Department of GeophysicsUniversity of BergenBergen, NorwayNERSCNERSC

G.C. Rieber Climate InstituteNansen Environmental andRemote Sensing Center,Bergen, Norway Nansen Environmental andRemote Sensing CenterThesis for the Dr. Scient. degree

Evaluation of the Ocean Ventilation Processes in an IsopycnicCoordinate Ocean General Circulation ModelYongqi Gao (yongqi.gao@nersc.no)Supervisors: Prof. Dr. Helge Drange and Prof. Ola M. JohannessenDecember, 2002

AbstractThe ocean uptake of heat and the greenhouse gas carbon dioxide isan important process of the Earth's climate system. The regional dis-tribution and rate of these uptakes are very much determined by thedecadal-scale ventilation of the upper waters of the oceans. In thisthesis, the ocean ventilation processes at high northern and southernlatitudes have been evaluated by performing a series of experimentswith the Miami Isopycnic Coordinate Ocean Model (MICOM). Theevaluation is mainly based on the ocean uptake and the subsequentsub-surface spreading of the chloro uorocarbons CFC-11 and CFC-12.It is shown that analyses of the observed and simulated distributions ofthe CFCs is a powerful method to perform a �rst order determinationof the poorly known isopycnal and diapycnal di�usion coe�cients inOcean General Circulation Models (OGCMs). In addition, transportand mixing of the radionuclides 137Cs and 90Sr from atmospheric falloutand from the Sella�eld reprocessing plant in the UK have been simu-lated for the period 1950-1999. It is argued that simulated distribu-tions of radionuclides like 137Cs and 90Sr can be used to complement therather few observations of radionuclides, and that a model system likethe one adopted in the present study has the potential to be used for thefuture prediction of the radioactive contaminations in the Nordic Seasand the Arctic Ocean, for instance under a global warming scenario.Finally, interannual variability of the water mass exchanges betweenthe North Atlantic and the Nordic Seas have been investigated basedon an intermediate-resolution version of MICOM. It is found that thevolume transports through the openings along the Greenland-ScotlandRidge vary on interannual time scales, and that the ow through theDenmark Strait and between the Faroes and Scotland is correlated withthe North Atlantic Oscillation (NAO). This result illustrates the needfor using realistic atmospheric forcing �elds in order to obtain realis-tic variability in the transport and spreading of tracers embedded inOGCMs.i

ii

AcknowledgmentsThis work has been supported by the G. C. Rieber Climate Institute at the Nansen Envi-ronmental and Remote Sensing Center, Norway. Finanical support from the G. C. RieberFoundation, the Norwegian Research Council and Norsk Hydro as are highly appreciated.Thanks to Prof. Dr. Helge Drange, my supervisor and close friend. It is Helge Drangewho helps me to be what I am now in scienti�c sense.Special thanks to Prof. Ola M. Johannessen for providing me the unique chance to persuethe Dr. degree at Nansen Environmental and Remote Sensing Center.Thanks also to the colleagues at the G. C. Rieber Climate Institute, and then in particularDr. Mats Bentsen for his continuous guidance during the work.Finally, I would like to give the thanks to my wife, Chao Song, for her constant under-standing and encouragement.

iii

iv

This dissertation consists of the following six parts:OverviewPaper IDutay J.C. and J.L. Bullister and S.C. Doney and J.C. Orr and R. Najjar and K.Caldeira and J.M. Campin and H. Drange and M. Follows and Y. Gao and N. Gruberand M.W. Hecht and A. Ishida and F. Joos and K. Lindsay and G. Madec and E.Marier-Reimer and J.C. Marshall and R.J. Matear and P. Monfray and G.K. Plattnerand J. Sarmiento and R. Schlitzer and R. Slater and I.J. Totterdell and M.F. Weirig andY. Yamanaka and A. Yool, 2002. Evaluation of ocean model ventilation with CFC-11:comparison of 13 global ocean models. Ocean Modelling, 4 89-120, 2002Paper IIGao, Y., H. Drange and M. Bentson, 2002. E�ects of diapycnal and isopycnal mixingon the ventilation of CFCs in the North Atlantic in an isopycnic coordinate OGCM,revised version submitted to Tellus.Paper IIIJ. E. �ie Nilsen, Gao, Y., H. Drange, T. Furevik and M. Bentsen, 2002. SimulatedNorth Atlantic-Noridc Seas water mass exchanges in an isopycnic coordinate OGCM,submitted to Geophys. Res. Lett..Paper IVGao, Y. and Drange, H. 2002. Sensitivity of Diapycnal Mixing on the Oceanic Venti-lation and Uptake of CFC-11 in the Southern Ocean, submitted to Journal of MarineSystems.Paper VGao, Y., H. Drange, M. Bentsen and O.M. Johannessen, 2002. Simulating transportof radionuclides in the North Atlantic-Arctic region, submitted to J. Environ. Radioac-tivity.

v

vi

Contents1 Overview 11.1 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.2 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.3 Passive tracers in OGCMs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.4 The ocean model used in this study . . . . . . . . . . . . . . . . . . . . . . . 51.5 Vertical mixing in MICOM . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61.6 Passive tracers used in this study . . . . . . . . . . . . . . . . . . . . . . . . . 61.7 Focus areas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101.8 Model versions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111.9 Summary of the papers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121.9.1 Paper I: Evaluation of ocean model ventilation with CFC-11: compar-ison of 13 global ocean models . . . . . . . . . . . . . . . . . . . . . . 121.9.2 Paper II: E�ects of diapycnal and isopycnal mixing on the ventilationof CFCs in the North Atlantic in an isopycnic coordinate OGCM . . . 131.9.3 Paper III: Simulated North Atlantic-Nordic Seas water mass exchangesin an isopycnic coordinate OGCM . . . . . . . . . . . . . . . . . . . . 131.9.4 Paper IV: Sensitivity of Diapycnal Mixing on the Oceanic Ventilationand Uptake of CFC-11 in the Southern Ocean . . . . . . . . . . . . . . 141.9.5 Paper V: Simulating transport of radionuclides in the North Atlantic-Arctic region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141.10 Future prospective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15References 172 Paper I: Evaluation of ocean model ventilation with CFC-11: comparisonof 13 global ocean models 21vii

3 Paper II: E�ects of diapycnal and isopycnal mixing on the ventilation ofCFCs in the North Atlantic in an isopycnic coordinate OGCM 553.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 593.2 Model Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 613.2.1 Vertical mass transfer in MICOM . . . . . . . . . . . . . . . . . . . . . 623.2.2 Sensitivity experiments . . . . . . . . . . . . . . . . . . . . . . . . . . 633.2.3 Spin-up of the physical model . . . . . . . . . . . . . . . . . . . . . . 643.2.4 Set-up of the CFCs simulation . . . . . . . . . . . . . . . . . . . . . . 643.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 653.3.1 Meridional overturning and zonal distribution of CFC-11 . . . . . . . 653.3.2 Ventilation of the sub-surface water masses . . . . . . . . . . . . . . . 693.3.3 Ventilation of the thermocline . . . . . . . . . . . . . . . . . . . . . . . 703.3.4 Ventilation of the sub-surface water masses . . . . . . . . . . . . . . . 723.3.5 Spreading of the Labrador Sea water . . . . . . . . . . . . . . . . . . . 743.3.6 Spreading of the Over ow Water . . . . . . . . . . . . . . . . . . . . . 773.4 Discussion and conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 783.5 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81References 834 Paper III: Simulated North Atlantic{Nordic Seas water mass exchanges inan isopycnic coordinate OGCM 874.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 914.2 Model Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 924.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 934.4 Discussion and conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 964.5 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99References 1005 Paper IV:Sensitivity of Diapycnal Mixing on the Oceanic Ventilation andUptake of CFC-11 in the Southern Ocean 1035.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1075.2 Model description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1085.2.1 Model con�guration . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1095.2.2 Model experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1105.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110viii

5.3.1 Ocean storage and uptake of CFC-11 . . . . . . . . . . . . . . . . . . . 1105.3.2 The meridional distribution of CFC-11 . . . . . . . . . . . . . . . . . . 1125.3.3 Ventilation of the SO . . . . . . . . . . . . . . . . . . . . . . . . . . . 1135.4 Discussion and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1225.5 Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123References 1246 Paper V: Simulating transport of radionuclides in the North Atlantic-Arcticareas 1276.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1316.2 Model description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1326.2.1 The tracer module . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1336.3 Observed distribution of water masses and radionuclides . . . . . . . . . . . . 1336.4 The radionuclides simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . 1366.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1376.6 Discussion and conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144References 148

ix

x

Chapter 1Overview

1

2

1.1 ObjectiveThe objective of the study has been to evaluate the decadal scale ventilation processes in anisopycnic coordinate Ocean General Circulation Model. Particular emphasis has been put onthe role and sensitivity of the strength of isopycnal and diapycnal mixing, and the role ofapplying synoptic (for instance daily re-analysed) versus climatological (i.e., monthly mean)atmospheric forcing �elds on the upper ocean dynamics and variability.The OGCM used in the study is the Miami Isopycnic Coordinate Ocean Model (MICOM;Bleck et al., 1992).1.2 MotivationThe World Oceans have been identi�ed as being critical to understand the Earth's climatesystem (Houghton et al., 1996). Of the components in climate system, the World Oceans arethe largest dynamically active reservoir of carbon, heat and fresh water on Earth (Hartman,1994; Siegenthaler & Sarmiento, 1993). A recent study shows that changes in the oceanheat content dominate the changes in the Earth's heat balance during the latter half ofthe 20th century (Levitus et al., 2001). Furthermore, numerical simulations (Levitus et al.,2001) indicate that the increase in ocean heat content may largely be due to the increase ofanthropogenic gases in Earth's atmosphere.To understand the role of the oceans in the climate system, OGCMs have been designedover the last decades. OGCMs o�er a method to evaluate past and present oceanic uptakeof the greenhouse gases and provide the only way to predict future ocean uptake in a prog-nostic way. Therefore, it is important to evaluate OGCMs against observations (England &Holloway, 1998; Dutay et al., 2002), particularly those that are used in the coupled climatesimulations. As such, validation of OGCMs is crucial to better understand the present stateof the ocean-atmosphere system, to quantify the ocean's role in the uptake of radiatively(and therefore climatically) important trace gases such as carbon dioxide, and to improvepredictions of tomorrow's climate system.The ocean uptake of the greenhouse gases is closely associated with the ventilation ofthe surface waters of the ocean. For instance, rapid ventilation of ocean can remove orsequester excessive heat and carbon from the atmosphere to great depths by deep convectionand subsequently mitigate climate change (Sarmiento & Qu�er�e, 1996).Temperature and salinity are included in OGCMs per se since these �elds are necessaryto determine the density of water, which in turn in uences the ocean circulation. Therefore,temperature and salinity can be viewed as dynamically active tracers.3

Another class of tracers are chemical tracers dissolved in sea water. The spatial-temporaldistribution of these tracers can be distinct from the hydrographic tracers, and can thusprovide additional information on the source, pathway and rate of the formation of the actualwater masses (Rhein et al., 2002; Orsi et al., 2002). Therefore, chemical tracers are powerfulin the study of ocean ventilation processes in general (Jenkins & Smethie, 1996). Recently,chemical tracers have been implemented into OGCMs for di�erent applications (England &Maier-Reimer, 2001), including validation of OGCMs.The most frequently used chemical tracers are tritium, chloro uorocarbons, natural andbomb-produced radiocarbon, and to a lesser extent, biogeochemical tracers, radionuclides andcertain noble gasses. A brief summary of the chemical tracers is given in Tab. 1.1.Tracers Sources Properties Applications Ref.Radiocarbon Natural and bomb-produced 5730-year Natural and transient 1, 2Tritium Bomb-produced 12.43-year Transient 3CFCs Industry released Stable Transient 4Helium Sea oor volcanism Stable Deep ow 5Phosphate Natural nutrient Biogeochemical Carbon-cycle 6Cesium-137 Fallout and reprocessing plants 30-year Regional modeling 7SF6 Deliberately released Stable Iso- and diapycnal mixing 8Oxygen-18 Natural isotope State fractionation Paleoceanographic study 9Table 1.1: The most frequently used chemical tracers in OGCM simulations (based on England& Maier-Reimer, 2001). Properties include key information about the half-life of the tracer,whether it is biogeochemically active, etc. 1Toggweiler et al. (1989); 2Toggweiler et al. (1989b);3Heinze and Maier-Reimer (1998); 4Dutay et al. (2002); 5Farley et al. (1995); 6Maier-Reimer(1993); 7Nies et al. (1998); 8Sundermeyer and Price (1998); and 9Wadley et al. (2002).As indicated in the table, chemical tracers can be used to assess the OGCMs behaviour ondi�erent temporal and spatial scales. The choice of the applied tracers is therefore applicationdependent. In addition, the source function of the tracer and observations need to be availablefor the tracer to be useful for evaluation of an OGCM.The most widely used tracer data originates from the World Ocean Circulation Experiment(WOCE1) conducted in the 1990s, and the Geochemical Ocean Sections Study (GEOSECS2)in the 1970s.1http://whpo.ucsd.edu2http://ingrid.ldeo.columbia.edu/SOURCES/.GEOSECS/.dataset documentation.html4

1.3 Passive tracers in OGCMsThe governing equation for the transport and mixing of a passive tracer in an OGCM can bewritten as: dCdt = L(C) +Q� �C : (1.1)Here C is the concentration of the dissolved chemical tracer, and L represents a 3-dimensionaloperator including advection and mixing of the tracer. Q is the source/sink term of the tracer,so if the tracer originates from the atmosphere, a source term will be added to the ocean mixedlayer. Finally, � is the decay term if the tracer is radioactive.1.4 The ocean model used in this studyA crucial point in the design of an OGCM is the choice of vertical coordinate. As discussedin Bentsen (2002), three main options are available:The �rst choice is to use the depth z as the vertical coordinate and this kind of models isusually referred to as z-models. Advantages of the z-models are a fairly simple discriminationof the equations, for a Boussinesq uid the horizontal pressure gradient do not generate anyproblems, the equation of state can be accurately represented, and the mixed layer can bewell resolved. Disadvantages are di�culties associated with tracer advection and di�usionalong neutral surfaces (McDougall, 1987), and di�culties associated with the representationof the ocean bottom topography.The second choice is �-models where the vertical coordinate is terrain following. Ad-vantages of the �-models are a natural representation of the bottom boundary layer and anaccurate representation of the equation of state. Disadvantages are cumbersome represen-tation of tracer advection and di�usion along the neutral surfaces and di�culties related toaccurate representation of the horizontal pressure gradient.Finally, �-models use potential density as vertical coordinate. MICOM (Bleck et al., 1992)is such a model, and this model has been used in the thesis. The general picture in the oceaninterior is that the sea water ows and mixes along the neutral surfaces. With a suitablechoice of reference pressure, the neutral surfaces are approximately equal to isopycnals, orto surfaces with constant potential density. Advantages of the �-models are the potentialaccurate representation of the tracer advaction and di�usion, a linear representation of thebottom topography and the easily represented horizontal pressure gradient. For the simula-tions performed in this study, potential density with reference pressure at the surface (�0)and potential density with reference pressure at the 2000 m-depth (�2), are used.The surface mixed layer is mostly unstrati�ed and the representation of it is di�cult in5

�-models. In MICOM, a non-isopycnic surface mixed layer acts as the linkage between atmo-spheric forcing and the ocean interior. The water mass in MICOM is, in addition to density,speci�ed by temperature and salinity and these variables are related by a simpli�ed (3rd de-gree polynomial �t) equation of state (Friedrich & Levitus, 1972). A simpli�ed equation ofstate is used since MICOM frequently derives the state variables from it. The request for thissimpli�cation is a disadvantage with �-models.Vertical mixing in the ocean has long been a fundamental problem to the climate re-searchers (Munk and Wunsch, 1998) since it controls the upward mixing of the deep water.Arti�cially released tracers, for instance SF6, can be used to estimate the strength of the ver-tical mixing. Such studies �nd that the vertical mixing varies signi�cantly with the verticaldensity strati�cation, and that the mixing over the rough topography is greatly enhanced(Ledwell et al., 2000; Ganachaud and Wunsch, 2000; Watson et al., 1999; Muench et al.,2002). Therefore, evaluation of the role of various mixing parameterizations in OGCMs isimportant to better simulate the ocean uptake of, for instance, heat and carbon dioxide.1.5 Vertical mixing in MICOMIn MICOM, the vertical mixing consists of surface water mixing and mixing between theisopycnals, also known as diapycnal mixing. The surface water mixing includes convection,and entrainment and detrainment of the upper mixed layer. When the generation of tur-bulence caused by wind stress or negative buoyancy forcing (caused by cooling or increasedsalinity of the mixed layer) dominates over the positive buoyancy forcing and dissipation,entrainment occurs. The opposite situation holds for the detrainment.In the versions of MICOM used in this thesis, convective mixing takes place if the densityof the surface mixed layer exceeds the density of one or more of the underlying isopycnals.The instability is removed by mixing all of the unstable water masses with the mixed layerwater, and by absorbing the new water mass into the mixed layer.1.6 Passive tracers used in this studyChloro uorocarbons (CFCs)The anthropogenic trace gases CFC-11 (CCl3F) and CFC-12 (CCl2F2) were introduced to theatmosphere in the early 1930s. Because of their importance in stratospheric chemistry, theatmospheric distribution of these gases are closely monitored. Though almost all productionand release take place in the northern hemisphere, rapid mixing in the lower atmosphere and6

0 10 20 30Temperature (oC)

20

25

30

35

40

Sal

inity

(psu

)

7.5

7.5

10

10

15

1517.5

17.5

20

22.5

22.525

2527.5

30

30 12.5

32.5 (a) 0 10 20 30Temperature (oC)

20

25

30

35

40

Sal

inity

(psu

)

2

2

2.5

2.5

3

3.5

3.5

4

4

4.5

4.5

5

55.

5

5.5

6

66.

5

6.5

7

77.

