Tarhankut and the Danube delta,with depths lower than 50 m, and a narrow belt ofabout 50 km wide, extending from the CrimeaPeninsula to the coast of Bulgaria, with depths varyingbetween 50 and 100 m. In the first (very shallow)region, the dynamics are strongly dominated by windsand dissipation, whereas in the second one, they aremuch more complex due to the interaction betweenshelf and open-sea processes.

The recent quasi-synoptic measurements carriedout under the CoMSBlack and EROS 21 projects arenot analysed in detail for the shelf area. However,salinity data [Figure 5(a)] demonstrate that thedynamics close to the Danube delta are dominated bythe river plume. A well-defined front separates theriver water from the open-sea waters. Below thepycnocline salinity reveals quite different patterns[second column of Figure 5(b)], indicating that theprocesses in the surface layers might be quite indepen-

dent from the ones in the deep layers. It is of particularinterest to show the vertical cross-sections in thesouthern part of the plume, also indicating thedecorrelation between surface and deeper waters.

The circulation on the shelf has been addressed ina number of experimental and model studies (e.g.Blatov et al., 1984; Simonov & Altman, 1991;Mikhailova & Shapiro, 1996). There is evidence tosuggest that the currents may rotate in a clockwise oranticlockwise manner depending on the wind direc-tion. Remote sensing data obtained from CZCS(Barale & Murray, 1995) show that the plume origi-nating from the Danube River often displaces to thenorth or intrudes the shelf interior. This supportsthe idea that the circulation is very changeable(e.g. Stanev & Beckers, 1999a, b), which can beeasily explained by the small mechanical inertia of ashallow-water column.

Physical processes: model–dataintercomparisons

Inventory of the physical processes affecting the coastal–open ocean exchange and their representation by theDMG-MOM and GHER model

The variability in the coastal ocean occurs over a widerange of space- and time-scales that necessitate con-sidering a wide range of phenomena, simultaneously.This is exactly the case in the Black Sea, where thescales of major processes in the coastal zone rangefrom regional to basin-wide. Direct atmospheric forc-ing and exchanges at the boundaries dominate thedynamics in the coastal zone, and both free and forcedmotions are important. The width of the continentalshelf and the characteristics of the slope area shape thegeometry of the processes.

An inventory of processes studied with theDMG-MOM and GHER models (with a shortspecification of them) is given in Table 2. We remindthe reader that some processes listed in Table 2 havealso been studied using other models. In the follow-ing, we will demonstrate the relevancy of some physi-cal processes to the coastal–open ocean exchange,using model results.

Waves

Coastal waves. Most of the modelling studies ongeneral ocean circulation focus on wind and thermo-haline currents, neglecting the short, periodic sea-surface variability by prescribing rigid lid boundaryconditions. Another large class of model studiesaddresses tidal motions, neglecting the baroclinicity.


44°N

31°E

45.5°N

Longitude

Lat

itu

de

April 1997

30°E29°E

45°N

44.5°N

12.5 13 13.5 14 14.5 15 15.5 16 16.5 17 17.5 18

31°ELongitude

April 1997

30°E29°E

0 3 6 9 12 15 18 21 24 27 30 33

6031°E

0

Longitude

Ver

tica

l cro

ss-s

ecti

on (

m)

April 1997

30°E29°E

20

40

14.5 15 15.5 16 16.5 17 17.5 18

50

30

10

Salinity Salinity Salinity

44°N

45.5°N

Lat

itu

de

April 1993

45°N

44.5°N

April 1993

60

0

Ver

tica

l cro

ss-s

ecti

on (

m)

April 1993

20

40

50

30

10

44°N

45.5°N

Lat

itu

de

July 1992

45°N

44.5°N

July 1992

60

0

Ver

tica

l cro

ss-s

ecti

on (

m)

July 1992

20

40

50

30

10

44°N

45.5°N

Lat

itu

deSeptember 1991

45°N

44.5°N

September 1991

60

0

Ver

tica

l cro

ss-s

ecti

on (

m)

September 1991

20

40

50

30

10

(S–18) × 100

(a) (b) (c)Salinity at 5 m Salinity at 40 m Cross section at 44.4°N

F 5. Salinity patterns on the north-western shelf from the CoMSBlack and EROS surveys. Horizontal plots at (a) 4 mand (b) 40 m. (c) Vertical cross-section at 44·4�N.

However, in deep baroclinic land-locked basins, boththe sea-surface oscillations and the baroclinicity areimportant. Closed boundaries provide a wave-guide

for Kelvin waves (coastal-trapped waves, in the case ofa basin with realistic bottom), and their relationshipwith other wave processes has been illustrated by


T 2. Inventory of the major physical processes and phenomena in the Black Sea studied using the DMG-MOM andGHER models

Processes/phenomena Black Sea reference Notes

WavesSurface Stanev and Beckers (1999a) subinertial oscillations with maximum

amplitudes at 1·2, 2·1, 4·4 and 6·3 h havebeen studied

Internal Stanev and Beckers (1999a) coupling between free surface andpycnocline oscillations

Basin oscillations Rachev and Stanev (1997a)Rachev and Stanev (1997b)Stanev and Staneva (2000)

western propagation caused by basinoscillations

Topographic Stanev and Rachev (1999) differentiation between basin andtopographic oscillation

Kelvin and Rossby Stanev and Beckers (1999a)Stanev and Rachev (1999)

analysis of characteristics of Kelvin waves ina small basin

Coastal trapped Stanev and Beckers (1999a) oscillations with periods exceeding inertialperiod, with the coast on their right

Front processes and eddiesUpwelling Stanev and Beckers (1999a) generation of upwelling at Cape KaliakraRim current Stanev and Staneva (2000)

Staneva and Stanev (1997)Stanev and Beckers (1999a, b)

characteristics of the rim current, transport,vertical shear of the currents

Baroclinic instabilities Rachev and Stanev (1997)Stanev and Staneva (1999)

free and forced baroclinic oscillations

Eddies (synoptic, quasi-permanent,basin-scale)

Gregoire (1998)Stanev and Staneva (2000)

the impact of eddies on the circulation

Small-scale processesBoundary layers and mixing Stanev et al. (1997) parameterization in models and their impact

on simulationsEkman transport Simeonov et al. (1997)

Staneva et al. (1995, 1998)surface and bottom plumes, gravity currents,ocean–atmosphere exchange

Breaking waves and turbulence Stanev and Beckers (1999a, b) mixing in intermediate layers caused bybreaking waves

Currents and water masses Staneva and Stanev (1997)Stanev and Beckers (1999a, b)Stanev et al. (1998)Staneva et al. (1999)

wind- and buoyancy-driven currents, andtheir relationships to water-mass formation

Stanev and Beckers (1999a) in the case of the BlackSea. The active free surface in the GHER modelmakes it possible to simulate short, periodic baro-tropic oscillations. Their periods range between 1 and10 h. Coupling between barotropic and baroclinicoscillations results in baroclinic wave excitation andwave shedding towards the open sea. This is shownin time vs distance from the coast dependency[Figure 6(a)]. It is clear that the coastal-trapped wavesare simulated in a narrow zone over the continentalslope. The wave speed can be measured by the slopeof the thick line in Figure 6(a) connecting equalphases at different distances from the coast (givingapproximately 0·5–1 m s�1). The characteristics ofthe oscillations change with time and, as indicated inFigure 6(a) at location 14, high salinity water from

the basin interior intrudes the coastal zone. Almostsimultaneously (after day 22–23), a decrease of theamplitude of oscillations is observed.

The appearance of internal waves in the Black Sea isvery specific and they are well traced in the time vsdepth plot [Figure 6(b)]. Our simulations reveal twolayers (surface and deep) with large stratifications andan intermediate homogeneous layer (the CIL). Theinternal wave oscillations in the intermediate layer, asseen in the temperature field, are insignificant. Thisproves that the CIL acts as a thermic buffer, not onlyat climatic time scales, but also at high frequencies.However, the oscillations of the interface (permanenthalocline) interacting with the shelf/continental slopemight become a key element in the mixing process.Breaking internal waves [or transformations of the

–25025

–150

Days from perturbation

(b)

Dep

th (

m)

20 21 22 23 24

GHER

Temperature at point 11 (°C)

2124

24

Point

(a)

Day

s fr

om p

ertu

rbat

ion

9 14 19

GHER

Temperature (°C)

7.43

7.437.43

7.55 7.55

7.80

7.80

7.80

7.80

8.05

7.80

7.55

7.43

7.43

7.67

7.43 7.43 7.437.55

7.67 7.67

7.80

7.80

7.67

7.55

7.55

7.43

7.43

7.67

7.80

7.80

7.80

7.43

7.927.

927.928.05 8.17

8.17

7.92

7.927.92

8.17

8.17

8.17

8.05

8.05

8.298.298.29 8.428.42

8.42

8.178.42

8.54

8.548.54

8.42

8.42

8.17

8.17

8.178.42

8.05

8.05

7.80

7.92

8.05

8.17

8.29

8.29

8.298.29

8.17

8.05

8.54

8.54

8.42

8.42

8.668.66

8.66

8.66

8.54

8.42

8.548.66

8.42

8.79

8.79

8.79

8.79

8.91

8.91

8.918.918.91

8.79

8.798.91

8.79

8.91

8.91

8.918.66

8.798.91

8.798.79

8.79

8.91

23

22

8.18

8.18

8.18

8.18

8.18

8.18

8.18

8.18

8.18

8.18

8.18

8.26

8.268.26

8.26

8.26

8.26

8.10

8.10

8.10

8.10

8.10

8.10

8.10

8.10

8.10

8.01

8.01

8.01

8.01

8.01

8.01

8.01

7.93

7.93

7.93

7.93

7.93

7.857.77

7.85

7.77

7.77

7.77

7.85

7.85

7.77

7.85

7.85

7.85

7.77

7.68

7.68

7.68

7.68

7.68

7.68

7.68

7.60

7.60

7.60

7.60

7.60

7.60

7.52

7.52

7.52

7.52

7.52

7.43

7.35

7.35 7.

437.

43

7.35

7.35 7.

437.

43

7.35

7.35 7.

43

7.30

F 6. (a) Time vs distance from the coast plot of temperature simulated with the GHER model with 15-km horizontalresolution at 150 m, 43�27�N. The abscise axis gives the distance from the coast (the number on the axis times 15 km).(b) Time vs depth plot of temperature at location 11 simulated with the GHER model.


signal in the coastal region, as shown in Figure 6(a)]might provide mixing in the halocline and affect thethickness of the CIL. This leads to changes in thestratification of the intermediate layer and depends onthe amount of energy provided by free-surface oscil-lation. These physical processes currently remain un-explored for the Black Sea, but might be responsiblefor the control of the exchange between upper andintermediate layers.

The oscillations of the halocline on differenttime-scales could be of particular importance for thephysical control of the biological systems. We canspeculate that mixing (affecting different chemical andbiological compounds) will also be governed by wavebreaking. This might explain the importance of lateralmixing in the Black Sea, providing the basin interiorwith strong signals generated in the coastal areas.