58 (b)Figure 1.1: Solubilities (mol m�3 atm�1) of (a) CFC-11 and (b) CFC-12 in sea water byWarner and Weiss (1985).the chemical stability of CFCs ensure relatively uniform distributions of these gases over thetroposphere (England & Maier-Reimer, 2001).CFCs enter the ocean from the atmosphere by absorption through the air-sea interface.The physical law governing air-sea gas exchange can be written as:F = k � (Csat � Cw); (1.2)where F (mol m�2 s�1) is the gas ux across the air-sea surface, k (m s�1) is the gas transfervelocity (or piston velocity), Csat (mol m�3) is the saturated sea surface concentration ofthe gas and Cw (mol m�3) is the simulated surface tracer concentration. The solubilities ofCFC-11 and CFC-12 are well documented (Warner & Weiss, 1985) and are more dependenton temperature than salinity (Fig. 1.1). Therefore, from a thermodynamic point of view,cold surface waters at high latitudes contain more CFC than warm waters. Given the fairlywell known solubilities and input functions of the CFCs, estimates of Csat is readily deter-mined. The value of the piston velocity k depends on which type of forcing that are applied,i.e., di�erent parameterizations are used for synoptic and climatological (and consequentlysmoothed) winds (Wanninkhof, 1992).Though the concentrations of CFCs in sea water is low (of the order 10�9 mol m�3),the present-day technology is able to measure the concentrations of CFC-11 and CFC-12accurately (detectable down to 0.005�10�12 mol kg�1 in small volume water samples, forinstance, in 30 cm3 water samples; England & Maier-Reimer, 2001). The detection limits ofCFCs are 3 orders of magnitude smaller than the typical near-surface concentrations presentlyfound in the ocean. Measurement of CFCs are available globally, with extended coverage fromthe recent WOCE programme (Fig 1.2 and Tab. 1.2). CFCs are chemically and biologically7

Table 1.2: Available CFC measurements in the WOCE programme (http://www.awi-bremerhaven.de/GEO/eWOCE).Atlantic Ocean Paci�c Ocean India Ocean Southern OceanA01E P17N I01 S04IA24N P16 I03 S04PA24 P04 I05PA05 P17 I06A07 P18A09 P19A17A23inactive in sea water. Based on the above mentioned reasons, CFCs are well suited tracersfor evaluating OGCMs on decadal time scales.It has been demonstrated that the measured distributions of CFCs can serve as a powerfultool to examine the realisms of OGCMs (Dutay et al., 2002; Doney & Hecht, 2002), and toinfer sources and mixing pathways of water masses (Smethie Jr et al., 2000, Orsi et al., 2002).Recently, England and Maier-Reimer (2001) pointed out that model simulation using CFCsis a cost-e�ective way to assess the ocean ventilation for decadal to interdecadal time scales.RadionuclidesStudies of anthropogenic radioactive tracers in the ocean can be used to quantify the radiationlevel and to de�ne the transport pathways of the tracers (Kershaw et al., 1997; Kershaw &Baxter, 1995). The main sources of two of the major anthropogenic radionuclides, Caesium-137 (137Cs) and Strontium-90 (90Sr) are atmospheric fallout from the world-wide nuclearweapons testing, the Chernobyl accident in 1986, and the two principal nuclear reprocessingplants (Sella�eld in the UK and Cap de La Hague in France). The Sella�eld plant has releasedradionuclides into the Irish Sea since 1952 and Cap de La Hague has released radionuclidesinto the English Channel since 1966. In terms of quantities discharged and the impact of theradionuclides concentrations, the Sella�eld release is dominant (Kershaw & Baxter, 1995).Observations of the Sella�eld signal have been performed during the last decades (Dahl-gaard, 1995; Kershaw & Baxter, 1995), faciliating model evaluation. In addition, the inputfunction of the Sella�eld discharge, as a point source, is well de�ned (Dahlgaard, 1995). Ra-dionuclide 137Cs represents a major fraction of the Sella�eld discharge and moreover, it ishighly soluble and can be considered a conservative tracer in sea water.8

(a) (b)

(c) (d)Figure 1.2: Maps of WOCE stations in the Atlantic Ocean (a), the Paci�c Ocean (b), theIndian Ocean (c) and the Southern Ocean (d). Maps from the WOCE Hydrographic ProgramO�ce; http://whpo.ucsd.edu.9

1.7 Focus areasThe Southern OceanBased on a series of numerical simulations performed with di�erent OGCMs, the SouthernOcean (south of 30�S) is thought to be the largest sink for anthropogenic CO2 (Orr et al.,2001). From the same simulations, the largest model di�erences are found in the SouthernOcean with the highest uptake being 70% larger than the lowest. In addition, most of themodels substantially overestimate the storage of anthropogenic CO2 in this region. The modelstudy by Caldeira and Du�y (2000) indicates that the ux of anthropogenic carbon into theSouthern Ocean is high and the storage is low, in agreement with observation-based estimatesof ocean storage of anthropogenic CO2 (Sabine et al., 1999; Gruber, 1998). Moreover, Caldeiraand Du�y (2000) con�rms that the mechanism transporting anthropogenic carbon out of theSouthern Ocean is isopycnal transport.Essentially all of the Southern Ocean model studies have been performed with the classicalz-level OGCMs. It is therefore of interest to examine how the surface water ventilation pro-cesses are represented in an isopycnic coordinate model, like MICOM, in the Southern Ocean,and how sensitive the simulated ocean ventilation is to the applied mixing parameterizations.The North Atlantic OceanIt is believed that the North Atlantic thermohaline circulation is an important componentin the climate system, particularly for the climate in northern Europe (Hartman, 1994).The warm Atlantic surface water transports large amounts of heat northward and at highlatitudes, dense water is formed and sinks and ows southward at intermediate to abyssaldepths, forming the lower limb of the thermohaline circulation cell. The southward owingintermediate to abyssal waters, forming the Deep Western Boundary Currents (DWBCs),transport large volumes of North Atlantic Deep Water (NADW) across Equator and into theSouthern Ocean (Schmitz Jr., 1995). Numerical simulations indicate that the intensity of thethermohaline circulation might be sensitive to changes in the climate system (Rahmstorf andGanopolski, 1999; Clark et al., 2002). Therefore, reliable predictions of the future climateand assessments of possible changes in the global ocean uptake of atmospheric CO2 callfor a realistic representation of this circulation cell. Based on this motivation, the oceanventilation processes in the North Atlantic and the Nordic Seas have been evaluated basedon the simulated uptake, transport and spreading of CFCs.10

1.8 Model versionsA total of four basic versions of MICOM have been used in the study, see Table 1.3 for anoverview. Furthermore, triple experiments have been performed with model version i) andii), yielding a total of eight di�erent model realizations.Model version i) is used in Papers I and IV, addressing the ventilation of the SouthernOcean. Common for all of the three model experiments performed with model version i)is the use of 2000 m as reference level for the isopycnal coordinate surfaces. The choice ofthe 2000 m reference level is motivated by the fact that the Antarctic Bottom Water is notproperly represented if the sea surface is used as the reference level (see also discussion in thefollowing section).For model versions ii) to iv), the focus have been on addressing the ventilation of thesurface waters in the North Atlantic, and to use the model system for addressing some of thenatural variability modes in the marine climate system in the region. In these integrations,sea surface is used as the reference level for the isopycnal coordinate surfaces.The di�erence between model versions ii) to iv) is mainly that version ii) uses a ratherregular horizontal grid system with one pole over the Antarctic continent, and the otherpole over Siberia (Bentsen et al., 1999). This grid con�guration prevents grid singularities inthe Arctic Ocean, and is a grid variant that is sometimes used in coupled atmosphere-oceanclimate models, like in the newly developed Bergen Climate Model (BCM; Furevik et al.,2002). Model versions iii) and iv) uses a horizontal grid con�guration with one pole overNorth America and one pole over Eur-Asia, yielding fairly high resolution in the Atlanticregion. The di�erence between model version iii) and iv) is that the horizontal grid spacingis doubled in version iii) compared to version iv).The two triple experiments that have been performed are characterized by the following:For model version version i), presented in Paper IV, the diapycnal di�usivity (m2 s�1) is setto 0, 0.5 �10�7=N and 2.0 � 10�7=N , where N (s�1) is the Brunt-V�ais�al�a frequency.In model version ii), presented in Paper II, the �rst experiment is characterized by adiapycnal di�usivity 3 � 10�7=N and an isopycnal di�usive velocity (i.e., di�usivity dividedby the size of the grid cell) of 0:01 m s�1. The second experiment follows experiment one butnow the diapycnal di�usivity is set to 5 � 10�8=N , plus a prescribed increase in the mixingagainst the sea oor. The third experiment follows experiment two but with the isopycnaldi�usive velocity set to 0:0025 m s�1. In Paper II therefore, model experiment one and twoform a twin experiment, and model experiment two and three form a second twin experiment.11

Ref. Diapycnal di�. Isopycnal di�. Model Grid Lon-lat # of Paperlevel �10�7=N velocity �10�2 domain system resolution layers(m2 s�1) (m s�1)i) 2000 m 2.0 0.5 69�S-65�N Mercator 2� � 2� cos� 16 Ii) 2000 m 0.0, 0.5, 2.0 0.5 69�S-65�N Mercator 2� � 2� cos� 16 IVii) Surface 3.0, 0.5, 0.5 1.0, 1.0, 0.25 Global Stretched 150 km in NA 24 IIiii) Surface 3.0 0.5 Global Stretched 90-120 km in NA 24 Viv) Surface 3.0 0.5 Global Stretched 45-60 km in NA 26 IIITable 1.3: Overview of the four model con�gurations used in the study, labelled i) to iv).N (s�1) is the Brunt-V�ais�al�a frequency, NA is the North Atlantic, Mercator denotes theMercator grid projection, � (rad) is latitude, and # of layers denotes the number of verticallayers in the model system.1.9 Summary of the papers1.9.1 Paper I: Evaluation of ocean model ventilation with CFC-11: com-parison of 13 global ocean modelsCFCs simulations were performed using 13 coarse-resolution models which participated inthe Ocean Carbon Model Intercomparison Project (OCMIP). The comparison between thesimulated and the observed CFC-11 evaluate the abilities of these ocean models in the ven-tilation on several decades time scale. It was found that the simulated global storage of theCFC-11 in the 13 models di�ers greatly (�30%), which is mainly due to the di�erence in theventilation from the high latitudes. In the Southern Ocean, the meridional distribution ofthe CFC-11 di�ers particularly in the intermediate water, whereas the zonal distribution ismainly controlled by the subgrid-scale parameterization. With isopycnal di�usion and eddy-induced velocity parameterization included, models generated more realistic ventilation in theintermediate water. In the ventilation of the deep and bottom water, the models also varysubstantially. The comparison also shows that the coupled ocean-ice models systematicallyproduced more realistic AABW formation region; however, these models still largely overes-timate AABW ventilation if no speci�c parameterization of brine rejection is included. Inthe North Paci�c Ocean, all the 13 models exhibit a systematic large underestimation of theCFC-11 uptake in the thermocline of the subtropical gyre, while no systematic di�erence to-ward the observation is found in the subpolar gyre. In the North Atlantic Ocean, the CFC-11uptake is globally underestimated in the subsurface. In the deep ocean, all but the reversemodel, failed to capture the two recently ventilated branches observed in the North AtlanticDeep Water (NADW). In addition, the simulated transport in the Deep Western BoundaryCurrent (DWBC) is too sluggish in all but the isopycnal model, where it is too rapid.12

1.9.2 Paper II: E�ects of diapycnal and isopycnal mixing on the ventilationof CFCs in the North Atlantic in an isopycnic coordinate OGCMA global version of the Miami Isopycnic Coordinate Ocean Model is used to performed theCFCs simulations. The simulated and the observed CFC-11 and the ratio of CFC-11/CFC-12 are compared to examine the ventilation in the North Atlantic Ocean. Three experimentsare carried out: One with a diapycnal di�usivity Kd = 3� 10�7=N m2 s�1 and an isopycnaldi�usive velocity (i.e., di�usivity divided by the size of the grid cell) vtrac = 0:01 m s �1(Exp. 1); one as Exp. 1 but with Kd = 5 � 10�8=N m2 s�1 plus increased bottom mixing(Exp. 2); and one as Exp. 2 but with vtrac = 0:0025 m s�1 (Exp. 3). The main featuresof the simulated ventilation are strong uptake of the CFCs in the Labrador, Irminger andNordic Seas, and a topographically-aligned geostrophically-controlled southward transportof CFC-enriched water in the Atlantic. It is found that the Over ow Waters (OW) fromthe Nordic Seas, the penetration of the western boundary currents, the ventilation of thesubtropical surface waters, the vertical density strati�cation and the meridional overturningare all critically dependent on the applied isopycnal and diapycnal di�usivities, with Exp. 3(Exp. 1) yielding the most (least) realistic CFC distributions. Furthermore, it is the combinedrather than the isolated e�ect of the isopycnal and diapycnal di�usivities that matter. Forinstance, the strength of the simulated Meridional Overturning Circulation (MOC) is similarin Exps. 1 and 3, but the simulated CFC-distributions are far too di�usive in Exp. 1 andfairly realistic in Exp. 3. It is shown that the simulated distributions of transient tracers likethe CFCs can be used to tune the strength of the applied isopycnal mixing parameterization,a tuning that is di�cult to conduct based on the simulated distributions of temperature andsalinity alone.1.9.3 Paper III: Simulated North Atlantic-Nordic Seas water mass ex-changes in an isopycnic coordinate OGCMThe variability in the volume exchanges between the North Atlantic and the Nordic Seasduring the last 50 years are investigated using a synoptic-forced, global version of the MiamiIsopycnic Coordinate Ocean Model (MICOM). The simulated volume uxes roughly matchthe existing observations. The net volume ux across the Faroe-Shetland Channel (FSC) ispositively correlated with the net ux through the Denmark Strait (DS; R=0.74 for 3 yearslow pass �ltering), but negatively correlated with the net ux across Iceland-Faroe Ridge(IFR; R=-0.80). For the Atlantic In ow (AtI) across the FSC and IFR, the correlation isR=-0.59. Furthermore, the simulation suggests that an atmospheric pattern with similaritiesto the North Atlantic Oscillation is the main driving force to the inter-annual variations in13

the winter-mean volume exchanges through the DS and the FSC. The model also shows a0.02 Sv yr�1 reduction of the AtI to the Nordic Seas since 1957.1.9.4 Paper IV: Sensitivity of Diapycnal Mixing on the Oceanic Ventilationand Uptake of CFC-11 in the Southern OceanThe sensitivity of diapycnal mixing on the oceanic uptake of anthropogenic CFC-11 andthe ventilation of the surface waters in the Southern Ocean (SO) are investigated with theMiami Isopycnic Coordinate Ocean Model (MICOM). Three model experiments have beenperformed: One with a diapycnal mixing Kd (m2 s�1) of 2 � 10�7=N (Exp. 1), one withvanishing diapycnal mixing (Exp. 2), and one with 5�10�8=N (Exp. 3; N (s�1) is the Brunt-V�ais�al�a frequency). The performed model simulations indicate that the observed verticaldistribution of CFC-11 along 272 �E in the SO is caused by local ventilation of the surfacewaters and westward-directed isopycnic transport and mixing from deeply ventilated waters inthe Weddell Sea region. It is found that at the end of 1997, the simulated net ocean uptake ofCFC-11 in Exp. 2 is 25% below that of Exp. 1. Of this reduction, the decreased uptake of CFC-11 in the SO accounts for 80% of the total di�erence. Furthermore, Exps. 2 and 3 yield farmore realistic vertical distributions of the ventilated CFC-waters than Exp. 1. The performedtriple-experiment also highlights the sensitivity of the SO surface water ventilation to thedistribution of the simulated ML. Rather small di�erences in the maximum ML depth mayeasily generate huge di�erences in the temporal-spatial properties of the ventilated waters. Itis argued that inclusion of CFC's in coupled climate models could be used as a test-bed forevaluating the decadal-scale ocean uptake of heat and CO2 in these models.1.9.5 Paper V: Simulating transport of radionuclides in the North Atlantic-Arctic regionThe spatial and temporal distributions of the anthropogenic radionuclides 137Cs and 90Sr, orig-inating from nuclear bomb testing and the Sella�eld reprocessing plants in the Irish Sea, aresimulated using a global version of the Miami Isopycnic Coordinate Ocean Model (MICOM)with daily atmospheric forcing (NCAR/NCEP). The obtained spreading of the Sella�eld dis-charge along the coast of Norway and into the Barents Sea closely match the observations.Since the observations are limited, our simulation gives the detailed distribution of the ra-dionuclides spreading in both time and space. Furthermore, the model can be used for thefuture prediction of the radioactive contaminations in the Nordic Seas and the Arctic Oceanunder the di�erent climate scenario, for instance, the global warming.14

1.10 Future prospectiveThe presented studies have mainly focussed on the ventilation processes of the surface watersat high northern (i.e., Atlantic) and southern latitudes. It is concluded that a reasonabledescription of the ventilation and the subsequent transport and mixing processes of the venti-lated surface waters in the Atlantic region can be obtained with proper choice of the diapycnaland isopycnal mixing coe�cients. A similar conclusion is drawn based on the performed sensi-tivity experiments in the Southern Ocean. It is also argued that proper evaluation of coupledclimate models can be achieved by incorporating CFCs into such models. A natural extensionof the work in the thesis is therefore to include CFCs into the recently developed BCM model(Furevik et al., 2002) to assess the decadal scale ventilation in this model.A challenge for coupled climate model studies using an isopycnic coordinate OGCM isthe choice of reference level for the coordinate system. Since the �0 coordinate system doesnot resolve the di�erence between the North Atlantic Deep Water and the Antarctic BottomWater, a proper representation of the World Oceans calls for, for instance, 2000 m as referencelevel for the isopycnic layers. In the present version of BCM, �0 is used as the verticalcoordinate system. In the ongoing build-up phase of the next version of BCM, extensiveevaluation of the high northern latitude surface water ventilation properties need to be carriedout with the OGCM using, for instance, �2 as the reference level. The latter exercise can bebased on the work presented in this thesis.It is furthermore of importance to assess the ventilation properties of the applied modelsystem by using realistic, for instance daily synoptic re-analysed, atmospheric forcing �elds.Such model studies are currently being conducted, and represent therefore a continuation ofthe climatological-forced model simulations presented in, for instance, Paper II.The parameterization schemes for diapycnal and isopycnal mixing used in this study aresimple, but is still more elaborate than �xed values used in many present day climate OGCMs(see for instance discussion in Nilsson and Walin, 2001). A more complete description of thediapycnal mixing would include parameterizations of, for instance, bottom gravity currentsand tidal mixing over rough topography. It is foreseen that basic research on these processesand improved model parameterizations should be run in parallel to optimize the output ofthe studies.Another model improvement that should be considered is the treatment of convective in-stabilities. In the present version of MICOM, static instability is removed by entraining allof the unstable water masses into the upper mixed layer. Since the size of the grid cells istypically of the order 100 km, sub-grid scale mixing schemes should be incorporated into themodel. The e�ect of improved treatment of convective mixing can be assesed by analysing15

simulated and observed water mass and tracer distribution characteristics in convective re-gions. The ongoing SF6 experiment in the Greenland Sea (Watson et al., 1999) appears idealfor this purpose, and such an activity is currently being conducted.Finally, ocean or climate modeling in general is little but an academic exercise if keyobservations of the climate system are not available. It is therefore of paramount importanceto secure high-quality observations of the climate system. One way to do this is to moreactively join groups working on observations and modeling. Such collaboration will strengthenthe need for both observations and modeling, and is certainly required to improve predictionsof the climate system over the next century or so.