Basin waves. While in the world ocean the Rossbywaves are responsible for the western propagation ofsignals, the wave reflection in enclosed basins creates amore specific regime, associated with the basin oscil-lations (Rachev & Stanev, 1997a; Demyshev et al.,1996; Stanev & Rachev, 1999), that could exist inbarotropic as well as in baroclinic fluids. What makesthe Black Sea an interesting case is that the Rossbyradius of deformation (several tenth kilometres) iscomparable with some fundamental basin scales(central narrow section, the narrow easternmost partof the basin). The simulations with the DMG-MOMdemonstrate that the Rossby waves emerge from theeasternmost area and propagate to the west with aphase speed of 2 cm s�1 [Figure 7(b)]. They domi-nate the dynamics when the model is forced withstationary boundary conditions (Rachev & Stanev,1997a), as well as when the forcing changes withtime (Stanev & Staneva, 2000), as is the case inFigure 7(a). The meridional wind-stress componentand the simulated meridional velocity at the sea sur-face show quite different patterns. Wind maximaoccur in winter and are more pronounced in thewestern Black Sea. The lack of slope in the windcontours demonstrates that there is no substantialsignal propagating in the zonal direction (we excludefrom this analysis the synoptic processes in the atmos-phere having very short time-scales). There is nosubstantial correlation between the two types of data(forcing and response), demonstrating that the west-ward propagation is rather an appearance of naturaloscillations. Thus, the atmospheric variability pro-vides perturbations for the ocean system, but does notshape the response. To give an idea of the consistencyof simulations with observations, Figure 7(c) showsthat the western propagation is also well pronounced

in the altimeter signal, although the phase speed isslightly lower.

The analysis of model simulations demonstratesthat the basin oscillations induce strong changes in thedepth of pycnocline (Rachev & Stanev, 1997a). Theycould also enhance the amplitude of the rim currentand affect the exchange between anticyclonic andcyclonic areas. Since this process is extremely sensitiveto complicated bottom relief, we could speculate thatthis type of wave could affect the anticyclonic eddiesbetween the main gyre and the coast. Some of themost energetic eddies of this type originate in the areaof the Crimea Peninsula (Figure 1), and their inter-action with the shelf could substantially affect themixing (including mixing of chemical and biologicalmatter) and the resulting water-mass formation.

Frontal processes and eddies

The Black Sea upwelling. In this section, we addresssome processes that are crucially important for thecirculation and synoptic variability of any oceanbasin using simulation results produced by theDMG-MOM and GHER models. In basins withstrong vertical stratification, the contrasts betweenthermohaline characteristics in coastal and open-searegions are usually well pronounced, providing thatthe mechanic forcing sufficiently maintains the slopeof the pycnocline steep. Under such conditions, theinstabilities associated with frontal oscillations and thedirect mechanic forcing from the atmosphere triggersan intense upwelling. It has been found that there aresome areas (Cape Kaliakra, the southern coast and thearea west of the Crimea Peninsula, Figure 8) wherethe upwelling is quasi-permanent (Stanev et al., 1988;Sur et al., 1994; Blatov & Ivanov, 1992; O} szoy &U} nluata, 1998; Gavarkievicz et al., 1999). In thepresence of strong changes of coastal line or topogra-phy (e.g. Cape Kaliakra), the oscillations of thepycnocline may amplify the transport of CIW into thesurface layers. What has not been addressed in anyfurther detail in previous studies is the specific verticalstratification in the Black Sea and the characteristicstimes of processes acting in the horizontal and in thevertical. While it takes about 1 year for the currents tomake one loop along the coast (Stanev et al., 1998),the vertical penetration of the signals is much slower,as shown by observations and modelling of thepenetration of radiotracers (see ‘ Penetration of thesignals from the sea surface into the pycnocline ’).Since the horizontal–isopycnal mixing is much largerthan the vertical–diapycnal mixing, the water proper-ties of each region tend to homogenize on isopycnals.Thus, the properties of coastal waters dominate in the


��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

��

!�"��

�#�$�#�� %��&�' ��

��

��

��

��

��

��

��

��

�

�

�

�

�

�

�

��

��

��

��

��

��

��

��

(��' "�''

��

��

��

��

��

��

��

�)�

�)�

�)�

�)�

�)�

�)�

��

��

��

�&�#%�*�'��+#�$�#��&&�


��

��

��

��

��

��

��

��

�)��)��)��)�� )� �)� �)� �)�

��

��

�)� �)��)�

,�" ��#�$�#�� %��&�'��

F 8. Monthly mean vertical velocities in March 1984at 7·5 m. Dark tones mark upwelling areas. Simulations arecarried out with the DMG-MOM model with 1

12� resolution.

F 7. Time–longitude diagrams of (a) wind stress magnitude and (b) meridional velocity at the sea surface, 43.5�N. Theslope of the contours gives a measure of the speed of westward propagation. The data used to plot this diagram are simulatedby the DMG-MOM model with 1

12� resolution. The model is forced with atmospheric analysis data from UKMO. (c) Sea-levelanomalies (mm) from the T/P data at the same latitude.

surface layers, while deep-water characteristicsdominate in the interior basin (Staneva et al., 1999;Stanev & Staneva, 2000). Under such conditions, onecould represent the Black Sea as being composed oftwo dynamically different sub-basins: coastal andopen sea. A similar division has already been sug-gested by Bulgakov and Korotaev (1984). However,what has been realized in recent years is that thecirculation in the coastal (anticyclonic) area isdominated by synoptic eddies, which requires theaddressing of the exchange between the two areasfrom the viewpoint of non-stationary dynamics.

Accordingly, the physical and biological character-istics of the coastal/interior basin tend to upper/deepsea characteristics. This ‘ regionalization ’ is consistentwith the general cyclonic circulation and the associ-ated upwelling in the basin interior. The two branchesof the vertical circulation communicate by theexchange in the slope area, the latter providing asubstantial part of the exchange in the Black Sea [seeFigure 4(b)].

The upwelling has not only important conse-quences for the vertical exchange of physical proper-ties in the upper layer, it also affects the biologicalproductivity. The basin-wide upwelling in the BlackSea interior supplies the intermediate layer withhydrogen sulphide-rich waters, as shown from obser-vations (Dobrujanskaya, 1967). Deep waters (rich in

ammonium) could also have a key significance for thecharacteristics of the trophic chains in the upwellingareas.

Systematic studies on the local aspects of coastalupwelling are still sparse (e.g. Blatov & Ivanov, 1992;Stanev et al., 1988; Oguz et al., 1992; Sur et al.,1994; Kosnyrev et al., 1996; Vlasenko et al., 1996;Gavarkievicz et al., 1999). The upward transport ofwaters from the CIL into the surface layer opposes thegeneral trend of anticyclonic circulation in the coastalzone bringing surface waters into the deeper layers.Since the temperature of the CIW is lower than thesea surface temperature in summer, this cold-watermass clearly traces the upwelling region by givingsignals in the AVHRR data (mostly with synoptic andmesoscale characteristics) in the warm part of theyear.

The generation of coastal upwelling in thewestern Black Sea has been shown by the model ofStanev and Beckers (1999a, b), but there are noclear estimates about the contribution of this processin the mixing of Black Sea waters. The localizedappearance of upwelling necessitates the very fineresolution of numerical models (Demirov, 1994), andwe will show in the next subsection an example ofsuch simulations.

Basin-scale/mesoscale circulation. The rim current is oneof the most interesting physical phenomena in theBlack Sea dynamics. It is very narrow, which is due toextremely strong density contrasts in the vertical, aswell as to very narrow continental slope. Realisticmodelling of this current requires a fine spatial reso-lution. This is shown in Figure 9, which comparessimulations of the basin-wide circulation carried outwith the GHER model with 15- and 5-km horizontalresolution at the end of May. One can clearly see thatalthough the general pattern of circulation does notdrastically differ in the two simulations, the eddy-resolving simulations are dominated by meanders,eddies, filaments and dipole structures with scales thatare subgrid for the coarse model. The comparison ofthe eddy scales with the ones in the observed data (seeFigure 9 and Figure 1, in addition to Oguz et al.,1993, 1994; O} zsoy & U} nluata, 1998) demonstratesthat a 5-km resolution resolves most of the importantmesoscale features.


40.705°

46.705°Velocity scale:

0.07 m s–1

Depth: 10 m

(b)

Lat

itu

de (

N)

27.272°Longitude (E)

40.75°

45.75°

Velocity scale:0.07 m s–1

Depth: 10 m

(a)

Lat

itu

de (

N)

27.334°Longitude (E)

45.705°

44.705°

43.705°

42.705°

41.705°

29.272° 31.272° 33.272° 35.272° 37.272° 39.272° 41.272°

44.75°

43.75°

42.75°

41.75°

29.334° 31.334° 33.334° 35.334° 37.334° 39.334° 41.334°

F 9. Horizontal currents at 10 m at the end of May simulated by the GHER model and forced with monthly climaticdata. (a) Simulations with horizontal resolution of 15 km. (b) Simulations with horizontal resolution of 5 km.

The baroclinic instabilities together with basinoscillations and mesoscale eddies, give a very compli-cated picture of the exchange occurring between anti-cyclonic and cyclonic areas, depending on thetransition of the circulation between different domi-nating states (intense winter- and less intensesummer-state, Stanev & Staneva, 2000). The synopticeddies present a key element in the energy exchange.The key point here is that the two states of circulationare characterized by different slopes of the pycnoclinein the area of the rim current, and that the diapycnalexchange between coastal and open waters mightbe dependent on this slope, so that any mesoscalefeatures could affect this mixing.

Subgrid-scale processesSmall-scale processes are of the utmost importanceto the behaviour of geophysical fluids, and some ofthem are listed in the third part of Table 2. Most ofthese processes are not resolved by the basin-wide numerical models, therefore they have tobe parameterized. For details on the differentparameterizations, and their impact on the modelperformance, we recommend the paper by Stanevet al. (1997). Here, however, we only mention that theestimation of the impact of horizontal mixing is ofparamount significance when addressing the exchangebetween coastal sea and open ocean. One couldexpect that increasing the coefficient of horizontal


6.5Nov

8.0

Time (months)

(b)

Tem

pera

ture

(°C

)

Jan

7.5

7.0

SepJulMayMar

6.5Nov

8.0

Time (months)

(a)

Tem

pera

ture

(°C

)Jan

7.5

7.0

SepJulMayMar

F 10. Time-series of basin-averaged temperaturesimulated by DMG-MOM model with resolutions 1

12� (fullline) and 1

4� (dashed line) at (a) 55 m and (b) 75 m.

mixing in the model would result in increasing theexchange between the coastal and open-sea areas.However, as our simulations demonstrate, this is notalways the case. Reducing the horizontal diffusion canresult in unrealistically large slopes of the halocline,followed by an increase of eddy activity and enhancedcross-gyre mixing (Stanev et al., 1997). Since the rimcurrent encompasses the entire basin, one could con-clude that increasing the instability of the jet currentwould increase the diapycnal mixing along the entireslope area, partly compensated by a decrease in themixing (small diffusion coefficient) in the shelf break,inhibiting the uptake of cold water from the coastalzone. In such situations, slope currents take control ofcross-shelf mixing (Staneva & Stanev, 1997).