16

ReferencesBentsen, M. (2002). Modelling Ocean Climate Variability of the North Atlantic and the Nordic Seas. Unpub-lished doctoral dissertation, Univ. of Bergen, Bergen, Norway.Bentsen, M., Evensen, G., Drange, H., & Jenkins, A. D. (1999). Coordinate Transformation on a Sphere UsingConformal Mapping. Mon. Wea. Rev., 127, 2733{2740.Bleck, R., Rooth, C., Hu, D., & Smith, L. T. (1992). Salinity-driven thermohaline transients in a wind-and thermohaline-forced isopycnic coordinate model of the North Atlantic. J. Phys. Oceanogr., 22,1486-1515.Caldeira, K., & Du�y, P. B. (2000). The Role of the Southern Ocean in Uptake and Storage of AnthropogenicCarbon Dioxide. Science, 287, 620-622.Clark, P. U., Pisias, N. G., Stocker, T. F., & Weaver, A. J. (2002). The role of the thermohaline circulationin abrupt climate change. Nature, 415, 863-869.Dahlgaard, H. (1995). Transfer of European Coastal Pollution to the Arctic: Radioactive Tracers. MarinePollution Bulletin, 31, 3-7.Doney, S. C., & Hecht, M. W. (2002). Antarctic BottomWater Formation and Deep-Water Chloro uorocarbonDistributions in a Global Ocean Climate Model. J. Phys. Oceanogr., 32, 1642-1666.Dutay, J., Bullister, J., Doney, S., Orr, J., Najjar, R., Caldeira, K., Campin, J., Drange, H., Follows, M., Gao,Y., Gruber, N., Hecht, M., Ishida, A., Joos, F., Lindsay, K., Madec, G., Marier-Reimer, E., Marshall,J., Matear, R., Monfray, P., Plattner, G., Sarmiento, J., Schlitzer, R., Slater, R., Totterdell, I., Weirig,M., Yamanaka, Y., & Yool, A. (2002). Evaluation of ocean model ventilation with CFC-11: comparisonof 13 global ocean models. Ocean Modelling, 4 (2), 89-120.Dutay, J.-C., Bullister, J. L., Doney, S. C., Orr, J. C., Najjar, R., Caldeira, K., Campin, J. M., Drange, H.,Follows, M., Gao, Y., Gruber, N., Hecht, M. W., Ishida, A., Joos, F., Lindsay, K., Madec, G., Marier-Reimer, E., Marshall, J. C., Matear, R. J., Monfray, P., Plattner, G. K., Sarmiento, J., Schlitzer, R.,Slater, R., Totterdell, I. J., Weirig, M. F., Yamanaka, Y., & Yool, A. (2002). Evaluation of ocean modelventilation with CFC-11: comparison of 13 global ocean models. Ocean Modelling, 4, 89-120.England, M., & Holloway, G. (1998). Simulations of CFC content and water mass age in the deep NorthAtlantic. J. Geophys. Res., 103 (C8), 15885-15901.England, M., & Maier-Reimer, E. (2001). Using Chemical Tracers to Assess Ocean Models. Rev. of Geophys.,39 (1), 29-70.Farley, K., Maier-Reimer, E., Schlosser, P., & Broecker, W. (1995). Constraints on mantle 3He uxes anddeep-sea circulation from an ocean general circulation model. J. Geophys. Res., 100, 3829-3839.17

Friedrich, H., & Levitus, S. (1972). An approximation to the equation of state for sea water, suitable fornumerical ocean models. J. Phys. Oceanogr., 2, 514-517.Furevik, T., Bentsen, M., Drange, H., Kindem, I., Kvamst�, N., & Sorteberg, A. (2002). Description andvalidation of the Bergen Climate Model: ARPEGE coupled with MICOM. Clim. Dyn. (accepted)Ganachaud, A., & Wunsch, C. (2000). Improved estimates of global ocean circulation, heat transport andmixing from hydrographic data. Nature, 408, 453-457.Gruber, N. (1998). Anthropogenic CO2 in the Atlantic Ocean. Global Biogeochem. Cycles, 12 (1), 165-191.Hartman, D. L. (1994). Global Physical Climatorology. Academic Press Inc., U.S.A.Heinze, C., & Maier-Reimer, E. (1998). Transient tracers in a global ogcm: Source functions and simulateddistributions. J. Geophys. Res., 103 (C8), 15903-15922.Houghton, J. T., Filho, L. G. M., Callander, B. A., Harris, N., Kattenberg, A., & Maskell, K. (Eds.). (1996).Climate Change 1995: The Science of Climate Change. Contribution of Working Group I to the SecondAssessment Report of the Intergovernmental Panel on Climate Change. Cambridge, UK: CambridgeUniversity Press.Jenkins, W., & Smethie, W. (1996). Transient tracers track ocean climate signals. Oceanus, 29, 29-32.Kershaw, P., & Baxter, A. (1995). The transfer of reprocessing wastes from north-west Europe to the Arctic.Deep Sea Res., 42, 1413-1448.Kershaw, P., Gurbutt, P., Woodhead, D., Leonard, K., & Rees, J. (1997). Estimates of Fluxes of 137Cs inNorthern Waters from recent Measurements. Sci. Total Environm., 202, 211-223.Ledwell, J. R., Montgomery, E., Polzin, K., Laurent, L., Schmitt, R., & Toole, J. (2000). Evidence for enhancedmixing over rough topography in the abyssal ocean. Nature, 403, 179-182.Levitus, S., Antonov, J., Wang, J., Delworth, T., Dixon, K., & Broccoli, A. (2001). Anthropogenic Warmingof Earth's Climate System. Science, 292, 267-270.Maier-Reimer, E. (1993). The biological pump in the greenhouse. Global and Planetary Change, 8, 13-15.McDougall, T. J. (1987). Thermobaricity, Cabbeling and Water-Mass Conversion. J. Geophys. Res., 92 (C5),5448-5464.Muench, R., Padman, L., Howard, S., & Fahrbach, E. (2002). Upper ocean diapycnal mixing in the north-western Weddell Sea. Deep Sea Res., 49, 4843-4861.Munk, W., & Wunsch, C. (1998). Abyssal recipes ii: energetics of tidal and wind mixing. Deep Sea Res.,45(I), 1977-2010.Nies, H., Harms, H., Karcher, M., Dethle�, D., Bahe, C., Kuhlmann, G., Oberhuber, J., Backhaus, J., Kleine,E., Loewe, P., Matishov, D., Stepanov, A., & Vasiliev, O. (1998). Anthropogenic Radioactivity in theNordic Seas and the Arctic Ocean - Results of a Joint Project. Deutsche Hydrographische Zeitschrift,50 (4), 313-343.Nilsson, J., & Walin, G. (2001). Freshwater forcing as a booster of thermohaline circulation. Tellus, 53A,628-640.Orr, J., Maier-Reimer, E., Mikolajewicz, U., Monfray, P., Sarmiento, J., Toggweiler, J., Taylor, N., Palmer,J., Gruber, N., Sabine, C., Quere, C., Key, R., & Boutin, J. (2001). Estimates of anthropogenic carbonuptake from four three-dimensional global ocean models. Global Biogeochem. Cycles, 15 (1), 43-60.18

Orsi, A. H., Jr, W. M. S., & Bullister, J. L. (2002). On the total input of antarctic waters to the deep ocean:A preliminary estimate from chloro uorocarbon measurements. J. Geophys. Res., 107 (C8), 31-1-31-17.Rahmstorf, S., & Ganopolski, A. (1999). Long-term global warming scenarios computed with an e�cientcoupled climate model. Climatic Change, 43, 353-367.Rhein, M., Fischer, J., Smethie, W., Smythe-Wright, D., Weiss, R., Mertens, C., Min, D.-H., Fleischmann, U.,& Putzka, A. (2002). Labrador Sea Water: Pathway, CFC Inventory, and Formation Rates. J. Phys.Oceanogr., 32, 648-665.Sabine, C., Key, R., Johnson, K., Millero, F., Poisson, A., Sarmiento, J., Wallace, D., & Winn, C. (1999).Anthropogenic CO2 inventory of the Indian Ocean. Global Biogeochem. Cycles, 13, 179-198.Sarmiento, J., & Qu�er�e, C. (1996). Oceanic CO2 uptake in a model of century-scale global warming. Science,274, 1346-1350.Schmitz Jr., W. J. (1995). On the interbasin-scale thermohaline circulation. Rev. Geophys., 33 (2), 151{173.Siegenthaler, U., & Sarmiento, J. (1993). Atmospheric carbon dioxide and the ocean. Nature, 365, 119-125.Smethie Jr, W. M., Fine, R., Putzka, A., & Jones, E. (2000). Tracing the ow of north atlantic deep waterusing chloro uorocarbons. J. Geophys. Res., 105 (C6), 14297-14323.Sundermeyer, M., & Price, J. (1998). Lateral mixing and the North Atlantic Tracer Release Experiment:Observations and numerical simulations of Lagrangian particles and a passive tracer. J. Geophys. Res.,103 (C10), 21481-21497.Toggweiler, J., Dixon, K., & Bryan, K. (1989a). Simulations of radiocarbon in a coarse-resolution world oceanmodel, 1, Steady state prebomb distributions. J. Geophys. Res., 94, 8217-8242.Toggweiler, J., Dixon, K., & Bryan, K. (1989b). Simulations of radiocarbon in a coarse-resolution world oceanmodel, 2, Distributions of Bomb-Produced Carbon 14. J. Geophys. Res., 94, 8243-8264.Wadley, M. R., Bigg, G. R., Rohling, E. J., & Payne, A. J. (2002). On modelling present-day and last glacialmaximum oceanic �18O distributions. Global and Planetary Change, 32, 89-109.Wanninkhof, R. (1992). Relationship Between Wind Speed and Gas Exchange Over the Ocean. J. Geophys.Res., 97 (C5), 7373-7382.Warner, M., & Weiss, R. F. (1985). Solubilities of chloro uorocarbons 11 and 12 in water and sea water. DeepSea Res., 32, 1485-1497.Watson, A., Messias, M., Fogelqvist, E., Van Scoy, K., Johannessen, T., Oliver, K., Stevens, D., Rey, F.,Tanhua, T., Olsson, K., Carse, F., Simonsen, K., Ledwell, J., Jansen, E., Cooper, D., Kruepke, J., &Gullyard, E. (1999). Mixing and convection in the Greenland Sea from a tracer-release experiment.Nature, 401, 902-904.

19

20

Chapter 2Paper I: Evaluation of ocean modelventilation with CFC-11:comparison of 13 global oceanmodels

21

22

Chapter 3Paper II: E�ects of diapycnal andisopycnal mixing on the ventilationof CFCs in the North Atlantic in anisopycnic coordinate OGCM

55

56

E�ects of diapycnal and isopycnal mixing on the ventilation of CFCs in theNorth Atlantic in an isopycnic coordinate OGCMYongqi Gao�, Helge Drange and Mats BentsenNansen Environmental and Remote Sensing CenterEdv. Griegsv. 3A, N-5059 Bergen, Norway(revised version submitted to Tellus, 2002)

* Corresponding author:Yongqi Gao (yong@nersc.no)Nansen Environmental and Remote Sensing CenterEdv. Griegsv. 3AN{5059 BergenNORWAY57

AbstractSimulated distributions of the chloro uorocarbons CFC-11 and CFC-12 are used to exam-ine the ventilation of the North Atlantic Ocean in a global version of the Miami IsopycnicCoordinate Ocean Model (MICOM). Three simulations are performed: One with a diapyc-nal di�usivity Kd = 3 � 10�7=N m2 s�1 and an isopycnal di�usive velocity (i.e., di�usivitydivided by the size of the grid cell) vtrac = 0:01 m s�1 (Exp. 1); one as Exp. 1 but withKd = 5 � 10�8=N m2 s�1 plus increased bottom mixing (Exp. 2); and one as Exp. 2 butwith vtrac = 0:0025 m s�1 (Exp. 3). The main features of the simulated ventilation are stronguptake of the CFCs in the Labrador, Irminger and Nordic Seas, and a topographically-alignedgeostrophically-controlled southward transport of CFC-enriched water in the Atlantic. It isfound that the Over ow Waters (OW) from the Nordic Seas, the penetration of the west-ern boundary currents, the ventilation of the subtropical surface waters, the vertical densitystrati�cation and the meridional overturning are all critically dependent on the applied isopy-cnal and diapycnal di�usivities, with Exp. 3 (Exp. 1) yielding the most (least) realistic CFCdistributions. Furthermore, it is the combined rather than the isolated e�ect of the isopy-cnal and diapycnal di�usivities that matter. For instance, the strength of the simulatedMeridional Overturning Circulation (MOC) is similar in Exps. 1 and 3, but the simulatedCFC-distributions are far too di�usive in Exp. 1 and fairly realistic in Exp. 3. It is shown thatthe simulated distributions of transient tracers like the CFCs can be used to tune the strengthof the applied isopycnal mixing parameterization, a tuning that is di�cult to conduct basedon the simulated distributions of temperature and salinity alone.

58

3.1 IntroductionThe North Atlantic thermohaline circulation is an active and important component of theclimate system, particularly on multi-annual to decadal time scales (e.g., Curry & McCartney,2001; Eden and Jung, 2001). The warm and saline surface water owing northward with theNorth Atlantic Current (NAC), the extension of the Gulf Stream current system further southin the Atlantic, transports large amounts of heat and salt poleward. Large parts of the heat isreleased to the atmosphere, especially during the cold and windy winter months, contributingto the anomalous (high) temperatures of especially the north-western Europe (e.g., Hartman,1994; Rahmstorf and Ganopolski, 1999). Likewise, the transport of salt is crucial for theformation of dense surface waters in the Labrador and the Nordic Seas (here de�ned as theGreenland, Norwegian and Iceland Seas) (Furevik et al., 2002).The dense water masses formed in the open ocean, together with dense brine water formedon the Arctic shelves during formation of ice (Jones et al., 1995; Rudels et al., 1994) formintermediate, and in some occasions, truly abyss water. The major branches of the cold anddense intermediate to deep currents of the North Atlantic Ocean are known as the DeepWestern Boundary Currents (DWBCs; Warren, 1981), and they form the lower limb of thethermohaline circulation (THC) cell (Gordon, 1986).Climatological proxy data and numerical simulations indicate that the intensity of thethermohaline circulation is sensitive to the actual state of the climate system, and vice versa(Manabe and Stou�er, 1999). Therefore, reliable realizations of the past, present and futureclimate system, including regional climate predictions for various climate scenarios and -for instance - assessments of possible changes in the global ocean uptake of heat and thegreenhouse gas CO2, require a realistic representation of this circulation cell.The chloro uorocarbon trace gases CCl3F (CFC-11) and CCl2F2 (CFC-12) were intro-duced to the atmosphere in the early 1930's. The atmospheric evolution of the CFCs havebeen monitored continuously since then (Fig. 3.1), and the oceanic concentration has beenmeasured at several occasions and locations over the last 20 years (see below). CFCs have nonatural sources and are chemically and biologically inactive in the ocean, so the CFCs are wellsuited tracers to shed light onto seasonal to decadal time scale processes in the ocean and onthe representation of these processes in Ocean General Circulation Models (OGCMs). There-fore, the measured distributions of CFCs have been used to examine the degree of realism ofOGCMs (Dixon et al., 1996; Dutay et al., 2002), and to infer sources and mixing pathways ofwater masses in the ocean (Smethie Jr et al., 2000) and in the OGCMs (England & Holloway,1998). In a recent paper, England and Maier-Reimer (2001) argue that OGCM simulationwith CFCs is a cost-e�ective approach to evaluate the inter-decadal ocean ventilation.59

1940 1950 1960 1970 1980 1990Year

0

100

200

300

400

500

600

CF

Cs

Par

tialP

ress

ure

(pic

oatm

)

CFC-12

CFC-11(a) 1940 1950 1960 1970 1980 1990Year

0

0.1

0.2

0.3

0.4

0.5

0.6

CF

C11

:CF

C-1

2R

atio

(b)Figure 3.1: (a) Reconstructed history of CFC-11 and CFC-12 partial pressure in dry air atone atmosphere pressure (Walker et al., 2000) and (b) ratio of CFC-11/CFC-12. The solid(dashed) line corresponds to the CFC concentrations and the CFC-11/CFC-12 ration in thenorthern (southern) hemisphere.Analysis of North Atlantic CFC observation has been performed to reveal the pathwayof the DWBCs and to quantify the speed and mixing of these currents (Andrie et al., 1999).Weiss et al. (1985) identi�ed a CFC-rich water mass near 1600 m depth at the TropicalAtlantic. Using CFC observations obtained in 1992, Smethie Jr et al. (2000) was able toidentify two prominent sub-surface CFC-11 maxima in the western North Atlantic along24�N. Both cores were intensi�ed to the west, extending into the ocean interior. The upperCFC-11 core was related to the water masses formed in the Labrador Sea, known as the UpperLabrador Sea Water (ULSW; Pickart et al., 1996), whereas the lower one originates from theOW (Hansen & �sterhus, 2000) from the Nordic Seas.In this paper, the CFC-11 burden on the North Atlantic Ocean water masses is sim-ulated using a truly global version of MICOM (Bleck et al., 1992), coupled to a dynamic-thermodynamic sea ice model. Guided by earlier studies performed with a semi-global versionof a similar model (Dutay et al., 2002), special attention has been paid to on the representationof the timing, location, strength, advective transport and dispersive mixing of the ventilatedwater masses, and the oceanic responses to the strength of the diapycnal and ispycnal mixingparameterizations. This is, to our knowledge, the �rst numerical sensitivity experiment usingCFCs in an isopycnic coordinate OGCM.For the CFC-experiments presented here, the choice of an isopycnic coordinate OGCM isappealing for mainly two reasons: The preferred plane of advective ow in the ocean is along,rather than across, planes of constant density, and the mixing is orders of magnitude weakerin the diapycnal than in the isopycnal direction (Ledwell & Watson, 1993; Toole et al., 1994).60

-120 -60 0 60 120Longitude

-90

-60

-30

0

30

60

90

Latit

ude

-120 -60 0 60 120Longitude

-90

-60

-30

0

30

60

90

Latit

ude

-120 -60 0 60 120Longitude

-90

-60

-30

0

30

60

90

Latit

ude

Figure 3.2: The horizontal grid layout (every forth grid line is shown) and grid size (km) inthe experiments. Grid size smaller than 150 km is shaded.These features are embedded into the isopycnic concept per se. In fact, in the absence ofdiapycnic mixing, the isopycnic interfaces can be viewed as material surfaces as there are nonumerical mixing across these interfaces.The paper is organized as follows: In section 3.2 we present a brief description of the modelused in the experiments. In section 3.3 modelled results are shown and comparison are madewith CFC-11, CFC-12 and hydrographic observations. The simulated CFC-distributions arediscussed and summarized in section 3.4.3.2 Model DescriptionAs already stated, a global version of MICOM has been used in this study. In the horizontal,a local orthogonal grid system with one pole over Siberia and one pole over Antarctica wasadopted (Bentsen et al., 1999). The horizontal grid scale is shown in Fig. 3.2 and variesbetween 50 and 200 km. Along Equator, a local grid re�nement is applied in a 10� latitudinalband. There are 24 layers in the vertical, of which the uppermost mixed layer (ML) hasa spatial and temporal varying density. The speci�ed potential densities of the sub-surfacelayers were chosen to ensure proper representation of the major water masses in the NorthAtlantic/Nordic Sea regions. The densities of the isopycnic layers (in �0-units) are 24.12,24.70, 25.28, 25.77, 26.18, 26.52, 26.80, 27.03, 27.22, 27.38, 27.52, 27.63, 27.71, 27.77, 27.82,27.86, 27.90, 27.94, 27.98, 28.01, 28.04, 28.07 and 28.10. The vertically homogeneous ML61

utilizes the Gaspar (1988) bulk parameterization for the dissipation of turbulent kinetic en-ergy, and has temperature, salinity and layer thickness as the prognostic variables. In theisopycnic layers below, temperature and layer thickness are the prognostic variables, whereassalinity is diagnostically determined using the simpli�ed equation of state by Friedrich andLevitus (1972). The bathymetry is computed based on the arithmetic mean of the ETOPO-5data set (from National Geophysical Data Center, USA). No smoothing or adjustment of thetopography were made.The model is forced with monthly mean values of the NCAR/NCEP reanalysis (Kalnayet al., 1996) wind stress, short wave, long wave, latent and sensible heat uxes, precipitationand sea level pressure �elds by applying the Fairall et al. (1996) bulk parameterization for themomentum, heat and fresh water uxes (Bentsen and Drange, 2000). The ML temperatureand salinity are relaxed towards the monthly mean climatological values of Levitus and Boyer(1994) and Levitus et al. (1994), with a relaxation time scale of 30 days for a 50 m thick ML.The relaxation is reduced linearly with ML thickness exceeding 50 m, and it is set to zeroin waters where sea ice is present in March (September) in the Arctic (Antarctic) to avoidrelaxation towards temperature or salinity outliers in the poorly sampled polar waters.The thermodynamic module incorporates freezing and melting of sea ice and snow coveredsea ice (Drange & Simonsen, 1996). The dynamic part of the sea ice module is based on theviscous-plastic rheology of Hibler (1979), where sea ice is considered as a two-dimensionalcontiunum. The dynamic ice module has been further modi�ed by Harder (1996) to includedescription of sea ice roughness and the age of sea ice, and utilizing the advection scheme ofSmolarkiewicz (1984).The continuity, momentum and tracer equations are discretized on an Arakawa and Lamb(1977) C- grid stencil. The di�usive velocities (di�usivities divided by the size of the gridcell) for layer interface di�usion and momentum dissipation are 0.01 m s�1 and 0.02 m s�1,respectively, yielding actual di�usivities of about 103 m2 s�1. For the base-line integration(hereafter Exp. 1), the di�usive velocity for tracer (i.e., for temperature, salinity and theCFCs) dispersion vtrac = 0:01 m s�1. A ux corrected transport scheme (Smolarkiewicz& Clark, 1986) is used to advect the model layer thickness and the tracer quantities. Theprocesses governing the vertical mass transfer are described in Sec. 3.2.1. Readers interestedin the intrinsic details about MICOM are referred to Bleck et al. (1992).3.2.1 Vertical mass transfer in MICOMThe seasonal cycle of the ML thickness in MICOM is governed by entrainment, detrainmentand convection. When the turbulent kinetic energy is positive, i.e., when the generation ofturbulence caused by wind stress or negative buoyancy forcing (caused by cooling or increased62

salinity of the ML) dominates over the positive buoyancy forcing and dissipation, entrainmentoccurs. The opposite situation holds for detrainment.Convective mixing takes place if the density of the surface ML exceeds the density of oneor more of the underlying isopycnals. The instability is removed by mixing all of the unstablewater masses with the ML water, and by absorbing the new water mass into the ML. Bothmomentum and tracers (temperature, salinity and the CFCs) are uniformly mixed duringconvective adjustment episodes.The interior layers exchange their properties with the ML if they outcrop to the base of theML. Therefore, the location and timing of the outcropping layers are of central importancein analyzing the ocean ventilation of any of the ML properties.The interior isopycnic layers which usually do not have direct contact with the atmosphereare considered to behave adiabatically, except for a density strati�cation dependent diapycnalmixing between the layers. For the base-line version of the model (Exp. 1), the diapycnalmixing coe�cient Kd (m2 s�1) is parameterized according to the Gargett (1984) expressionKd = 3� 10�7N ; (3.1)where N = qg� @�@z (s�1) is the Brunt-V�ais�al�a frequency, g (m s�2) is the gravity acceleration,� (kg m�3) is the density and z (m) is the depth. The numerical implementation of Eq. 3.1follows the scheme of McDougall and Dewar (1998).3.2.2 Sensitivity experimentsTo examine the model sensitivity to the value of the diapycnal mixing Kd, a twin integration(hereafter Exp. 2) was performed in which Kd = 5�10�8=N (m2 s�1), or a factor 6 below thevalue used in Exp. 1. To further account for increased mixing towards the sea oor (Munkand Wunsch, 1998; Ledwell et al., 2000), Kd in Exp. 2 was increased by a factor four inthe layers located within 100 m from the sea oor. The diagnosed di�usivities in the NorthAtlantic are illustrated in Fig. 3.3, together with a global estimate of Kd based on the studyof Hasumi and Suginohara (1999).A second twin integration was performed based on Exp. 2 by reducing the di�usive velocityfor tracer dispersion vtrac by a factor four (hereafter Exp. 3). The three experiments, formingtwo twin experiments, are therefore characterized by the following:Exp. 1: Kd = 3� 10�7=N m2 s�1; vtrac = 0:01 m s�1,Exp. 2: Kd = 5� 10�8=N m2 s�1 plus increased bottom mixing; vtrac = 0:01 m s�1, andExp. 3: Kd = 5� 10�8=N m2 s�1 plus increased bottom mixing; vtrac = 0:0025 m s�1.63

0 0.5 1 1.5 2 2.5 3Diffusivity (cm2 s-1)

0

500

1000

1500

2000

2500

3000

3500

4000D

epth

(m)

Figure 3.3: Diagnosed di�usivity (cm2 s�1) in Exp. 1 (solid line), Exp. 2 (dotted line) andExp. 2 but without enhanced bottom mixing (dash-dotted line). For comparison, the param-eterization adopted for the global ocean by Hasumi and Suginohara (1999) is also displayed(long dashed line).3.2.3 Spin-up of the physical modelAll of the three model simulations were initialized by the January Levitus and Boyer (1994)and Levitus et al. (1994) climatological temperature and salinity �elds, respectively, a 2 mthick sea ice cover based on climatological sea ice extent (Gloersen et al., 1992), and an oceanat rest. The three experiments were then integrated in parallel for 110 years by applyingthe monthly mean NCAR/NCEP atmospheric forcing �elds, and temperature and salinityrelaxation, as described above.A spin-up integration of O(100 yrs) is far from su�cient to reach an annually steadystate circulation of the deepest waters in the World Ocean, but is su�cient for the upperand intermediate waters, including the rapidly ventilated deep waters of the North AtlanticOcean.3.2.4 Set-up of the CFCs simulationFollowing the atmospheric record (Fig. 3.1), the CFCs simulations begin in 1931 with zeroconcentrations in the atmosphere and the ocean. All experiments were run to the end of 1997.The ux of the CFCs at the air-sea surface are expressed as:F = Kw � (Csat � Csurf) (3.2)where F (mol m�2 s�1) is the uxes of the CFCs, Csat (mol m�3) is the saturated CFCs64

concentration in moist air near the sea surface, Kw (m s�1) is the transfer (or piston) velocity,and Csurf (mol m�3) is the modelled surface ocean CFCs concentration. The transfer velocityKw is computed using the Equation 3 of Wanninkhof (1992). Further details are given byDutay et al. (2002).3.3 Results3.3.1 Meridional overturning and zonal distribution of CFC-11As already mentioned, the CFCs are passive tracers that are redistributed by advective trans-port and dispersive mixing along the isopycnals, and by mixing across the isopycnals. Thesimulated meridional overturning circulation (MOC), illustrating the major features of themeridional thermohaline circulation cell, is useful in displaying the modelled CFCs distribu-tion.Figure 3.4 shows the temporal evolution of maximum overturning at 24�N in the NorthAtlantic during the CFC simulations. The maximum overturning is fairly stable during theintegrations. The maximum overturning occurs on the 27.52-isopycnal in Exps. 1 and 2,and on the 27.63-isopycnal in Exp. 3 (cfr. Fig. 3.5 and discussion below), implying that thenet southward ow of water masses occur with �0 > 27:52 in Exps. 1 and 2, and 27.63 inExp. 3. The average overturning rates are 13.0 Sv, 9.1 Sv and 12.8 Sv in Exps. 1, 2 and 3,respectively. The di�erence in the overturning rates illustrate the dependency on the strengthof the diapycnal (e.g., Bryan, 1987; Sun and Bleck, 2001), as well as on the strength of theisopycnal (e.g., (Schmittner and Weaver, 2001)), mixing.The simulated annual mean structure of the overturning in Exps. 1 and 3 are displayedin Fig. 3.5. The zonally averaged concentration of CFC-11 in year 1990 and the maximumML thickness are also shown in the �gure. The MOC and zonally averaged concentration ofCFC-11 are provided in both �0 and �xed depth (or z) coordinates, the latter by applyinga vertical interpolation of the isopycnals with a vertical resolution of 20 m. Overall, bothexperiments capture the so-called Global Conveyor Belt (GCB; Broecker, 1991) circulation.At the surface, the warm and light water becomes denser as it moves poleward. At highnorthern latitudes, the dense surface water sinks and ows southward at intermediate toabyssal depths to close the thermohaline circulation cell.Highest CFC-11 concentrations occur at the surface at the high latitudes mainly becauseof the wind and buoyancy generated mixing and temperature dependency on the solubility ofthe CFCs. Deep penetration of CFC-11 therefore occurs where deep surface mixing and wintertime convection are present with subsequent subduction of the ventilated waters (McCartneyand Talley, 1982; New et al., 1995). The shallow penetration of CFC-11 near the Equator is65

1930 1940 1950 1960 1970 1980 1990 2000Year

8

9

10

11

12

13

14

15

Ove

rtu

rnin

g(S

v)

Figure 3.4: Temporal variation of maximum overturning (Sv = 106 m3 s�1) along 24�N inthe North Atlantic. The mean value for Exp. 1 (solid line), Exp. 2 (dotted line) and Exp. 3(dash-dotted line) are 13.0 Sv, 9.1 Sv and 12.8 Sv, respectively.mainly due to the strong thermocline (implying weak diapycnal mixing), and consequentlyshallow ML there. The southward ow of dense water in the North Atlantic Ocean dominatesthe lower parts of the GCB, illustrating the unique properties of these water masses.The out ow of NADW with enhanced CFC-11 concentration at intermediate depths iscaptured in both experiments (panels 3.5b and d) at about 2000 m. Moreover, a deep out owof NADW with a high CFC-11 signal is captured in Exp. 3 (and to some extent in Exp. 2, notshown) at a depth of about 4000 m. Figures 3.5a and c con�rm that the southward movingdeep water masses have potential density > 27.52 in Exp. 1 (and Exp. 2, not shown), and >27.63 in Exp. 3.In Exp. 1, the out ow of NADW is about 13 Sv in the subtropical North Atlantic(Figs. 3.5a and b), which is in accordance with the estimated value of 13�2 Sv accord-ing to Ganachaud and Wunsch (2000) within the similar density range. The transport ofNADW across Equator is also about 13 Sv, which is consistent with the estimate of 14 Sv bySchmitz Jr. (1995), and is in the range of 12 to 15 Sv estimated by Dickson and Brown (1994).In the sub-surface waters, the most southward-reaching water mass tagged with CFC-11 hasa potential density between 27.71 and 27.77 (Fig. 3.5a), and the zonal average depth of thisCFC-11 tongue is 2000-2600 m at Equator (Fig. 3.5b). It is seen that the sub-surface CFC-11tongue originating from the North Atlantic has passed Equator in 1990, which implies an e�-cient ventilation and spreading of the corresponding water mass. Furthermore, the northwardconcentration gradient of the CFC-11 tongue indicates that the source of the ventilation of66

-30 0 30 60Latitude (oN)

24.1225.2826.1826.8027.2227.5227.7127.8227.9027.9828.0428.10

σ 0

1

1

1

13 119

179

-1

5

1-1

-30 0 30 60Latitude (oN)

24.1225.2826.1826.8027.2227.5227.7127.8227.9027.9828.0428.10

σ 0

(a) Exp. 1 (�0- coordinate) -30 0 30 60Latitude (oN)

0

1000

2000

3000

4000

5000

6000

Dep

th(m

) 119

5

1

1193 -1

13

-17

-1

1 -1

-30 0 30 60Latitude (oN)

0

1000

2000

3000

4000

5000

6000

Dep

th(m

)

(b) Exp. 1 (z-coordinate)

-30 0 30 60Latitude (oN)

24.1225.2826.1826.8027.2227.5227.7127.8227.9027.9828.0428.10

σ 0

0.015 0.03 0.05 0.1 0.2 0.4 0.8 1.6 3.2

CFC-11 (pmol kg-1)

1

9

3

31

5

11

13

1

7

5

11

-1

-30 0 30 60Latitude (oN)

24.1225.2826.1826.8027.2227.5227.7127.8227.9027.9828.0428.10

σ 0

(c) Exp. 3 (�0- coordinate) -30 0 30 60Latitude (oN)

0

1000

2000

3000

4000

5000

6000

Dep

th(m

)

0.015 0.03 0.05 0.1 0.2 0.4 0.8 1.6 3.2

119

119

-17

9

-1

1313

1

3

57

-30 0 30 60Latitude (oN)

0

1000

2000

3000

4000

5000

6000

Dep

th(m

)CFC-11 (pmol kg-1)

(d) Exp. 3 (z-coordinate)Figure 3.5: Zonally averaged CFC-11 concentration in 1990 (in colors, unit pmol kg�1, cut-o� value 0.005 pmol kg�1) from Exp. 1 (upper panels) and Exp. 3 (lower panels). The leftpanels show the CFC-11 concentration and overturning (1 Sv = 106 m3 s�1; 2 Sv isolines)as function of potential density and latitude. The right panels show the same quantities asfunction of depth and latitude. Solid (dashed) lines indicate clockwise (counter-clockwise)circulation looking west, and the thick lines in the right panels indicate the maximum MLthickness in the model. The black areas in the �0-panels correspond to water masses that areabsent at the given latitude.67

-10 -8 -6 -4 -2 0 2 4 6 8 10Mass transport (Sv)

-20 -16 -12 -8 -4 0 4 8 12 16 20Overturning (Sv)

24.12

25.28

26.18

26.80

27.22

27.52

27.71

27.82

27.90

27.98

28.04

28.10

σ 0

(a) Exp. 1 -10 -8 -6 -4 -2 0 2 4 6 8 10Mass transport (Sv)

-20 -16 -12 -8 -4 0 4 8 12 16 20Overturning (Sv)

24.12

25.28

26.18

26.80

27.22

27.52

27.71

27.82

27.90

27.98

28.04

28.10

σ 0

(b) Exp. 3Figure 3.6: The mass transport (Sv; bars) on each isopycnal and the overturning (Sv; solideline) across 24�N in the North Atlantic from CFC-year 1990. (a) from Exp. 1, and (b) fromExp. 3. Positive (negative) values indicate northward (southward) transports.this water mass originates north of 50�N.The overturning features of Exps. 2 (not shown) and 3 di�er from those of Exp. 1 bythe following: A deep CFC-11 tongue located at about 4000 m is present in Exps. 2 and 3,but not in Exp. 1. This �nding is further addressed in section 3.3.6. Maximum overturningis weak in Exp. 2, but consistent with Exp. 1 (and with observational based estimates) inExp. 3. In addition, the southward extension of the CFC-enriched water masses at depths of2000{2500 m is less profound in Exps. 2 and 3 compared to that in Exp. 1.To further elucidate on the relationship between the CFC-enriched water masses and theoverturning, the mass transport of each isopycnic layer and the overturning across 24�N inthe North Atlantic for CFC-year 1990 are depicted in Fig. 3.6 (the corresponding transportsfor the other years are very similar).In Exp. 1 (Fig. 3.6a), the major part of the water masses with density less than 27.52 ownorthward, and the water masses with �0 > 27:52 ow southward. As already mentioned, thewater mass 27.71 dominates the transport of the southward- owing waters. The transportrate of this water mass reaches 9.1 Sv, and contributes about 67% of the total southward masstransport across 24�N. The dominant transport rate of the 27.71-isopycnal is the reason forthe prominent southward-reaching tongue of CFC-11 in Exp. 1. The value of the overturning68

400

800

1600

200

800

1600

400

800

200

400

800

2000

50oN

70oN

60oN

80oN

0 1030 1020405060

Greenland

(a) Exp. 1, ML depth 27.52

28.04

27.63

27.63

27.6327.82

27.927.94

28.04

50oN

70oN

60oN

80oN

0 1030 1020405060

Greenland

(b) Exp. 1,outcropping isopycnals27.52

28.04

27.63

27.63

27.63

27.82

27.9

27.98

28.0427.98

50oN

70oN

60oN

80oN

0 1030 1020405060

Greenland

(c) Exp. 2, outcropping isopycnals 27.52

28.1

27.63

27.71 27.63

27.82

27.52

27.82

27.9428.04

28.0428.04

27.63

50oN

70oN

60oN

80oN

0 1030 1020405060

Greenland

(d) Exp. 3, outcropping isopycnalsFigure 3.7: Simulated ML thickness (m) (a) and location of the outcropping isopycnals (b) inApril 1990 of Exp. 1. (c) and (d) are as (b) but for Exps. 2 and 3, respectively. ML thicknessexceeding 800 m and potential densities > 27.71 are shaded.at the bottom is about 1.2 Sv. This transport originates from the ow through the BeringStrait, and is close to the observational based estimate of 1 Sv (Roach et al., 1995).For Exps. 2 (not shown) and 3, the transports are generally weaker than those in Exp. 1.There are two exceptions to this: The ML transport in Exps. 2 and 3 exceeds the transportin Exp. 1 by a factor two to three, and a second sub-surface maximum occurs on the 27.90-isopycnal.3.3.2 Ventilation of the sub-surface water massesAfter examining at the meridional penetration of CFC-11, the attention is turned to thelocation of the deep mixing in the North Atlantic, and the water masses that are ventilatedwith the ML during winter time. Figure 3.7 shows the ML thickness in April 1990 in Exp. 1,and the locations of the outcropping isopycnals in the northern North Atlantic and in theNordic Seas for all of the experiments.Deep convective mixing reaches a depth of 1600-2000 m in the Labrador and IrmingerSeas, and here the 27.63-isopycnal outcrops to the surface. Therefore, a strong CFC-11 signal69

is expected on the 27.63-isopycnal (or at a depth of 1600-2000 m). Similarly, the simulatedconvective mixing reaches a depth of 800 m in the Greenland Sea and here the 27.94-isopycnaloutcrops to the surface. As a result, strong CFC-11 signals originating from the Nordic Seasare expected on the isopycnals with potential densities greater than those of the simulatedLabrador Sea Water (LSW).The simulated ML depth of Exps. 2 and 3 are close to that of Exp. 1 (not shown). Theotcropping region of the isopycnals in Exps. 2 and 3 are also similar to those of Exp. 1(Figs. 3.7c and d). However, compared to Exp. 1, the densities of the outcropping layersare slightly higher in Exp. 2 and, for the Nordic Seas, signi�cantly higher in Exp. 3. Thedi�erence in the vertical density structure in the Nordic Seas between the three experimentshas important implications for the simulated OW characteristics as discussed in section 3.3.6.3.3.3 Ventilation of the thermoclineThe panels in Fig. 3.8 show the observed and simulated concentrations of CFC-11 and thevertical position of the 27.52 and 27.77 isopycnals along the Tropical Atlantic Study (TAS),Leg 2 (Weiss et al., 1985). For Exp. 3, the simulated distributions of CFC-11 in the upperwaters resemble the observation closely, whereas Exp. 1 is clearly too di�usive. It is also seenthat, even after a total integration time of 162 years, the position of the 27.52 isopycnal inExp. 3 is almost indistinguishable from the climatological mean position (the latter based onLevitus and Boyer (1994) and Levitus et al. (1994)). For comparison, the 27.52 isopycnal hasdescended by about 100 m in Exps. 1 and 2.Based on the simulated vertical density strati�cation, the strength of the applied diapyc-nal di�usivity in the upper water column is about 0.16�10�4 m2 s�1 in Exp. 3, or within theestimated value of (0.11-0.17)�10�4 m2 s�1 based on a purposefully tracer release experimentin the thermocline of the subtropical Atlantic Ocean (Ledwell & Watson, 1993; Ledwell et al.,1998), and also similar to the value adopted by Hasumi and Suginohara (1999) (see Fig. 3.3).In Exp. 1, the diagnosed di�usivity is about 1.5 �10�4 m2 s�1, or more than one order ofmagnitude above the cited Ledwell and Watson (1993) and Ledwell et al. (1998) values. Thecorresponding value for Exp. 2 is 0.2�10�4 m2 s�1. Thus, the ventilation of the thermoclinein the subtropical Atlantic is highly sensitive to the strength of the diapycnal mixing. Fur-thermore, the di�erence between Exps. 2 and 3 show that also the applied isopycnal mixingis of importance for the vertical distribution of the tracer (see below). For the applied modelsystem, it follows that Kd = 5 � 10�8=N m2 s�1 and vtrac = 0:0025 m s�1 yield a realisticventilation rate of the thermocline waters in the sub-tropical Atlantic Ocean.70