The interrelationship between horizontal and verti-cal exchange is demonstrated below, in the simula-tions with the DMG-MOM model; by analysing thedifferences between mixing properties simulated bycoarse- and fine-resolution models (Table 1). Theatmospheric forcing is identical in both models,except that the high-resolution model admits meso-scale structures in the sea surface temperature field,and thus in the interactive heat fluxes. Since meso-scale heat fluxes generally enhance water formations,we would expect a change in the CIL average tem-perature in a high-resolution model compared to oneof a coarse resolution. However, the more realisticresolution of the physical processes in the fine-resolution model did not result in any substantialdifference in the annually averaged temperatures inthe CIL, only in large amplitudes of the seasonalsignal (Figure 10). This comparison suggests that, byincreasing the eddy activity along the rim current(which is also the usual situation in the real basin), therate of ventilation of the coastal zone increases. Thishas far-reaching consequences. By exchanging watersbetween the two areas (often diapycnally), the modeltends to change the vertical stratification. It is clearlyseen in Figure 10(a,b) that the vertical temperaturegradient is smaller at the end of winter and greater bythe end of autumn in the fine-resolution model. So,the increase in the amplitude of seasonal temperaturesshould affect the biology: directly, by influencing therate of biological processes, and indirectly, by actingon the depth of the mixed layer.

Currents and water masses

Currents. The simulations of the DMG-MOM surfacecurrents have magnitudes of about 20–40 cm s�1,the total horizontal mass transport reaches severalsverdrups, with a larger part of it being located abovethe pycnocline. This transport is approximately two-

fold larger under the fine-resolution model than underthe coarse one. One could ask whether this drasticdifference has a pronounced impact on the ventilationof coastal regions and the intermediate layer, orwhether a compensation between eddy and meantransport occurs, as in some ocean models (Cox,1985; Bryan, 1986). Since the vertical circulation inthe Black Sea is much weaker than the horizontal one,and large changes in the horizontal circulation are notaccompanied by large (in absolute values) changes in


��

��

�� )�� )�� )�� )�� )�� )�� )��

��

��

��

��

��

��

��

��

��

- �"�� "$�#��&��'��

.�&&�"

��

��

�� )�� )�� )�� )�� )�� )�� )��

��

��

��

��

��

��

��

��

��

- �"�� "$�#��&��'��

�� &

��

��

�� )�� )�� )�� )�� )�� )�� )��

��

��

��

��

��

��

��

��

��

- �"�� "$�#��&��'��

(� �"

��

��

�� )�� )�� )�� )�� )�� )�� )��

��

��

��

��

��

��

��

��

��

- �"�� "$�#��&��'��

./"��

F 11. Seasonal mean zonally-averaged vertical mass transport simulated by the DMG-MOM model with 112� resolution,

forced with UKMO atmospheric analysis data.

the vertical circulation, the intensification of thehorizontal circulation does not automatically result inpronounced vertical overturning (see Figure 11 andalso Stanev, 1990; Rachev & Stanev, 1997a) and doesnot have a clear impact on the water–mass forma-tion. The results of simulations are consistent with

observations and simple theoretical considerations(Bulgakov et al., 1996). The simulated seasonal meanhorizontal mass transport ranges in values: 6, 8, 4and 5�106 m3 s�1 in winter, spring, summerand autumn, respectively. The vertical overturningis much smaller, amounting in the upper layer to


6�104 m3 s�1 in winter, 4�104 m3 s�1 in spring,4�104 m3 s�1 in summer and 5�104 m3 s�1 inautumn. The corresponding rates of vertical transport

in the deep layer are 16�104 m3 s�1 in winter,9�104 m3 s�1 in spring, 10�104 m3 s�1 in summerand 13�104 m3 s�1 in autumn. Thus, we couldconclude that the vertical circulation is: (1) muchweaker than the horizontal one (about two orders ofmagnitude, which is typical for stagnant basins;Stanev, 1990), (2) about one order of magnitudestronger than the river discharge or the transportthrough the Straits of Bosphorus (�104 m3 s�1,O} zsoy & U} nluata, 1999, and (3) the patterns are moreirregular and variable compared to those of horizontaltransport. The above results support the idea that thehorizontal processes in the Black Sea are much moreactive than the vertical ones, which is also due to thestagnant conditions. The evidence that the verticalcirculation is about one order of magnitude largerthan the straits inflow supports the independentresults based on the analysis of radionuclide penetra-tion into the pycnocline, and simulations on theMediterranean plume in the Black Sea, giving a valueof about 10 for the rate of entrainment of Black Seawater by the Mediterranean plume (Buesseler et al.,1991; Simeonov et al., 1997; Staneva et al., 1999).The correlation of our estimates with this fundamen-tal number, obtained independently, gives a credibilityfor the simulated ventilation of intermediate and deeplayers.

In the context of the present study, the area west ofthe Crimean Peninsula is very important as a key areawhere the rim current attacks the continental slope.The anticyclonic eddies simulated in this regionpropagate westward (Figure 12) and shape theexchange between the shelf and the open sea.These eddies are also found in the survey data[Figure 1(a,c)], as well as in the T/P data [Figure 1(d)and Figure 3]. Their impact on the shelf–open oceanexchange is associated with the large thermal capacityof anticyclones caused by the deeper position of thehalocline. This has at least two important conse-quences for the physical and biological system relatedto the deepening of the pycnocline in this area: (1) anincrease in the volume of the biologically active layerand (2) mixing between coastal and open-sea watersmight become more efficient.