0

1000

2000

3000

Dep

th(m

)

0.8

0.03

0.015

0.8

0.012

0.2 0.10.05

0.4

5 10 15 20Latitude (oN)

ObservationTAS/Leg2

5 10 15 20Latitude (oN)

ObservationTAS/Leg2

-80 -60 -40 -20 00

30

60

TAS/Leg2

(a) Observed CFC-11 along TAS/Leg 20.2

0.030.05 0.03

0.012

0.80.8

0.1

0.015

5 10 15 20Latitude (oN)

0

1000

2000

3000

Dep

th(m

)

5 10 15 20Latitude (oN)

0

1000

2000

3000

Dep

th(m

)(b) CFC-11 in Exp. 1

0.05

0.80.8

0.050.03

0.012

5 10 15 20Latitude (oN)

0

1000

2000

3000

Dep

th(m

)

0.005

5 10 15 20Latitude (oN)

0

1000

2000

3000

Dep

th(m

)

(c) CFC-11 in Exp. 20.05

0.80.8

0.03

0.012

0.012

5 10 15 20Latitude (oN)

0

1000

2000

3000

Dep

th(m

)

5 10 15 20Latitude (oN)

0

1000

2000

3000

Dep

th(m

)

(d) CFC-11 in Exp. 3Figure 3.8: Observed (a) and simulated (b, c, d) CFC-11 concentrations (pmol kg�1) alongTAS/Leg 2 (inlet in panel a) in early 1983 (Weiss et al., 1985). The cut-o� value is 0.01 pmolkg�1, with the indicated 0.005 pmol kg�1 as concentration of the sub-surface maximum inExp. 2 (panel c). The long dashed lines show the position of the 27.52 and 27.77 isopycnals.71

3.3.4 Ventilation of the sub-surface water massesThe Upper Labrador Sea Water (ULSW) is one component of the NADW (Bower and Hunt,2000). The presence of recently formed and CFC-enriched ULSW in the Tropical Atlantichas been discussed, for instance, by Weiss et al. (1985) and Andrie et al. (1999). In theobservation (Fig. 3.8a), the out ow of ULSW, which is believed to be formed in the SouthernLabrador Sea (Pickart, 1992), is clearly present by the core with high CFC-11 concentrationnear 1700 m depth (Weiss et al., 1985) and with potential density of close to 27.77 (the lattercalculated from Levitus and Boyer (1994) and Levitus et al. (1994)).The simulated CFC-11 distributions yield a sub-surface core of CFC-11, but there areapparent di�erences compared to the observation: For Exps. 1 and 2, the simulated coreshave a potential density of 27.71 (or 0.06 �0 units below the observed density), whereasthe density of the core of Exp. 3 matches the observation. For all of the experiments, thedepths of the cores of CFC-enriched water are located at about 2200 m depth (or 600 mbelow the observation). In addition, the simulated CFC-11 concentrations are signi�cantlyunderestimated.The simulated strati�cation in the uppermost 1100 m (�0 < 27:52) is, as already described,quite close to the climatological density strati�cation. However, the strati�cation in the deeperwaters (�0 > 27:52) has experienced a drift, which will be addressed in Sec. 3.4. For instance,the depth of 27.77-isopycnal is about 1500 m in the climatology, 2900 m in Exp. 1, 2600 min Exp. 2, and 2200 m in Exp. 3.To illustrate the presence of both ULSW and OW using CFC-11, the observed and simu-lated CFC-11 along 24�N in the North Atlantic, together with the corresponding distributionof the 27.77 and 27.90 isopycnals, are displayed in Fig. 3.9. In the observation, two sub-surfacemaxima of CFC-11 are present, one at about 1500 m, the other between 3000 m and 4500 m.These sub-surface CFC-11 maxima correspond to the out ow of ULSW and OW, respectively(Smethie Jr et al., 2000). The water mass corresponding to the shallow CFC-11 maximum haspotential density of about 27.77 (cfr. Fig. 3.8a), and the deep CFC-11 maximum has potentialdensity > 27:86 (calculated from Levitus and Boyer (1994) and Levitus et al. (1994)).In Exp. 1, the model shows a CFC-11 maximum at the depth of 2000 m, but no deep core.The sub-surface maximum of CFC-11 corresponds to water with potential density of 27.71,which is the same as in the section along TAS/Leg 2 in Fig. 3.8b. An additional mismatch ispresent in the CFC-11 penetration depth in the Eastern North Atlantic Ocean (NEA), wherethe simulated CFC-11 signal penetrates far deeper than in the observation. The situation forExp. 2 is close to that of Exp. 1, and is still far from the observed distribution.A signi�cant improvement is obtained in Exp. 3. Here a CFC-maximum is found on the72

0.4

0.2

0.1

0.03

0.10.015

1.6

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

Longitude (oE)

1000 1000

2000 2000

3000 3000

4000 4000

5000 5000

6000 6000

Dep

th(m

)0.015 0.03 0.05 0.1 0.2 0.4 0.8 1.6 3.2

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

Longitude (oE)

1000 1000

2000 2000

3000 3000

4000 4000

5000 5000

6000 6000

Dep

th(m

)

(a) Observed CFC-11 along 24�N in August, 19920.2

0.10.05

1.6

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

Longitude

1000 1000

2000 2000

3000 3000

4000 4000

5000 5000

6000 6000

Dep

th(m

)

0.015 0.03 0.05 0.1 0.2 0.4 0.8 1.6 3.2

-70 -60 -50 -40 -30 -20Longitude

1000

2000

3000

4000

5000

6000

Dep

th(m

)

(b) CFC-11 in Exp. 10.2 0.1

1.6

0.05 0.03

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

Longitude

1000 1000

2000 2000

3000 3000

4000 4000

5000 5000

6000 6000

Dep

th(m

)

0.015 0.03 0.05 0.1 0.2 0.4 0.8 1.6 3.2

-70 -60 -50 -40 -30 -20Longitude

1000

2000

3000

4000

5000

6000

Dep

th(m

)

(c) CFC-11 in Exp. 20.2 0.015

1.6

0.4

0.03

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

Longitude

1000 1000

2000 2000

3000 3000

4000 4000

5000 5000

6000 6000

Dep

th(m

)

0.015 0.03 0.05 0.1 0.2 0.4 0.8 1.6 3.2

-70 -60 -50 -40 -30 -20Longitude

1000

2000

3000

4000

5000

6000

Dep

th(m

)

(d) CFC-11 in Exp. 31.8

1.8 1.6

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

Longitude

1000 1000

2000 2000

3000 3000

4000 4000

5000 5000

6000 6000

Dep

th(m

)

-70 -60 -50 -40 -30 -20Longitude

1000

2000

3000

4000

5000

6000

Dep

th(m

)

(e) CFC-11/CFC-12 in Exp. 21.8

1.8

1.8

1.6

1.4

1.2

1.2

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

Longitude

1000 1000

2000 2000

3000 3000

4000 4000

5000 5000

6000 6000

Dep

th(m

)

-70 -60 -50 -40 -30 -20Longitude

1000

2000

3000

4000

5000

6000

Dep

th(m

)

(f) CFC-11/CFC-12 in Exp. 3Figure 3.9: (a) Observed and (b, c, d) simulated CFC-11 concentration (in grey shading,unit pmol kg�1, cut-o� value 0.01 pmol kg�1), and (e, f) the diagnosed CFC-11/CFC-12ratio along WOCE Section A05 along 24�N in August, 1992 (Bryden et al., 1996). The longdashed lines show the position of the 27.77 and 27.90 isopycnals.73

western part of the basin with potential density of 27.77. Apparently, the second sub-surfaceCFC maximum between 3000 m and 4500 m is not present in Exp. 3. However, OW formsthe lower part of the CFC-plume on the western side of the section. This is illustrated inFig. 3.9f, displaying the CFC-11/CFC-12 ratio of the CFC-enriched water masses. Since theatmospheric CFC-11/CFC-12 ratio increases with time (see Fig. 3.1b), the more recentlyventilated water masses have a higher CFC-11/CFC-12 ratio than older water masses. Itfollows from Fig. 3.9f that the lower limb of the plume at a depth of about 4000 m and withdensity of 27.90 is recently ventilated, and it will be shown in section 3.3.6 that this watermass originates from the Nordic Seas. Also the CFC-enriched water in the eastern side of thebasin (Fig. 3.8d) originates from the Nordic Seas (section 3.3.6). For comparison, the CFC-11/CFC-12 ratio of the CFC-enriched sub-surface water in Exp. 2 show no indications of asecondary sub-surface CFC maximum (Fig. 3.9e). The observed CFC- 11/CFC-12 ratio isconsistent with the �ndings in Fig. 3.9f (not shown), with the exception of the deep CFC-plume in the NEA in the simulation.3.3.5 Spreading of the Labrador Sea waterGiven that the simulated surface mixing in the subtropical North Atlantic can only reach thedepth of several hundred meters (panels 3.5b and d), the presence of the excessive CFC-11penetration in the NEA (and partly also in the NWA) is attributed to too strong isopycnictransport or mixing. To check this, the distribution of CFC-11 are examined on the ventilatedisopycnals.Figure 3.10 displays the observed and the simulated CFC-11 concentrations in early 1990on the isopycnal in the centre of the shallow sub-surface core. The distribution of CFC-11shows that the source water originates form the Labrador and Irminger Seas (and, caused bythe lack of observations, possibly further upstream into the Nordic Seas) in the observations(Smethie Jr et al., 2000), and in the Labrador, Irminger and Nordic Seas in the simulations.The simulated and observed concentrations of CFC-11 in the sub-polar source region are com-parable and all simulations capture the observed southward transport of CFC-11. However,the observed spreading of CFC-11 along the western boundary is faster than simulated, inparticular for Exp. 1.In Exp. 1, the simulated equator-ward speed of ULSW in the Sub-tropical and TropicalAtlantic is less than 0.02 m s�1, which is below the observed estimates of 0.02 to 0.04 m s�1(Bower and Hunt, 2000). The simulated depth of the isopycnic interfaces is also displayed inthe �gure. This clearly shows that the CFC-11 core is at the depth between 2200 and 2300 min the Subtropical and Tropical Atlantic, as was shown in the vertical sections of the CFC-11distribution (Fig. 3.8b). 74

-80

-80

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

Longitude

-20 -20

0 0

20 20

40 40

60 60

80 80

Lat

itude

(o N)

-80 -70 -60 -50 -40 -30 -20 -10-20

0

20

40

60

80

(a) Observed CFC-11 -80

-80

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

-20 -20

0 0

20 20

40 40

60 60

80 80

Lat

itude

(o N)

-80

-80

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

-20 -20

0 0

20 20

40 40

60 60

80 80

Lat

itude

(o N)

-80

-80

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

Longitude

2200

23002200

2300

2300

(b) CFC-11 in Exp. 1

-80

-80

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

-20 -20

0 0

20 20

40 40

60 60

80 80

Lat

itude

(o N)

-80

-80

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

-20 -20

0 0

20 20

40 40

60 60

80 80

Lat

itude

(o N)

0.015 0.03 0.05 0.1 0.2 0.4 0.8 1.6 2.4 3.2

-80

-80

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

Longitude

-20 -20

0 0

20 20

40 40

60 60

80 80

Lat

itude

(o N) 2000

2100

2050

2050(c) CFC-11 in Exp. 2 -80

-80

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

Longitude

-20 -20

0 0

20 20

40 40

60 60

80 80

Lat

itude

(o N)

-80

-80

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

Longitude

-20 -20

0 0

20 20

40 40

60 60

80 80

Lat

itude

(o N)

0.015 0.03 0.05 0.1 0.2 0.4 0.8 1.6 2.4 3.2

-80

-80

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

Longitude

-20 -20

0 0

20 20

40 40

60 60

80 80

Lat

itude

(o N)

2200

2300

(d) CFC-11 in Exp. 3Figure 3.10: CFC-11 concentration (pmol kg�1, cut-o� value is 0.01 pmol kg�1) in the out owof UNADW in early 1990. In panel (a), observed CFC-11 (Smethie Jr et al., 2000) is givenwith stations included as black dots. Black contour line denotes the depth of the isopycnal,grey color shows the presence of the simulated �0 = 27:71 water mass in Exps. 1 and 2, and�0 = 27:77 in Exp. 3.75

-80

-80

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

-20 -20

0 0

20 20

40 40

60 60

80 80

Lat

itude

(o N)

0.2 0.5 1 2 5

Figure 3.11: Ratio of the apparent di�usive velocity to advective velocity on the 27.71-isopycnal in Exp. 1.In Exps. 2 and 3, the simulated equator-ward velocity at the ULSW level are < 0.01 ms�1 and < 0.02 m s�1, respectively. Consequently, Exp. 2 has a simulated CFC-11 tonguethat is even more sluggish than that of Exp. 1, whereas Exp. 3 gives the most realistic (butstill underestimated) transport along the western boundary.Figure 3.10b and c indicate that the reason for the deep presence of CFC-11 in the centraland eastern North Atlantic at 24�N as seen in Fig. 3.9b and c is caused by isopycnal transportor mixing of the waters originating from the the high northern latitudes.The apparent di�usive velocity ~Vdi� (m s�1)on an isopycnic layer can be diagnosed as(Smolarkiewicz, 1984) ~Vdi� = KisohC riso(hC) (3.3)where Kiso (m2 s�1), riso (m�1), h (m) and C (mol m�3) are the di�usion coe�cient, theisopycnic gradient operator, the layer thickness and the CFCs concentration on the actualisopycnal, respectively. The ratio between the advective and the apparent di�usive velocityon the 27.71-isopycnal in Exp. 1 is shown in Fig. 3.11. It is seen that the advective transportdominates in the OW regions and along the DWBC in the model, whereas the apparentdi�usive transport is responsible for the spreading of the CFCs into the central and easternNorth Atlantic Basin. The latter �nding indicates that the applied isopycnal di�usive velocityof 0.01 m s�1 is too large by a factor two to �ve. This also explains why Exp. 3, with ahorizontal di�usive velocity of one fourth compared to Exps. 1 and 2, greatly improves thesimulated distribution of the sub-surface CFC-enriched water masses.76

-80

-80

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

0

0

Longitude

-20 -20

0 0

20 20

40 40

60 60

80 80L

atitu

de(o N

)

-80 -70 -60 -50 -40 -30 -20 -10 0-20

0

20

40

60

80

(a) Observed CFC-11 -80

-80

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

-20 -20

0 0

20 20

40 40

60 60

80 80

Lat

itude

(o N)

-80

-80

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

-20 -20

0 0

20 20

40 40

60 60

80 80

Lat

itude

(o N)

-80

-80

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

Longitude

44004400

40004000

4000

(b) Exp. 1

-80

-80

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

Longitude

-20 -20

0 0

20 20

40 40

60 60

80 80

Lat

itude

(o N)

-80

-80

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

-20 -20

0 0

20 20

40 40

60 60

80 80

Lat

itude

(o N)

0.015 0.03 0.05 0.1 0.2 0.4 0.8 1.6 2.4 3.2

-80

-80

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

Longitude

-20 -20

0 0

20 20

40 40

60 60

80 80

Lat

itude

(o N)

3800

3800

3600

3800

(c) Exp. 2 -80

-80

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

-20 -20

0 0

20 20

40 40

60 60

80 80L

atitu

de(o N

)

-80

-80

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

-20 -20

0 0

20 20

40 40

60 60

80 80L

atitu

de(o N

)

0.015 0.03 0.05 0.1 0.2 0.4 0.8 1.6 2.4 3.2

-80

-80

-70

-70

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

Longitude

-20 -20

0 0

20 20

40 40

60 60

80 80L

atitu

de(o N

)

3600

3700

3700 3600

3600

(d) Exp. 3Figure 3.12: As Fig. 3.10, but for the OW.3.3.6 Spreading of the Over ow WaterFigure 3.12 shows the observed and simulated CFC-11 of the OW. In Exp. 1, the CFC-11tongue reaches about 50�N, which is far less than in the observation. In addition, most ofthe OW remains in the NEA, contrary to the main observed pathway of these water masses.In fact, it is believed that most of the Iceland-Scotland Ridge (ISR) OW ows southwardand parallel to the Reykjanes Ridge, crossing the North Atlantic Ridge at the Charlie-GibbsFracture Zone at 57 �N, and thereafter continues along the continental slope of the NWA asa part of the DWBCs.In Exp. 2, the out ow of CFC-enriched OW reaches about 20�N, with the out ow throughthe Denmark Strait ventilating the NWA, and the out ow across the ISR ventilating the NEA.77

Again Exp. 3 is closest to the observed distribution of the CFCs. Here the OW throughthe Denmark Strait, and partly also the OW across the ISR, ow southward in the NWA asthe deep part of the DWBC. A signi�cant fraction of the ISR OW is, however, still beingfound in the NWA.3.4 Discussion and conclusionTo illuminate the decadal scale ventilation of the North Atlantic surface waters, and to performa �rst order assessment of the response of the ventilation to diapycnal and isopycnal mixing,the atmospheric tracers CFC-11 and CFC-12 have been embedded into a global version ofthe isopycnic coordinate ocean model MICOM.Three model experiments have been performed: The base-line experiment (Exp. 1) ischaracterized by a diapycnal mixing coe�cient Kd = 3 � 10�7=N (m2 s�1; N (s�1) is theBrunt-V�ais�al�a frequency), and an isopycnal di�usion velocity vtrac = 0:01 m s�1. The secondexperiment (Exp. 2) is identical to Exp. 1 but withKd = 5�10�8=N plus an enhanced mixingnear the sea oor. The third experiment (Exp. 3) follows Exp. 2, but with vtrac = 0:0025 ms�1. The three experiments, forming two twin experiments, show that the local to basin scaleocean dynamics and the associated transport and mixing of the CFCs are highly sensitive tothe strength of the diapycnal and isopycnal mixing parameterizations.It is shown that it is generally needed to consider the combined, rather than the isolated,e�ect of the diapycnal and isopycnal mixing. For instance, adjustments of the diapycnal(or vertical) di�usion coe�cient can be used to tune the simulated strength of the MOC.However, changes in the isopycnal (or horizontal) di�usion coe�cient will also induce changesin the strength of the MOC (cfr. Fig. 3.4). It is therefore not given which combination of thediapycnal and isopycnal mixing coe�cients that yield both a realistic MOC and a realisticventilation of the surface and sub-surface water masses.By introducing the CFCs as tracers to the model system, an additional and potentiallypowerful constraint is added to the model validation. From the present study, this is especiallyevident for the ventilation of the tropical thermocline (Figs. 3.8b and d, and Figs. 3.9b and d),the spreading of the sub-polar and the lighter over ow waters (Fig. 3.10), and the OW formingthe DWBC (Fig. 3.12). The conclusion from the present model exercise is that the experimentwith both reduced diapycnal and isopycnal di�usivity (Exp. 3) is superior to the other modelexperiments for all of the model-data comparisons that have been assessed.The main di�erences between the simulated features of Exp. 3 and the observed CFCs arelinked to too weak and sluggish DWBCs in the model. The problem with the lower part ofthe DWBC is related to a part of the source waters for this current, namely the part of the78

Nordic Seas OW that ows across the ISR. It was demonstrated in section 3.3.6 that all of themodel experiments show a tendency for the ISR OW to be con�ned to the NEA, whereas themajor part of this water should cross the North Atlantic Ridge at the Charlie-Gibbs FractureZone at 57 �N.The reason for this model de�ciency is related to the vertical position of the isopycnalscarrying the OW. If the isopycnals with the OW are located deeper than the North AtlanticRidge in the model, the OW will necessarily be con�ned to the NEA. It follows from Fig. 3.9that the vertical position of the 27.90 isopycnal is too deep in all experiments. The situationis clearly poorest for Exp. 1, whereas Exp. 3 is slightly improved compared to Exp. 2.The amount and characteristics of the OW down-stream of the Greenland-Scotland Ridgeare determined by at least three factors: The formation rate of intermediate to deep waters inthe Nordic Seas (plus the contribution of sub-surface waters from the Arctic owing throughthe Fram Strait), the amount of ambient water entrained into the OW, and the diapycnalmixing acting on the OW.In the absence of formation of intermediate to deep waters and in the presence of somediapycnal mixing, the OWwill disappear, and the deepest isopycnals will dry out and thereforedescend in the water column. The OW isopycnals will also experience a net loss of mass andtherefore descend in the water column if the diapycnal mixing ux exceeds the transport rateof the corresponding OW over the Greenland-Scotland Ridge. This is clearly the situation inExp. 1; here the diapycnal mixing far exceeds the formation of the OW, leading to an e�cientconversion of dense to less dense water masses (cfr. the 27.90 isopycnal in Fig. 3.9a and b).The combined e�ect of isopycnal and diapycnal di�usion on changes in the vertical densitystrati�cation is exempli�ed in Fig. 3.13. The �gure shows the vertical layer drift (positivevalues means deepening of the isopycnals) across the DWBC at 24 �N between CFC-year 1992and 1931. It follows that the layer drift is signi�cantly reduced in Exps. 2 and 3 compared tothat of Exp. 1.Further up-stream, the 27.90 isopycnal in Exp. 3 is located su�ciently high in the watercolumn (Fig. 3.9d) that part of the OW is able to cross the North Atlantic Ridge (Fig. 3.12d).However, a part of the OW in the NEA is not able to cross the ridge, leading to a sub-surfaceCFC-maximum in the basin (Fig. 3.9d and e), and consequently to a reduced intensity of thesimulated DWBC.A realistic representation of the DWBC in OGCMs is further complicated by a usuallycoarse horizontal model resolution, and consequently a model bathymetry that most likely donot resolve important topographic features as the Faroe-Bank Channel and the Charlie-GibbsFracture Zone.Further complications are linked to the fact that mixing across surfaces of constant density79

0 100 200 300 400 500 600 700Vertical layer drift (m)

24.12

25.28

26.18

26.80

27.22

27.52

27.71

27.82

27.90

27.98

28.04

28.10

σ 0

Figure 3.13: Mean vertical layer drift (m) during the CFC integrations (i.e., the di�erencebetween CFC-year 1992 and 1931) across the DWBC along A05 section at 24�N in the NorthAtlantic. Exp. 1 in dased line, Exp. 2 in dotted line, and Exp. 3 with bars.80

(or neutral surfaces) are generated by a variety of processes, including mixing caused bydissipation of energy from internal waves in the open ocean, mixing against topography, andmixing generated by (surface, internal or near sea bed) shear ows.In addition, the natural variability of the mixing in the sup-polar region is particularlyimportant to the ventilation of the deep North Atlantic Ocean, and this variability contributessigni�cantly to the variability of the MOC (Curry & McCartney, 2001). A consequence ofthe applied climatological monthly mean atmospheric forcing �elds is a fairly constant MOCin the simulations presented here (see Fig. 3.4). In reality, year-to-year variability in theforcing, possibly in combination of internal ocean circulation modes (Bentsen, 2002), leads tosigni�cant variations in the MOC. Therefore, a model system like the system presented here,should also be validated based on the use of synoptic (i.e., daily) atmospheric forcing �elds.Such integrations are ongoing, and will be presented elsewhere.The parameterization for the diapycnal mixing used in this study is simple, but is stillmore elaborate than �xed values used in many present day climate OGCMs (see for instancediscussion in Nilsson and Walin, 2001). A more complete description of the diapycnal mixingwould include parameterizations of, for instance, bottom gravity currents and tidal mixingover rough topography, which is beyond the scope of this paper.It is concluded that the presented simulations provide detailed insight into the e�ectthe isopycnal and diapycnal di�usion parameterizations have on surface to abyss ventilation,dispersive mixing and advective spreading on some of the key water masses in the NorthAtlantic. It is encouraging that the simulated distribution of the CFCs in Exp. 3 are asclose to the observed distributions as is the case. Furthermore, the major model de�ciencies(i.e., too sluggish DWBCs) are believed to be linked to an intricate interplay between modelresolution, model forcing and possibly unresolved mixing processes in the model. Therefore,continuous improvements and re�nements of this and similar model systems are required toimprove the model performance. Irrespective of the detected problems, the CFC-study showpromising prospects for simulating multi-annual to dedacal scale variability of the presentday climate system, and thereby also for assessing possible changes in the variability of themarine climate system as the human induced global warming continues.3.5 AcknowledgmentsThis study has been supported by the G. C. Rieber Foundations, Norsk Hydro as, the Norwe-gian Research Council through the RegClim and the Programme of Supercomputing projects,and the project GOSAC under the EC Environment and Climate Programme. We are gratefulto Drs. W. M. Smethie, R. F. Weiss, P. Salameh, and G. Parrilla for providing CFC-data from81

the cruises used in the study. The reviewers' comments greatly improved the manuscript.

82

ReferencesAndrie, C., Ternon, J., Bourles, B., Gouriou, Y., and Oudot, C. 1999. Tracer distributions and deep circulationin the western tropical Atlantic during CITHER 1 and ETAMBOT cruises, 1993-1996. J. Geophys. Res.,104, C9, 21195-21215.Arakawa, A., and Lamb, V. 1977. Computational design of the basic processes of the UCLA General CirculationModel. Methods Comput. Phys., 17, 174-265.Bentsen, M., Evensen, G., Drange, H., and Jenkins, A. D. 1999. Coordinate Transformation on a Sphere UsingConformal Mapping. Mon. Wea. Rev., 127, 2733{2740.Bentsen, M., and Drange, H. 2000. Parameterizing surface uxes in ocean models using the NCEP/NCARreanalysis data. Tech. Rep. No. 4. Nansen Environmenal and Remote Sensing Center, Bergen, Norway.Bentsen, M. 2002. Modelling Ocean Climate Variability of the North Atlantic and the Nordic Seas. Doctoraldissertation, Univ. of Bergen, Bergen, Norway.Bleck, R., Rooth, C., Hu, D., and Smith, L. T. 1992. Salinity-driven thermohaline transients in a wind-and thermohaline-forced isopycnic coordinate model of the North Atlantic. J. Phys. Oceanogr., 22,1486-1515.Bower, A., and Hunt, H. 2000. Lagrangian observations of deep western boundary current in the north atlanticocean. part 1: Large-scale pathways and spreading rates. J. Phys. Oceanogr., 30, 764-783.Broecker, W. 1991. The great ocean conveyor. Oceanography, 4, 79-89.Bryan, F. 1987. Parameter sensitivity of primitive equation ocean general circulation models. J. Phys.Oceanogr., 17, 970-985.Bryden, H., Gri�ths, M., Lavin, A., Millard, R., Parrilla, G., and Smethie, W. 1996. Decadal Changes inWater Mass Characteristics at 24�N in the Subtropical North Atlantic Ocean. J.Climate, 9, 3162-3186.Curry, R., and McCartney, M. S. 2001. Ocean gyre circulation changes associated with the North AtlanticOscillation. J. Phys. Oceanogr., 31, 3374{3400.Dickson, R. R., and Brown, J. 1994. The production of North Atlantic Deep Water: Sources, rates andpathways. J. Geophys. Res., 99, C6, 12319{12341.Dixon, K., Bullister, J., Gammon, R., and Stou�er, R. 1996. Examining a coupled climate model using CFC-11as an ocean tracer. Geophys. Res. Letters, 23, 1957-1960.Drange, H., and Simonsen, K. 1996. Formulation of air-sea uxes in the ESOP2 version of MICOM. Tech.Rep. No. 125. Edv. Griegsv. 3A, N-5059 Solheimsviken, Norway: Nansen Environmental and RemoteSensing Center. 83

Dutay, J., Bullister, J., Doney, S., Orr, J., Najjar, R., Caldeira, K., Campin, J., Drange, H., Follows, M., Gao,Y., Gruber, N., Hecht, M., Ishida, A., Joos, F., Lindsay, K., Madec, G., Marier-Reimer, E., Marshall,J., Matear, R., Monfray, P., Plattner, G., Sarmiento, J., Schlitzer, R., Slater, R., Totterdell, I., Weirig,M., Yamanaka, Y., and Yool, A. 2002. Evaluation of ocean model ventilation with CFC-11: comparisonof 13 global ocean models. Ocean Modelling, 4, 89-120.Eden, C., and Jung, T. 2001. North Atlantic Interdecadal Variability: Oceanic Response to the North AtlanticOscillation. J. Climate, 14, 676-691.England, M., and Holloway, G. 1998. Simulations of CFC content and water mass age in the deep NorthAtlantic. J. Geophys. Res., 103, C8, 15885-15901.England, M., and Maier-Reimer, E. 2001. Using Chemical Tracers to Assess Ocean Models. Rev. Geophys.,39, 29-70.Fairall, C. W., Bradley, E. F., Rogers, D. P., Edson, J. B., and Young, G. S. 1996. Bulk parameterization ofair-sea uxes for Tropical Ocean-Global Atmosphere Coupled-Ocean Atmosphere Response Experiment.J. Geophys. Res., 101, C2, 3747{3764.Friedrich, H., and Levitus, S. 1972. An approximation to the equation of state for sea water, suitable fornumerical ocean models. J. Phys. Oceanogr., 2, 514-517.Furevik, T., Bentsen, M., Drange, H., Johannessen, J. A., and Korablev, A. 2002. Temporal and spatialvariability of the sea surface salinity in the Nordic Seas. J. Geophys. Res. in pressGanachaud, A., and Wunsch, C. 2000. Improved estimates of global ocean circulation, heat transport andmixing from hydrographic data. Nature, 408, 453-457.Gargett, A. 1984. Vertical eddy in the ocean interior. J. Marine. Res., 42, 359-393.Gaspar, P. 1988. Modeling the seasonal cycle of the upper ocean. J. Phys. Oceanogr., 18, 161{180.Gloersen, P., Campbell, W. J., Cavalieri, D. J., Comiso, J. C., and Zwally, C. L. P. H. J. 1992. Artic andAntarctic sea ice, 1978{1987. National Aeronautics and Space Administration, Washington, D.C., USA.Gordon, A. 1986. Interocean exchange of thermocline water. J. Geophys. Res., 91, 5037-5046.Hansen, B., and �sterhus, S. 2000. North Atlantic-Nordic Seas exchanges. Prog. Oceanogr., 45 (2), 109-208.Harder, M. 1996. Dynamik, Rauhigkeit und Alter des Meereises in der Arktis. Doctoral dissertation, Alfred-Wegner-Institut f�ur Polar- ind Meeresforschung, Bremerhaven, Germany.Hartman, D. L. 1994. Global Physical Climatology. Academic Press Inc., U.S.A. p. 171-203Hasumi, H., and Suginohara, N. 1999. E�ects of locally enhanced vertical di�usivity over rough bathymetryon the world ocean circulation. J. Geophys. Res., 104, C10, 23367-23374.Hibler, W. 1979. A dynamic thermodynamic sea ice model. J. Phys. Oceanogr., 9, 815-846.Jones, E., Rudels, B., and Anderson, L. 1995. Deep waters of the Arctic Ocean: origins and circulation. DeepSea Res., 42, 737-760.Kalnay, E., et al.. 1996. The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77 (3),437{471.Ledwell, J. R., and Watson, A. J. 1993. Evidence for slow mixing across the pycnocline from an open oceantracer-release experiment. Nature, 364, 701-703.Ledwell, J. R., Watson, A. J., and Law, C. S. 1998. Mixing of a tracer in the pycnocline. J. Geophys. Res.,103, C10, 21449-21529. 84

Ledwell, J. R., Montgomery, E., Polzin, K., Laurent, L., Schmitt, R., and Toole, J. 2000. Evidence for enhancedmixing over rough topography in the abyssal ocean. Nature, 403, 179-182.Levitus, S., and Boyer, T. P. 1994. World Ocean Atlas 1994 Volume 4: Temperature. NOAA Atlas NESDIS4. Washington, D.C., USA.Levitus, S., Burgett, R., and Boyer, T. P. 1994. World Ocean Atlas 1994 Volume 3: Salinity. NOAA AtlasNESDIS 3. Washington, D.C., USA.Manabe, S., and Stou�er, R. 1999. The role of thermohaline circulation in climate. Tellus, 51A-B, 92-109.McCartney, M. S., and Talley, L. D. 1982. The subpolar mode water of the North Atlantic Ocean. J. Phys.Oceanogr., 12, 1169{1188.McDougall, T., and Dewar, W. 1998. Vertical mixing, cabbeling and thermobaricity in layered models. J.Phys. Oceanogr., 1458-1480.Munk, W., and Wunsch, C. 1998. Abyssal recipes ii: energetics of tidal and wind mixing. Deep Sea Res.,45(I), 1977-2010.New, A., Bleck, R., Jia, Y., Marsh, R., Huddleston, M., and Bernard, S. 1995. An ispoycnic model study ofthe North Atlantic. Part I: Model experiment. J. Phys. Oceanogr., 25, 2679{2711.Nilsson, J., and Walin, G. 2001. Freshwater forcing as a booster of thermohaline circulation. Tellus, 53A,628-640.Pickart, R. 1992. Water mass components of North Atlantic deep western boundary current. Deep Sea Res.,Part A, 39, 1553-1572.Pickart, R., Smethie Jr, W., Lazier, J., Jones, E., and Jenkins, W. 1996. Eddies of newly formed upperLabrador Sea Water. J. Geophys. Res., 101, 20711-20726.Rahmstorf, S., and Ganopolski, A. 1999. Long-term global warming scenarios computed with an e�cientcoupled climate model. Climatic Change, 43, 353-367.Roach, A., Aagaard, K., Pease, C., Salo, S., Weingartner, T. ., Pavlov, V., and Kulakov, M. 1995. Directmeasurements of transport and water properties through the Bering Strait. J. Geophys. Res., 100,18443-18457.Rudels, B., Jones, E., Anderson, L., and Kattner, G. 1994. On the Intermediate Depth Waters of the ArcticOcean. In: O. Johannessen, R. Muench, and J. Overland (Eds.), The Polar Oceans and Their Rolein Shaping the Global Environment: The Nansen Centennial Volume (Vol. 85 of AGU GeophysicalMonograph, p. 33-46). American Geophysical Union.Schmittner, A., and Weaver, A. 2001. Dependence of multiple climate states on ocean mixing parameters.Geophys. Res. Letters, 28, 1027-1030.Schmitz Jr., W. J. 1995. On the interbasin-scale thermohaline circulation. Rev. Geophys., 33, 151{173.Smethie Jr, W. M., Fine, R., Putzka, A., and Jones, E. 2000. Tracing the ow of north atlantic deep waterusing chloro uorocarbons. J. Geophys. Res., 105, C6, 14297-14323.Smolarkiewicz, P. K. 1984. A Fully Multidimensional Positive De�nite Advection Transport Algorithm withSmall Implicit Di�usion. J. Comp. Phys., 54, 325{362.Smolarkiewicz, P. K., and Clark, T. L. 1986. The multidimensional positive de�nite advection transportalgorithm: Further development and applications. J. Comp. Phys., 67, 396-438.85

Sun, S., and Bleck, R. 2001. Thermohaline Circulation Studies with an Isopycnic Coordinate Ocean Model.J. Phys. Oceanogr., 31, 2761-2782.Toole, J. M., Polzin, K. L., and Schmitt, R. W. 1994. Estimates of Diapycnal Mixing in the Abyssal Ocean.Science, 264, 1120{1123.Walker, S., Weiss, R., and Salameh, P. 2000. Reconstructed histories of the annual mean atmospheric molefractions for halocarbons CFC-11, CFC-12, CFC-113, and carbon tetrachloride. J. Geophys. Res., 105,C6, 14285-14296.Wanninkhof, R. 1992. Relationship Between Wind Speed and Gas Exchange Over the Ocean. J. Geophys.Res., 97, C5, 7373-7382.Warren, B. 1981. Deep circulation of the world ocean. In B.A. Warren and C. Wunsch (Ed.), Evolution ofPhysical Oceanography, Scienti�c Surveys in Honor of Henry Stommel (p. 6-41). MIT Press.Weiss, R. F., Bullister, J. L., Gammon, R. H., and Warner, M. J. 1985. Atmospheric chloro uoromethanes inthe deep equatorial Atlantic. Nature, 314, 608{610.

86

Chapter 4Paper III: Simulated NorthAtlantic{Nordic Seas water massexchanges in an isopycniccoordinate OGCM

87

88

Simulated North Atlantic{Nordic Seas water mass exchanges in anisopycnic coordinate OGCMJ. E. �. Nilsen1;2, Y. Gao1, H. Drange1;2, T. Furevik2 and M. Bentsen11Nansen Environmental and Remote Sensing CenterEdv. Griegsv. 3A, N-5059 Bergen, Norway2 Geophysical Institute, University of BergenAllegt. 70, 5007 Bergen, Norway(submitted to Geophys. Res. Lett., 2002)

89

AbstractThe variability in the volume exchanges between the North Atlantic and the Nordic Seasduring the last 50 years are investigated using a synoptic-forced, global version of the MiamiIsopycnic Coordinate Ocean Model (MICOM). The simulated volume uxes roughly matchthe existing observations. The net volume ux across the Faroe-Shetland Channel (FSC) ispositively correlated with the net ux through the Denmark Strait (DS; R=0.74 for 3 yearslow pass �ltering), but negatively correlated with the net ux across Iceland-Faroe Ridge(IFR; R=-0.80). For the Atlantic In ow (AtI) across the FSC and IFR, the correlation isR=-0.59. Furthermore, the simulation suggests that an atmospheric pattern with similaritiesto the North Atlantic Oscillation is the main driving force to the inter-annual variations inthe winter-mean volume exchanges through the DS and the FSC. The model also shows a0.02 Sv yr�1 reduction of the AtI to the Nordic Seas since 1957.