Conclusions

We have demonstrated that the altimeter signal isreliable for analysis of the annual variability of circu-lation. The large amount of such data enables aprecise mapping of basin-wide dynamics, as well as ofthe variability in some dynamically important areas(coastal anticyclones and the well-known areas ofthe Batumi and Sevastopol eddies, Figure 3). The

43°N34°E

46.5°N

Longitude

(c)

Lat

itu

de

28°E

30

29°E 30°E 31°E 32°E 33°E

46°N

45.5°N

45°N

44.5°N

44°N

43.5°N

43°N34°E

46.5°N

Longitude

(b)

Lat

itu

de

28°E

30

29°E 30°E 31°E 32°E 33°E

46°N

45.5°N

45°N

44.5°N

44°N

43.5°N

43°N34°E

46.5°N

Longitude

(a)

Lat

itu

de

28°E

40

29°E 30°E 31°E 32°E 33°E

46°N

45.5°N

45°N

44.5°N

44°N

43.5°N

F 12. Snapshots of velocities at 2·5 m in the north-western Black Sea. The plots are based on simulations withthe DMG-MOM model with 1

12� resolution, forced withUKMO atmospheric analysis data. (a) 20 January 1993;(b) 29 June 1993; (c) 19 September 1993.


multiple time-scales were demonstrated over a widerange of frequencies (from several hours, governinginternal gravity oscillations and convection, to inter-annual). Such an illustration is the coupling betweenbarotropic and baroclinic oscillations. This process isof significance for the dynamics in the coastal zone,and one important fact has been given here associatedwith the baroclinic wave excitation and wave sheddingtowards the open sea with a phase speed of 0·5–1 m s�1. A fundamental characteristic of the BlackSea, intimately related to the coastal–open seaexchange, is extreme vertical stratification. It preventsvertical mixing, in which case the oscillations of theinterface (permanent halocline) interacting with theshelf/continental slope are a key process. The trans-formations of the wave signal in the coastal region(Figure 6) provide mixing in the halocline that mightaffect the thickness of the CIL. This explains, at leastpartially, the importance of lateral mixing in the BlackSea, providing the basin interior with strong signalsgenerated in the coastal areas. Unfortunately, studiesin this field are almost non-existent for the Black Sea,but what is already known from ocean studies is thatthis is a potentially very important area for exploringdifferent scenarios regarding the transport and trans-formation of physical and biological matter on theshelf break.

What has not yet been addressed in enough detail inthe context of horizontal mixing is the interactionbetween the vertical stratification and the processesacting in the horizontal. It takes about 1 year for thecurrents to make a loop along the coast; however, thevertical penetration of the signals is much slower.Under such conditions, the horizontal–isopycnicalmixing is much larger than the vertical–diapycnalmixing, and the water properties tend to align to theisopycnals. The latter exhibit a large slope over thenarrow continental slope and split the Black Sea intotwo dynamically distinct areas: coastal and openocean, where water properties are dominated bysurface- and deep-water characteristics, respectively.What has not yet been analysed in the current modelsis the extent to which the mixing parameterizations,aligned along the model co-ordinates, are applicablein areas of sharp slopes (even under very fine resolu-tion). More elaborate parameterizations (Gent &McWilliams, 1990; Griffies et al., 1998) have to befurther applied in order to reach a better consistencybetween the models and the real mixing processes.Developing new mixing parameterizations in extremeareas such as the Black Sea is another challenging taskfor the future. Since the vertical overturning (i.e. ameasure of internal mixing) appears to be about oneorder of magnitude stronger than the river discharge

or the transport through the Straits of Bosphorus(this factor is in agreement with the estimates fromobservations of Buesseler et al., 1991), it is of funda-mental interest to test the sensitivity of this estimateto different parameterizations and water balancescenarios.

Among the interesting dynamical features deservingfuture interest, and in particular more profoundquantification, are the two branches of vertical circu-lation (i.e. upwelling in the interior and downwellingin the coastal zone) communicating by exchangingwater and other properties in the slope area. Theposition, slope and variability of the pycnocline in thecoastal area might have important consequences notonly for the physical system, but also, by increasingthe volume of the surface layer, the pelagic system.What has now become clear is that the ventilation ofthe coastal zone is controlled by the eddy activityalong the rim current, which has far-reaching conse-quences. By exchanging waters between the two areas(often diapycnally), the model tends to change thevertical stratification and the amplitude of the seasonalsignal in the intermediate layer. As mentionedearlier, these exchange processes have to be furtherinvestigated with models that have more elaborateparameterizations of the mixing.

Acknowledgements

The authors thank M. H. Calvex for altimeter dataprocessing and V. Belokopytov for providing us withgridded climatic data. The help of E. Cholakov, whoplotted some figures, is also acknowledged. Thanksare also due to G. Korotaev for useful comments onthe manuscript. Data from the CoMSBlack surveyshave been prepared in the framework of the NATOBlack Sea project and been made available throughBlack Sea Environment Internet Node (BSEIN). Thiswork has been supported by an EROS 21 researchcontract with CEC grant IC20-CT96-0065. Weacknowledge the help of the UKMO for makingavailable meteorological analysis data under theresearch contract with CEC EV5V-CT92 0121,supplementary agreement CIPD CT93 0016. This ispublication No. 194 of the EU-ELOISE initiative.

References

Altman, E. N., Gertman, I. F. & Golubeva, Z. A. 1987 Climato-logical fields of salinity and temperature in the Black Sea. Report.State Oceanography Institute, Sevastopol, Ukraine, 109 pp.(in Russian).

AVISO 1996 User Handbook: Merged TOPEX/POSEIDON Products.AVI-NT-02-101-CN, Edition 3.0.


Barale, V. & Murray, C. N. 1995 The surface color field of enclosedmarine basins: Pigment patterns of the Black Sea. Remote SensingReviews 12, 61–82.

Beckers, J. M. 1991 Application of a 3D model to the westernMediterranean. J. Mar. Sys. 1, 315–332.

Beckers, J. M., Gregoire, M., Nihoul, J. C. J., Stanev, E., Staneva,J. & Lancelot, C. 2002 Modelling the Danube-influencednorth-western continental shelf of the Black Sea. I: Hydro-dynamical processes simulated by 3-D and box models. Estuarine,Coastal and Shelf Science 54, 453–472.