90

4.1 IntroductionThe Atlantic Thermohaline Circulation (ATHC) is a dynamically active component of theclimate system, in particular on multi-annual to decadal time scales (Curry & McCartney,2001; Bentsen et al., 2002). The heat and salt carried northward across the Greenland-Iceland-Scotland (GIS) ridge are substantial, and both quantities are of importance for thewater mass and ice distribution of the Nordic Seas and Arctic Ocean, and possibly also thedeep mixing and water mass transformations taking place in the region (Furevik et al., 2002).The gateways for the exchange of water masses between the two ocean basins are the 290 kmwide and 200-620 m deep Denmark Strait (DS) between Greenland and Iceland, the 400 kmwide and 300-500 m deep Iceland-Faroe Ridge (IFR), and the 200 km wide Faroe-ShetlandChannel (FSC) with the 850 m deep Faroe-Bank Channel (FBC) at its entrance (Fig. 4.1).The northward ow of Atlantic Water (AW) and the southward ow of dense Over owWater (OW) through the three passages have recently been reviewed by Hansen and �sterhus(2000) (hereafter cited as H�). A schematic overview of the surface current system in theregion is provided in Fig. 4.1. Gulf Stream waters from the southwest spread with the NorthAtlantic Current (NAC) via diverse pathways across the whole Northeast Atlantic, from theIrminger Sea in the west to the Scottish slope in the east. Most of the AW in the IrmingerCurrent (IC) joins the southbound East Greenland Current (EGC), while only a minor partenters the Nordic Seas through the Denmark Strait.East of Iceland, surface water in the northern Iceland Basin cross the IFR. Upon meetingthe Arctic Waters north of the ridge, the two form the Iceland Faroe Front and the AW turnseastward along the ridge as the Faroe Current (FC). This ow continues along the 2000 misobath into the Norwegian Sea as the western branch of the Norwegian Atlantic Current(NWAC).Through the FSC the major source of AW is the slope current along the Scottish Slope,and this ow continues northwards along the Norwegian Continental shelf-break as the easternbranch of the NWAC. Some of the FC turns southwards along the Faroese continental slopeinto the FSC, but most of it recirculates into the slope current on the eastern side of thechannel. Furthermore, the distribution of AW in the Northeast Atlantic is thought to besusceptible to atmospheric forcing, in that the westerlies may shift the AW eastwards.The over ow across the GIS ridge from the Nordic Seas into the North Atlantic is animportant source for the North Atlantic Deep Water (Dickson & Brown, 1994), and it formsthe deepest part of the ATHC. In recent years, much attention has been put on the volumetransport of the AW (Orvik et al., 2001) and the OW (H� ; Hansen et al., 2001; Dicksonet al., 1999; Girton et al., 2001) across the GIS ridge. In this paper, the mean value and the91

36 oW

24oW 12oW 0o

12o E

57 oN

60 oN

63 oN

66 oN

69 oN

DS

EGC

FBC

FSCFAROES

GREENLAND

IC

ICELAND

ICELANDBASIN

IFRIRMINGERSEA

NORTHSEA

NO

RW

AY

NORDIC SEAS

NWAC

SCOT−

LAND

WTR

S

FS

FN

M

M

M

NORTHEAST ATLANTIC

FC

Figure 4.1: The GIS-Ridge and its surrounding waters. Isobaths are drawn for every 500 m.Schematic surface currents with key references are indicated. Abbreviations are explained inthe text. Grey lines indicate model (M), Faroe north (FN) and south (FS), and Svin�y (S)sections.multi-annual to decadal scale variability in volume transports through the DS, across the IFR,and through the FSC are investigated using a 52 years hind-cast simulation with a medium-resolution, truly global version of the Miami Isopycnic Coordinate Ocean Model (MICOM).The model output is compared with available observations and observational-based estimatesfrom the region, and a possible link between the simulated variability and the atmosphericforcing is presented.4.2 Model DescriptionThe model system applied in this study is MICOM (Bleck et al., 1992), fully coupled toa dynamic-thermodynamic sea-ice module. The model set-up and integration follow thedescription of the synoptic hind-cast simulations in Furevik et al. (2002); Bentsen et al.(2002), and only key features are provided here.The model has 25 vertical layers with �xed potential densities, and an uppermost mixedlayer with temporal and spatial varying density. In the horizontal, the model is con�guredwith a local orthogonal grid mesh with one pole over North America and one pole over westernpart of Asia (Bentsen et al., 1999), yielding a grid-spacing of 30 to 40 km in the entire NorthAtlantic-Nordic Seas region. The bathymetry is computed as the arithmetic-mean value basedon the ETOPO-5 data base (Data Announcement 88-MGG-02, Digital relief of the Surfaceof the Earth, NOAA, National Geophysical Data Center, Boulder, Colorado, 1988).The di�usive velocities (di�usivities divided by the size of the grid cell) for layer interface92

Greenland

Norw

ay

20 cm s-1

Figure 4.2: Simulated mean surface currents for the period 1948{1999.di�usion, momentum dissipation, and tracer dispersion are 0.015 m s�1, 0.01 m s�1 and0.005 m s�1, respectively. The diapycnal mixing coe�cient Kd (m2 s�1) is parameterisedaccording to the expression Kd = 3� 10�7=N , where N (s�1) is the Brunt-V�ais�al�a frequency.The model is integrated with daily NCEP/NCAR re-analysis forcing �elds (Kalnay et al.,1996) for the period 1948-1999. No relaxation is used for temperature, whereas a diagnosed,weekly-resolved, annually-repeated restoring ux is applied for the sea surface salinity.4.3 ResultsThe model manages to capture the major features of the observed ow �elds in the NorthAtlantic-Nordic Sea region (Figure 4.2). The waters from the Gulf Stream is seen to spreadover the whole Northeast Atlantic via diverse pathways. Only a small fraction of the waters inthe Irminger Current enters through the Denmark Strait, while the bulk part turns southwardand follows the EGC. The major part of the NAC enters the Nordic Seas on both sides of theFaroe Islands. The general surface ocean circulation within the Nordic Seas is dominated bythe warm and saline NWAC to the south and east, and the cold and fresh EGC to the northand west.For the model calculations, all the water masses which ow into the Nordic Seas acrossthe GIS ridge are in the following denoted \in ow", and the term \out ow" comprises bothdense over ow and surface out ows. 93

0

2

4a)

0

2

4

6

10

12b)

1950 1955 1960 1965 1970 1975 1980 1985 1990 1995

2

4

c)

Vol

ume

flux

[Sv]

Net

Inflo

wO

utflo

w

Figure 4.3: Temporal variation of the 3-years low pass �ltered simulated net (a), in ow (b),and out ow (c) volume transports through the FSC (solid lines), IFR (dashed lines), DS(dot-dashed lines), and total in ow (dotted line).In order to put emphasis on the interannual to decadal variability, time series of thesimulated transports over the GIS ridge have been low-pass �ltered using a Butterworth �lterwith cut-o� period of 3 years (Fig. 4.3). All sections reveal substantial variability, with typicalamplitudes for the transport anomalies of the order 1-2 Sv. The net ow through the FSC,for instance, was very low during the �rst 20 years (1.8 Sv), then for the next 17 years theaverage was 3.3 Sv, before decreasing to near 2 Sv for the remaining part of the simulation.The total AtI however has highest transport from the late 50's to early 70's. The transportsthrough the other two sections show similar behaviour, although the changes here are less.The mean simulated transports over the GIS-ridge are summarized and compared toestimates cited in the literature in Table 4.1. Taking into consideration the variability onmulti-annual to decadal time scales in the simulated transports (Fig. 4.3), the model-datacomparison has been performed for the relevant timespans of data collection.The net volume ux through the FSC is positively correlated with the net volume uxthrough the DS, while negatively correlated with the net volume ux accross the IFR. The cor-relation coe�cients calculated from the low-pass �ltered time series are R=0.74 and R=-0.80,respectively. The in ows over the IFR and through the FSC are also negatively correlated(R=-0.59).The role of the atmospheric forcing has been investigated by correlating the model sim-ulated volume transports with the NCEP/NCAR re-analysis data used to force the model.Point correlations with the sea-level pressure (SLP) �eld (Fig. 4.4a,c,d,f) reveal that the in- ow transport and the net transports across the DS and the FSC are highly correlated tothe large scale pressure system during winter. For the DS, the net volume transport is domi-nated by the EGC and the over ow, and is therefore directed towards south. The correlation94

Table 4.1: Simulated and observational based mean northward (N) and southward (S) volumetransports in Sv (1 Sv =106 m3 s�1) over the GIS-ridge. The observational based transportestimates without indicated measurement period are general. Values in parenthesis are thenorthward ow corrected for recirculation of the FC, and the southward ow of the deepwaters alone. 1Hansen and �sterhus (2000); 2Fissel et al. (1988); 3Hansen et al. (2002);4Turrell et al. (2002); 5Orvik et al. (2001); 6Turrell et al. (1999); 7�sterhus et al. (1999);8Ellett (1998). Section Simulated Observed1948-99 Period Value Period SourcesDS N 0.5 1.0 1S 4.3 4.3 1, 2IFR N 5.6 5.8 3.3 1995-1998 15.4 3.5 1997-2001 3S 3.6 1.0 1FSC N 4.4 4.2 4.3 (3.2) 1994-2000 44.3 4.2 1995-1998 5S 2.1 2.2 4.5 (2.6) 1994-1998 1, 6, 7, 8

95

a) Net DS b) Net IFR c) Net FSCd) In ow DS e) In ow IFR f) In ow FSC

Figure 4.4: Correlation between the simulated net transports (upper panels) and in ow (lowerpanels) through the three GIS-ridge openings, and the NCAR/NCEP (Kalnay et al., 1996)SLP in winter (December{March). Isolines are drawn at 0.25, 0.40, and 0.60 correlations withincreasing line thickness for increasing correlations. For all panels, the 95% signi�cance (bya student-t method) is 0.27. Dashed Dashed lines indicate negative correlations.resembles a NAO-like pattern (Hurrell, 1995) with centres of action located in the central toeastern Nordic Seas and in a zonal belt extending eastward from the Azores. There is also acorrelation between the net (but not the in ow) volume transport through the IFR and thelarge scale winter SLP �eld (Fig. 4.4b). The latter pattern has centres of action located overthe Greenland and Irminger Seas and the central sub-tropical Atlantic Ocean. The correla-tions for the summer months June{September are weak, and hardly signi�cant at the 95%con�dence level (not shown).4.4 Discussion and conclusionsThe simulated northward ow through the DS of 0.5 Sv is below the observational basedestimate (see Table 4.1). However, the in ow estimates are dependent on water mass de�ni-96

tions, and values for the northward DS transport range from 0.6 to 2 Sv (H�). For the netsouthward transport through the DS and the Canadian Archipelago, H� obtained a value of6.0 Sv, while Fissel et al. (1988) obtained a transport estimate of 1.7 Sv through the CanadianArchipelago alone. These numbers yield an estimate of DS out ow of 4.3 Sv, a value thatmatches the simulated transport precisely. The use of this estimate is further substantiatedby the simulated out ow through the Canadian Archipelago of 1.8 Sv. It should be mentionedthat the estimates of H� and Fissel et al. (1988) are both uncertain, as discussed by H�.Firm conclusions based on these numbers are therefore di�cult. For the sake of completeness,the simulated in ow through the Bering Strait is 1.0 Sv, the same value as Roach et al. (1995)calculated from direct measurements.Over the IFR the simulated ow both northward (5.6 Sv) and southward (3.6 Sv) exceedestimates from measurements (Table 4.1; the observational based out ow of 1.0 Sv representsthe sum of modi�ed East Icelandic Water in the surface and the OW, each carrying 0.5 Sv;H�). However, the simulated net transport (2.0 Sv) matches the observational based trans-port of 2.3 Sv quite well. This inconsistency might imply the existence of some recirculationthrough the IFR section in the model.The simulated in ow through the FSC closely matches the recent observational estimatesin the channel (Turrell et al., 2002), as well as downstream in the eastern branch of the NWAC(Orvik et al., 2001, in the Svin�y-section; Fig. 4.1). The interpretation of this depends onwhether the observed recirculation in the channel is properly simulated (See Figs. 4.1 and4.2). If the recirculation in the model is realistic, the match indicates a realistic mean AtI,and the simulated in ow should be corrected for the southward surface ow. On the otherhand, if the recirculation is too weak in the model, then the ow from the south or southwestis equally overestimated.The observational based estimate of southward ow of 4.5 Sv is the sum of the out owbelow 450 m through the FBC (2.5 Sv; �sterhus et al., 1999) and OW over the Wyville-Thompson Ridge (WTR; 0.1 Sv; Ellett, 1998), and the southward ow in the upper layersof the FSC (1.9 Sv; Turrell et al., 1999). The simulated out ow of 2.2 Sv is far to weak tomatch these observations, but given the abovementioned possibility of a weak or non-existentsouthward surface ow through the model section, proper comparison might be with deepout ow alone (2.6 Sv). Finally, the simulated net transport of 4.3 Sv across the Iceland-Scotland ridge is close to the recent estimate of 4.0 Sv by H�.To our knowledge, only indirect observations are available to check the time evolutionof the simulated uxes. Based on hydrographic observations from the Svin�y section in theperiod 1955 to 1996, Mork and Blindheim (2000) calculated the geostrophic volume transportsin the two branches of the NWAC, and found that they appeared to be in opposite phase97

and NAO-controlled during summer (since 1978). This out-of-phase relationship is also foundin the model, where the transports through the IFR and the FSC are negatively correlated(R=-0.59).The relation between the atmospheric pressure systems and the model uxes are supportedby Blindheim et al. (2000) who found that the width of the NWAC is negatively correlatedwith the NAO-index on long (> 3 years) time scales. They attribute this to changes in thepathways of the ow over the ridge rather than the local wind forcing, since a high NAO-index may cause the westerlies to shift the NAC eastward. This is supported by Belkin andLevitus (1996), who report large meridional displacements of the NAC near the Charlie GibbsFracture Zone.Furthermore, based on nearly four years direct measurement (April 1995 to February1999) along the Svin�y section, Orvik et al. (2001) revealed a strong connection between theNAO-index and the transport in the eastern branch of the NWAC even on interannual timescales (i.e., high FSC-in ow coincides with high NAO-index).Not only the westerlies in the North Atlantic change with NAO-like forcing. The norther-lies modulating the southward ow of the Arctic Waters in the western parts of the NordicSeas are also controlled by these large scale changes in the pressure system (Blindheim et al.,2000). In this way, the winds together with increased southward extension of Arctic Watersmay inhibit in ow at both the Denmark Strait and over the Iceland Faroe ridge (Blindheimet al., 2000).In concordance with the above theories, the correlation patterns for the DS and the FSCin the low western/high eastern in ow situation (Fig. 4.4a,c,d,f) all clearly indicate pressuregradients giving both northerly winds along the East Coast of Greenland and westerlies inthe north Atlantic.Furthermore, it is interesting to note that Hansen et al. (2001) found indirect evidence ofa reduction in over ow through the FBC of 0.5 Sv over the last 50 years, implying that theAtI has been reduced to a similar degree. However, without simultaneous direct observationsacross the entire GIS ridge, it remains unclear whether the total over ow has been reduced.In the model, the AtI (Fig. 4.3b) has a negative trend giving a 0.7 Sv reduction in in ow tothe Nordic Seas since 1957 (also when considering the net in ow through the other openingsto the Arctic Mediterranean). Given the assumption that 75% of the AtI is returned to theAtlantic through the over ow (Hansen et al., 2001), the two studies agree. This issue, andthe more fundamental roles of the ATHC and wind driven transport, certainly need to beaddressed in order to improve the skill of future climate scenarios.98

4.5 AcknowledgmentsThe model development and analyses have been supported by the Research Council of Norwaythrough several projects, and in particular RegClim and KlimaProg's \Spissforskningsmidler",and the Programme of Supercomputing. The work has also received support from the EU-project PREDICATE (EVK2-CT-1999-00020). Support from the G. C. Rieber Foundationsis also highly acknowledged.

99

ReferencesBelkin, I., & Levitus, S. (1996). Temporal variability of the Subarctic Front near the Charlie-Gibbs FractureZone. J. Geophys. Res., 101, 28317-28324.Bentsen, M., Drange, H., Furevik, T., & Zhou, T. (2002). Variability of the Atlantic Meridional OverturningCirculation in an isopycnic coordinate OGCM. Climate Dynamics. (submitted)Bentsen, M., Evensen, G., Drange, H., & Jenkins, A. D. (1999). Coordinate Transformation on a Sphere UsingConformal Mapping. Mon. Weather Rev., 127, 2733{2740.Bleck, R., Rooth, C., Hu, D., & Smith, L. T. (1992). Salinity-driven thermohaline transients in a wind-and thermohaline-forced isopycnic coordinate model of the North Atlantic. J. Phys. Oceanogr., 22,1486-1515.Blindheim, J., Borovkov, V., Hansen, B., Malmberg, S. A., Turrell, W. R., & �sterhus, S. (2000). Upper LayerCooling and Freshening in the Norwegian Seas in Relation to Atmospheric Forcing. Deep-Sea Res., 47,655-680.Curry, R., & McCartney, M. S. (2001). Ocean gyre circulation changes associated with the North AtlanticOscillation. J. Phys. Oceanogr., 31, 3374{3400.Dickson, B., Meincke, J., Vassie, I., Jungclaus, J., & �sterhus, S. (1999). Possible predictability in over owfrom the Denmark Strait. Nature, 397, 243-246.Dickson, R., & Brown, J. (1994). The production of North Atlantic Deep Water: Sources, rates, and pathways.J. Geophys. Res., 99 (C6), 12319-12341.Ellett, D. (1998). Norwegian Sea Deep Water over ow across the Wyville Thompson Ridge during 1987{1988.ICES Coop. Res. Rep.(225), 195-205.Fissel, D., Birch, J., Melling, H., & Lake, R. (1988). Non-tidal ows in the Northwest Passage (Tech. Rep.No. 98). Canadian Institute of Ocean Science, Sidney, British Columbia, Canada.Furevik, T., Bentsen, M., Drange, H., Johannessen, J. A., & Korablev, A. (2002). Temporal and spatialvariability of the sea surface salinity in the Nordic Seas. J. Geophys. Res. (in press)Girton, J. B., Sanford, T. B., & K�ase, R. H. (2001). Synoptic sections of the Denmark Strait Over ow.Geophys. Res. Lett., 28 (8), 1619-1622.Hansen, B., & �sterhus, S. (2000). North Atlantic{Nordic Seas Exchanges. Prog. Oceanog., 45, 109-208.Hansen, B., �sterhus, S., Hatun, H., Kristiansen, R., & , K. M. H. L. . (2002). The Iceland-Faroe in ow ofAtlantic water to the Nordic Seas. (in press)Hansen, B., Turrell, W. R., & �sterhus, S. (2001). Decreasing over ow from the Nordic Seas into the AtlanticOcean through the Faroe Bank Channel since 1950. Nature, 411, 927-930.100

Hurrell, J. (1995). Decadal trends in the North Atlantic Oscillation: Regional temperatures and precipitation.Science, 269, 676-679.Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., Deaven, D., Gandin, L., Iredell, M., Saha, S., White, G.,Woollen, J., Zhu, Y., Y, Chelliah, M., Ebisuzaki, W., Higgins, W., Janowiak, J., Mo, K. C., Ropelewski,C., Wang, J., Leetmaa, A., Reynolds, R., Jenne, R., & Joseph, D. (1996). The NCEP/NCAR 40-yearReanalysis Project. Bull. Am. Meteor. Soc., 77 (3), 437-471.Mork, K. A., & Blindheim, J. (2000). Variation in the Atlantic In ow to the Nordic Seas, 1955{1996. Deep-SeaRes. I, 47 (6), 1035-1057.Orvik, K., Skagseth, �., & Mork, M. (2001). Atlantic In ow to the Nordic Seas. current structure and volume uxes from moored current meters, VM{ADCP and SeaSoar-CTD observations, 1995{1999. Deep-SeaRes. I, 48, 937-957.�sterhus, S., Hansen, B., Kristiansen, R., & Lundberg, P. (1999). The over ow through the Faroe BankChannel. Int. WOCE Newsl.(35), 35-37.Roach, A., Aagaard, K., Pease, C., Salo, S., Weingartner, T. ., Pavlov, V., & Kulakov, M. (1995). Directmeasurements of transport and water properties through the Bering Strait. J. Geophys. Res., 100,18443-18457.Turrell, W., Hansen, B., Hughes, S., & �sterhus, S. (2002). Hydrographic variability during the decade ofthe 1990's in the northeast Atlantic and southern Norwegian Sea. In Ices symposium on hydrobiologicalvariability in the ices area, 1990-99. ICES.Turrell, W., Hansen, B., �sterhus, S., Hughes, S., Ewart, K., & Hamilton, J. (1999). Direct observations ofin ow to the Nordic Seas through the Faroe Shetland Channel 1994{1997. ICES CM 1999/L(1), 1-15.