Blatov, A. S. & Ivanov, V. 1992 Hydrology and Hydrodynamics ofthe Black Sea Shelf Zone. Naukova Dumka, Kiev, 242 pp.(in Russian).

Blatov, A. S., Bulgakov, N. P., Ivanov, V. A., Kosarev, A. N. &Tujilkin, V. S. 1984 Variability of Hydrophysical Fields in the BlackSea. Gidrometeoizdat, Leningrad, 240 pp. (in Russian).

Bryan, K. 1986 Poleward buoyancy transport in the ocean andmesoscale eddies. Journal of Physical Oceanography 16, 927–933.

Buesseler, K. O., Livingston, H. D. & Casso, S. A. 1991 Mixingbetween oxic and anoxic waters of the Black Sea as traced byChernobyl cesium isotopes. Deep-Sea Research 38, S725–S745.

Bulgakov, S. V. & Korotaev, G. K. 1984 Possible mechanism ofstationary circulation of the Black Sea. In Integrated Investigationsof the Black Sea. Marine Hydrophysical Institute, Sevastopol,pp. 32–40 (in Russian).

Bulgakov, S. V., Korotaev, G. K. & Whitehead, J. A. 1996 The roleof buoyancy fluxes in the formation of a large scale circulationand stratification of sea water. Part 1 and 2. Atmospheric andOcean Physics 32, 548–564 (in Russian).

Cox, M. 1985 An eddy resolving numerical model of the ventilatedthermocline. Journal of Physical Oceanography 15, 1312–1324.

Demirov, E. K. 1994 Numerical modelling of the Black Seaeigler-oscillations on a curvilinear boundary fitted coordinatesystem. Dynamics of Atmospheres and Oceans 21, 83–103.

Demyshev, S. G., Korotaev, G. K. & Kuftarkov, A. Y. 1996Adjustment of the Black Sea geophysical fields to a newnumerical model. Atmos. Ocean Phys. 32, N5.

Dobrujanskaya, M. A. 1967 Regularities in the space distribution ofoxygen in the Black Sea. In Oceanographic Research in the BlackSea. Naukova Dumka Press, Kiev, pp. 154–169.

Gavarkievicz, G., Korotaev, G. K., Stanichny, S. V., Repetin, L. N.& Soloviev, D. M. 1999 Synoptic upwelling and cross-shelftransport processes along the Crimea coast of the Black Sea.Continental Shelf Research 19, 977–1005.

Gent & McWilliams 1990 Isopycnal mixing in ocean circulationmodels. Journal of Physical Oceanography 20, 150–155.

Gregoire, M. 1998 Modelisation 3D couplee physique-ecosystemede la mer Noire. Ph.D. Thesis. Liege University, 280 pp.

Griffies, S. M., Gnanadesikan, A., Pacanowski, R. C., Larichev,V. C., Dukowicz, J. K. & Smith, R. D. 1998 Isoneutral diffusionin z-coordinate ocean model. Journal of Physical Oceanography 28,805–830.

Ivanov, L. I., Besiktepe, S. & O} zsoy, E. 1997 The Black Sea coldintermediate layer. In Sensitivity to Change: Black Sea, Baltic Seaand North Sea (O} zsoy, E. & Mikaelyan, A., eds). KluwerAcademic Publishers, pp. 253–266.

Korotaev, G. K., Saenko, V. A., Koblinsky, C. J. & Knysh, V. V.1998 Study of the Black Sea circulation using Topex-Poseidenaltimetry. Earth Research from Space 3, 3–17.

Kosnyrev, V. K., Mikhailova, E. N. & Shapiro, N. B. 1996Numerical modelling of upwelling in the northwestern Black Sea.Meteorologia i Gidroloia 12, 65–72.

Lancelot, C. & Egorov, V. 1997 EROS 21, Danube–Black Seacruise. Cruise Report.

Lancelot, C., Staneva, J., Van Eeckhout, D., Beckers, J.-M. &Stanev, E. 2002 Modelling the Danube-influenced north-westerncontinental shelf of the Black Sea. II: Ecosystem response tochanges in nutrient delivery by the Danube River after itsdamming in 1972. Estuarine, Coastal and Shelf Science 54,
473–499.
Le Traon, P. Y., Nadal, F. & Ducet, N. 1998 An improvedmapping method of multi-satellite altimeter data. J. Atm. Ocean.Techn. 25, 522–534.

Lewis, B. L. & Landing, W. N. 1991 Manganese and iron in theBlack Sea. Deep-Sea Research 38, S773–S805.

Mikhailova, E. N. & Shapiro, N. B. 1996 Modelling the transportand transformation of river waters on the northwestern shelf andin the deep Black Sea. Marine Hydrophys. Journal 3, 30–40(in Russian).

Nihoul, J. C. J., Deleersnijder, E. & Djenidi, S. 1989 Modelling thegeneral circulation of shelf seas by 3D k-� models. Earth ScienceReviews 26, 163–189.

Oguz, T., La-Violette, P. & Unluata, U. 1992 Upper layer circula-tion of the southern Black Sea: Its variability as inferred fromhydrographic and satellite observations. J. Geoph. Res. 97 (C8),12569–12584.

Oguz, T., Malanotte-Rizzoly, P. & Aubrey, D. 1995 Wind andthermohaline circulation of the Black Sea driven by yearly meanclimatological forcing. J. Geoph. Res. 100 (C4), 6846–6865.

Oguz, T., Latun, V. S., Latif, M. A., Vladimirov, V. V., Sur, H. I.,Markov, A. A., O} zsoy, E., Kotovshchikov, B. B., Eremeev, V. V.& U} nluata, U} . 1993 Circulation in the surface and intermediatelayers of the Black Sea. Deep-Sea Research 40, 1597–1612.

Oguz, T., Aubrey, D., Latun, V., Demirov, E., Kolesnikov, L., Sur,H., Diaconu, V., Besiktepe, S., Duman, M., Limeburner, R. &Eremeev, V. 1994 Mesoscale circulation and thermohaline struc-ture of the Black Sea during HydroBlack 91. Deep-Sea Research41, 603–628.

O} zsoy, E. & U} nluata, U. 1998 The Black Sea. In The Sea Vol. 11(Robinson, A. & Brink, K., eds). John Wiley & Sons, Inc.,pp. 889–914.

Pacanowski, R. C., Dixon, K. & Rosati, A. 1991 The GFDLmodular ocean model users guide, version 1.0. GFDL OceanGroup Technical Report, No. 2, 176 pp.

Rachev, N. H. & Stanev, E. V. 1997a Eddy processes in semi-enclosed seas. A case study for the Black Sea. Journal of PhysicalOceanography 27, 1581–1601.

Rachev, N. H. & Stanev, E. V. 1997b Eddy dynamics controlled bybasin scale, coastline and topography. In Sensitivity to Change:Black Sea, Baltic Sea and North Sea, NATO ASI Series, Vol. 27(Ozsoy, E. & Mikaelyan, A., eds). Kluwer Academic Publishers,pp. 341–364.

Simeonov, J., Stanev, E., Backhaus, J., Jungclaus, J. & Roussenov,V. 1997 Heat and salt intrusions in the picnocline from sinkingplumes. Test case for the entrainment in the Black Sea. InSensitivity to Change: Black Sea, Baltic Sea and North Sea, NATOASI Series, Vol. 27 (O} zsoy, E. & Mikaelyan, A., eds). KluwerAcademic Publishers, pp. 417–438.

Simonov, A. I. & Altman, E. N. 1991 Hydrometeorology andHydrochemistry of the USSR seas. Vol. IV. The Black Sea.Gidrometeoizdat, Moscow, 430 pp.

Sorkina, A. I. (ed.) 1974 Reference Book on the Black Sea Climate.Gidrometeoizdat, Moscow, 406 pp. (in Russian).

Stanev, E. V. 1990 On the mechanisms of the Black Sea circulation.Earth-Science Review 28, 285–319.

Stanev, E. V. & Beckers, J. M. 1999a Barotropic and baroclinicoscillations in strongly stratified ocean basins. Numerical studyfor the Black Sea. J. Mar. Sys. 19, 65–112.

Stanev, E. V. & Beckers, J. M. 1999b Numerical simulations ofseasonal and interannual variability of the Black Sea thermo-haline circulation. J. Mar. Sys. 22, 241–267.

Stanev, E. V. & Rachev, N. H. 1999 Numerical study on theplanetary Rossby waves in the Black Sea. J. Mar. Sys. 21,283–306.

Stanev, E. V. & Staneva, J. V. 2000 The impact of the barocliniceddies and basin oscillations on the transitions between differentquasi-stable states of the Black Sea circulation. J. Mar. Sys. 24,1–2, 3–26.

Stanev, E. V., Staneva, J. V. & Roussenov, V. M. 1997 On theBlack Sea water mass formation. Model sensitivity study to


atmospheric forcing and parameterization of some physicalprocesses. J. Mar. Sys. 13, 245–272.

Stanev, E. V., Staneva, J. V. & Gospodinov, I. G. 1998 Numericaltracer model of surface and intermediate water formation in theBlack Sea. In Ecosystem Modelling as a Management Tool for theBlack Sea Vol. 2 (Ivanov, L. & Oguz, T., eds). Kluwer AcademicPublishers, Dordrecht, The Netherlands, pp. 197–219.

Stanev, E. V., Trukhchev, D. I. & Roussenov, V. M. 1988 TheBlack Sea Circulation and Numerical Modelling of the Black SeaCurrents. Sofia University Press, Sofia, 240 pp. (in Russian).

Staneva, J. V. & Stanev, E. V. 1997 Cold intermediate watermass formation in the Black Sea. Analyses on numerical modelsimulations. In Sensitivity to Change: Black Sea, Baltic Sea andNorth Sea (O} zsoy, E. & Mikaelyan, A., eds). NATO Series,Kluwer Academic Publishers, Dordrecht, The Netherlands,pp. 375–393.

Staneva, J. V. & Stanev, E. V. 1998 Oceanic response toatmospheric forcing derived from different climatic data sets.Intercomparison study for the Black Sea. Oceanol. Acta 21,393–417.

Staneva, J. V., Stanev, E. V. & Rachev, N. H. 1995 Heat balanceestimates using atmospheric analysis data. A case study for theBlack Sea. J. Geoph. Res. 100, C9, 18581–18596.

Staneva, J. V., Buesseler, K. O., Stanev, E. V. & Livingston,H. D. 1999 Application of radiotracers to study Black Seacirculation: Validation of numerical simulations against observedweapon testing and Chernobyl 137Cs data. J. Geoph. Res. 104,11099–11114.

Sur, H. I., O} zsoy, E. & U} nluata, U} . 1994 Boundary currentinstabilities, upwelling, shelf mixing and eutrophication processesin the Black Sea. Progress in Oceanography 249–302.

Turgul, S., Basturk, O., Saidam, C. & Yilmaz, A. 1992 Changes inthe hydrochemistry of the Black Sea inferred from water densityprofiles. Nature 359, 137–139.

Vlasenko, V. I., Ivanov, V. A. & Stashchuk, N. M. 1996 Generationof quasi-inertial oscillations during upwelling near the south coastof the Crimea. Okealologya 36, 43–51 (in Russian).

coastal–open ocean exchange in the black sea: … · cuniversite libre de bruxelles, ecologie des...

Documents