101

102

Chapter 5Paper IV:Sensitivity of DiapycnalMixing on the Oceanic Ventilationand Uptake of CFC-11 in theSouthern Ocean

103

104

Sensitivity of Diapycnal Mixing on Oceanic Ventilation and Uptake ofCFC-11 in the Southern OceanYongqi Gao� and Helge Drange1Nansen Environmental and Remote Sensing CenterEdv. Griegsv. 3A, N-5059 Bergen, NorwayAlso at Geophysical Institute, University of Bergen, Nowrway(submitted to Journal of Marine Systems, 2002)

�Corresponding author:Yongqi Gao (yong@nersc.no)Nansen Environmental and Remote Sensing CenterEdv. Griegsv. 3AN{5059 BergenNORWAY105

AbstractThe sensitivity of diapycnal mixing on the oceanic uptake of anthropogenic CFC-11 andthe ventilation of the surface waters in the Southern Ocean (SO) are investigated using theMiami Isopycnic Coordinate Ocean Model (MICOM). Three model experiments have beenperformed: One with a diapycnal mixing Kd (m2 s�1) of 2 � 10�7=N (Exp. 1), one withvanishing diapycnal mixing (Exp. 2), and one with 5�10�8=N (Exp. 3; N (s�1) is the Brunt-V�ais�al�a frequency). The performed model simulations indicate that the observed verticaldistribution of CFC-11 along 272�E in the SO is caused by local ventilation of the surfacewaters and westward-directed isopycnic transport and mixing from deeply ventilated waters inthe Weddell Sea region. It is found that at the end of 1997, the simulated net ocean uptake ofCFC-11 in Exp. 2 is 25% below that of Exp. 1. Of this reduction, the decreased uptake of CFC-11 in the SO accounts for 80% of the total di�erence. Furthermore, Exps. 2 and 3 yield farmore realistic vertical distributions of the ventilated CFC-waters than Exp. 1. The performedtriple-experiment also highlights the sensitivity of the SO surface water ventilation to thedistribution of the simulated ML. Rather small di�erences in the maximum ML depth mayeasily generate huge di�erences in the temporal-spatial properties of the ventilated waters. Itis argued that inclusion of CFCs in coupled climate models could be used as a test-bed forevaluating the decadal-scale ocean uptake of heat and CO2 in these models.Keywords: Chloro uorocarbons, Diapycnal mixing, Ocean modelling, Ocean ventilation,Southern Ocean, Transient tracers

106

5.1 IntroductionOcean General Circulation Models (OGCMs) are key tools in the study of the future oceanuptake of atmospheric greenhouse gases and heat. Furthermore, whereas nature experiencesone realisation of the climate state, climate models can be used as a laboratory to produce amultitude of climate realisations, and by that contribute to the understanding of the variabilityand stability properties of the system. It is in this respect crucial to evaluate the climatemodels against observed quantities to assess the degree of realism of the models.Numerical simulations performed with OGCMs indicate that the SO is the largest sink ofhuman-induced CO2 in the World Oceans (Orr et al., 2001), and that the simulated tracerdistributions vary signi�cantly between di�erent models (Dutay et al., 2002). Recent studiesshow that the vertical or, for an isopycnic coordinate OGCM, the diapycnal mixing plays akey role in the ventilation of the SO (Robitaille & Weaver, 1995; Sloyan & Rintoul, 2000). Itis therefore interesting to test the dependence of diapycnal mixing on the ventilation of thesurface waters, and consequently on the ocean uptake and storage of greenhouse gases andheat.Chloro uorocarbon CFC-11 (CCl3F) and CFC-12 (CCl2F2) were introduced into atmo-sphere in the early 1930s, and their atmospheric evolution (Fig. 5.1) is fairly well known.CFCs are man-made and are chemically and biologically inactive in the ocean, and are there-fore well suited to evaluate ocean ventilation processes on multi-annual to decadal time scales.The recent World Ocean Circulation Experiment (WOCE) has conducted global surveyes ofthe distributions of the CFCs (see http://whpo.ucsd.edu/distmaps.htm). Furthermore, nu-merical studies have demonstrated that validation against observed distributions of CFCs isa powerful method to explore mixing and transport properties in OGCMs (Dixon et al., 1996;Dutay et al., 2002), and to infer sources and mixing pathways of water masses in general(Orsi et al., 1999; Smethie Jr et al., 2000).The objective of the study is to explore the ventilation processes in the SO, hereafterde�ned as the region between 45�S and 69�S, by an isopycnic coordinate OGCM. For this, theobserved CFC-11 distribution along WOCE section P19 at 272�E during 1993 (see the P19cand P19s cruises on http://whpo.ucsd.edu/data/onetime/pac�c/p19/) have been comparedwith the simulated CFC-11 distributions along the same section.Model-data evaluations are, in general, based on classical level (or z-coordinate) OGCMs.The newly developed isopycnic coordinate OGCMs (e.g., Bleck et al., 1992; Oberhuber, 1993)have the advantage that mixing across the coordinate interfaces can be set to zero as thereis no (false) numerical mixing across these surfaces. In this con�guration, the coordinateinterfaces are material surfaces, and thereby truly isolate the e�ect of transport and mixing107

1940 1950 1960 1970 1980 1990Year

0

100

200

300

400

500

600

CF

Cs

Par

tialP

ress

ure

(pic

oatm

)

CFC-12

CFC-11

Figure 5.1: Reconstructed history of the partial pressure of CFC-11 and CFC-12 in dry airat one atmosphere pressure (Walker et al., 2000). The solid (dashed) line corresponds to theCFC concentrations in the northern (southern) hemisphere.along the coordinate layers. The latter property is intriguing for the ocean environment asthe preferred plane of ow and mixing is along, rather than across, planes of constant density(Ledwell & Watson, 1993; Toole et al., 1994).The paper is organized as follows: Section 5.2 provides a brief description of the physicalmodel and the setup of the CFCs simulations. Section 5.3 shows the results of the experiments,and the paper is concluded in section 5.4.5.2 Model descriptionThe OGCM used in this study is the Miami Isopycnic Coordinate Ocean Model (MICOM;Bleck et al., 1992). MICOM is a primitive equation model utilizing surfaces of constantdensity as the vertical coordinate. In the presented study, potential density at 2000 dbar (�2)is used as the reference level for the vertical coordinate surfaces.The modeled ocean consists of a surface mixed layer (ML) in which the potential densityvaries in time and space, and 15 interior isopycnic layers. For the ML, the Gaspar (1988)bulk-parameterization is used. All the air-sea exchanges of momentum, heat and fresh waterare incorporated into the ML using conventional bulk formulae. No surface restoration ofSST and SSS is applied. Convective mixing takes place if the density of the ML exceedsthe density of one or more of the underlying isopycnals. The instability is then removed bymixing all of the unstable water masses, and by absorbing the new water mass into the ML.Both momentum and tracers are uniformly mixed in the case of convective mixing.The interior isopycnals exchange their properties with the ML if they outcrop to the ML.Therefore, the location and timing of the outcropping of the isopycnic layers are of centralimportance in analysing the ocean ventilation of any of the ML properties. In addition, mixing108

is prescribed in the direction normal to the isopycnic interfaces. The associated diapycnalmixing coe�cient Kd (m2 s�1) is proportional to N�1 (Gargett, 1984), where N = qg� @�@z(s�1) is the Brunt-V�ais�al�a frequency, g (m s�2) is the gravity accelaration, � (kg m�3) is thedensity and z (m) is the depth. For the base-line integration (Exp. 1 in the following), theproportionality coe�cient is set to 2 � 10�7 m2 s�2. The numerical implementation of thediapycnal mixing follows the scheme of McDougall and Dewar (1998). Readers interested inthe intrinsic model features are referred to Bleck et al. (1992).5.2.1 Model con�gurationThe set-up of the applied version MICOM follows Sun (1997). The model domain spansthe region from 65�N to 69�S. A regular grid on a Mercator projection is used and a rathercoarse latitude-by-longitude resolution of 2� � 2� cos� (where � is latitude) is chosen for theexperiments. The �2 values for the isopycnic layers are 33.22, 34.26, 35.04, 35.62, 36.05,36.37, 36.61, 36.79, 36.92, 37.01, 37.07, 37.11, 37.14, 37.17 and 37.20, and the the equationsare di�erentiated on an Arakawa and Lamb (1977) C-grid stencil.The di�usive velocities (di�usivities divided by the size of the grid cell) for layer interfacedi�usion, momentum dissipation and temperature/salinity mixing are 0.005 m s�1, 0.01 ms�1 and 0.005 m s�1 respectively, yielding actual di�usivities of the order of 103 m2 s�1 for a100 km grid spacing. The presence of sea ice has been mimicked by setting the surface windstirring and the heat uxes to zero wherever the modeled mixed layer temperature is below�1.8 �C. No ice-related physical processes, for instance brine rejection (Anderson & Jones,1991), are incorporated in the model.Atmospheric surface forcings applied to the model include monthly mean climatologicalsurface wind stress, atmospheric relative humidity and surface temperature from the Com-prehensive Oceanographic and Atmospheric Data Set (COADS; Woodru� et al., 1987). Themonthly mean climatological net radiation ux is from the Oberhuber (1988) Atlas, and theprecipitation is from the NOAA microwave sunder (Spencer, 1993).Following the atmospheric record (Fig. 5.1), the CFCs simulation begins in 1931 with zeroconcentrations in the atmosphere and the ocean. The uxes of CFCs at the air-sea surfaceare expressed as: F = Kw � (Csat �Csurf ) (5.1)where F (mol m�2 s�1) is the ux of the CFCs, Csat (mol m�3) is the saturated CFCsconcentration in moist air near the sea surface, Kw (m s�1) is the transfer (or piston) velocity,and Csurf (mol m�3) is the modeled surface ocean CFCs concentration. The transfer velocityKw is computed using the Eq. 3 in Wanninkhof (1992). For details about introducing CFCs109

in OGCMs, see the description in Dutay et al. (2002).5.2.2 Model experimentsThe OGCM was �rst spun-up for 164 years with Kd = 2� 10�7=N . Three CFCs simulationswere then performed for the period 1931 to 1997: One experiment representing a continuationof the spin-up integration (Kd = 2 � 10�7=N ; Exp. 1), a second experiment with vanishingKd (Exp. 2), and a third experiment with Kd = 5� 10�8=N (Exp. 3).The value of Kd in Exp. 1 is identical to a recently estimated value of Kd in the pycnoclineover the South Scotia Ridge at the north-western entrance of the Weddell Sea (Muench et al.,2002). Since, for instance, tidal dissipation is expected to be particularly strong over roughtopography (e.g., Ledwell et al., 2000; Muench et al., 2002), it is expected that Exp. 1represents an upper (but realistic) bound on Kd.The main motivation for the performed triple experiment is, as already described, to fullyexploit the isopycnal versus diapycnal in uence on the ventilation of the SO waters. It shouldbe stressed that the experiments should be viewed as sensitivity experiments as the spin-upintegration is similar for all of the experiments. The use of identical initial conditions forthe physical model imply that the isolated e�ect of isopycnal versus diapycnal transport andmixing is more transparent than for the case with three independent model spin-ups. A morerealistic model con�guration would include fully independent model integrations (i.e., threeseparate model spin- ups), the use of realistic (i.e., synoptic) atmospheric forcing �elds, anda truly global model domain including a full dynamic-thermodynamic sea ice module.5.3 Results5.3.1 Ocean storage and uptake of CFC-11The CFC-11 absorbed by the model increases with the atmospheric concentration of CFC-11(Fig. 5.2). At the end of 1997, the CFC-11 inventory is 820�106 mol in Exp. 1 and 614�106mol in Exp. 2, or 75% of that of Exp. 1. These �gures clearly indicate the model dependencyon Kd.The zonally integrated inventory and the accumulated ux of CFC-11 (Fig. 5.3) illustratethat by far the most e�cient uptake and storage of CFC-11 take place in the SO. Theshift between the accumulated CFC- uxes at about 60�S and the maximum inventories atabout 45�S are consistent with an equatorward transport of CFC-11 from the high southernlatitudes.The accumulated CFC-11 ux in the SO can be deduced from Fig. 5.3. The obtained110

values are about 475�106 mol for Exp. 1 and 311�106 mol for Exp. 2, accounting for asmuch as 58% and 50% of the simulated World Ocean uptake, respectively. Furthermore, thedi�erence in the SO's uptake between the two experiments, 164�106 mol, accounts for 80%of the total di�erence of 206�106 mol between the two experiments, despite the area of theSO is only about 30% of the model domain. Finally, the averaged accumulated CFC-11 ux

1940 1950 1960 1970 1980 1990Year

0

200

400

600

800

1000

CF

C-1

1up

take

(106

mol

)

1940 1950 1960 1970 1980 19900

200

400

600

800

1000

Figure 5.2: Time evolution of the simulated oceanic uptake of CFC-11 (106 mol) in Exps. 1(solid line) and 2 (dashed line).

-60 -50 -40 -30 -20 -10 00

5

10

15

20

25

30

35

40

45

-60 -50 -40 -30 -20 -10 00

5

10

15

20

25

30

35

40

45

-60 -50 -40 -30 -20 -10 00

5

10

15

20

25

30

35

40

45

-60 -50 -40 -30 -20 -10 0Latitude (oS)

0

5

10

15

20

25

30

35

40

45

CF

C-1

1in

v.an

dac

c.flu

x(1

06m

ol)

Figure 5.3: Zonally integrated inventories of CFC-11 (106 mol) (solid lines) and accumulated ux of CFC-11 (106 mol) (dashed lines) at the end of 1997. Exp. 1 in thick lines and Exp. 2in thin lines.in the SO are 8.63�10�6 mol m�2 and 5.65�10�6 mol m�2 in Exps. 1 and 2, respectively,yielding a di�erence of 35%. 111

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

0

0

Latititude (oS)

0 0

1 1

2 2

3 3

4 4

5 5

6 6

CF

C-1

1flu

x(1

0-5pm

olm

-2s-1

)

Figure 5.4: Zonally averaged CFC-11 ux F (solid lines) and virtual ux Fv (dashed lines)(10�5 pmol m�2 s�1) in September 1990. Exp. 1 in thick lines and Exp. 2 in thin lines.5.3.2 The meridional distribution of CFC-11To qualitatively illustrate the reason for the di�erence in the meridional distribution of theCFC-11 uxes, Eq. 5.1 can be expressed as follows:F = Kw � Csat � �1� CsurfCsat � = Kw � Csat � (1� Psurf ) ; (5.2)where Psurf is the saturation degree of the CFCs. The above expression yields the ux F gridpoint by grid point. To simplify the analyses, the virtual ux Fv (mol m�2 s�1) is de�ned as:Fv = Kw � Csat � 1� �CsurfCsat �! = Kw � Csat � �1� Psurf� ; (5.3)where the overbar denote zonal averages.In the performed experiments, the strongest uptake of CFC-11 in the SO takes place inSeptember, when the ML temperature is low and subsequently the ML is deep (see below).The zonally averaged CFC-11 ux obtained by zonally averaging F in Eq. 2 and the virtual ux Fv from Eq. 3 in September 1990 are depicted in Fig. 5.4 for Exps. 1 and 2. In bothexperiments, the meridional variation in Fv follows, in general, that of F . Therefore, thefactors governing the meridional variation in the virtual ux Fv , rather than the grid point ux F , are addressed next.The simulated zonal mean CFC-11 saturation degree, the CFC-11 transfer velocity, thesaturated CFC-11 concentration, and the ML potential temperature, salinity and thickness inSeptember 1990 are provided in Fig. 5.5 for Exps. 1 and 2. The surface CFC-11 is signi�cantlyundersaturated in the SO implying that the SO acts as a sink of CFC-11. The saturation112

degree is about 95% from 10�S to 40�S and decreases south of 45�S. At the southern boundary,the saturation degree is about 52% and 80% in Exps. 1 and 2, respectively.The transfer velocity Kw of CFC-11, representing the e�ectiveness of ocean uptake ofCFC-11 given an air-sea disequilibrium of the gas, reaches the maximum value of 7.0�10�5m s�1 (or about 6.0 m day�1) at about 56�S and the minimum value of 1.2�10�5 m s�1 (or1.0 m day�1) at the southern boundary. The combined e�ect of the high transfer velocityand the low saturation degree result in an e�cient uptake of CFC-11 in the SO.In addition, the increase in the surface CFC-11 concentration towards the high southernlatitudes shows the temperature dependence on the CFC-11 solubility. In the SO, the potentialtemperature in Exp. 2 is slightly lower than that in Exp. 1. As a result, the saturated CFC-11 concentration is slightly higher in Exp. 2. The salinity, however, is signi�cantly higherin Exp. 1. Therefore, the ML goes deeper in Exp. 1 than in Exp. 2. This is of importancefor the ocean uptake of the CFCs as a deep ML implies a volumetric dilution of the MLCFC-11 concentration and consequently a prolonged air-sea equilibration time, and by thatan enhanced ocean uptake of the CFCs.5.3.3 Ventilation of the SOThe penetration of the CFCs from the CFC-enriched surface waters into the intermediateto deep waters are governed by mixing and transport. In the model, the spreading of theCFC-signal onto the isopycnals is dictated by the isopycnal transport and mixing. Therefore,the simulated ML thickness and the locations of the outcroping isopycnals are key quantitiesin assessing the ventilation of the di�erent water masses. In Exp. 2, the ML water is theonly source for the ventilation of the sub-surface water masses due to the vanishing diapycnalmixing.The simulated maximum ML thickness distribution in the SO are displayed in Figure 5.6in Exps. 1 and 2. In Exp. 1, surface mixing reaches a depth of 2400 m in the Weddell Sea(315�E), 1200 m along the southern boundary in the Paci�c Ocean sector, and more than1600 m between 62�S and 66�S in the Indian Ocean sector (10�E{60�E). As a result, thepenetration depth of the CFC-11 signal will be large in the areas where the simulated deepmixing occurs. In Exp. 2, the simulated deep convection in the Weddell Sea is similar withthat in Exp. 1. However, convective mixing reaches only 400 m along the southern boundaryin the Paci�c Ocean sector and 800-1600 m in the Indian Ocean sector. As a result, CFC-11does not, in general, penetrate as deep in Exp. 2 as in Exp. 1.The outcroping of the isopycnals to the ML determine the ventilation pathway of watermasses in the model. To illustrate this, the density of the outcroping isopycnals in September,the month of maximum ML thickness, is displayed in Fig. 5.7. In Exps. 1 and 2, the 37.17-113

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

0

0

Latititude (oS)

40 40

60 60

80 80

100 100

CF

C-1

1S

atu

ratio

n(%

)

(a) -60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

0

0

Latititude (oS)

0 0

0.2 0.2

0.4 0.4

0.6 0.6

0.8 0.8

1 1

Tra

nsfe

rve

loci

ty(1

0-4m

s-1)

(b)-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

0

0

Latititude (oS)

1 1

2 2

3 3

4 4

5 5

6 6

7 7

8 8

9 9

10 10

CF

C-1

1co

nc.

(pm

olkg

-1)

(c) -60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

0

0

Latititude (oS)

0 0

5 5

10 10

15 15

20 20

25 25

30 30

ML

pot.

tem

per

atur

e(o

C)

(d)-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

0

0

Latititude (oS)

33 33

34 34

35 35

36 36

ML

salin

ity(p

su)

(e) -60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

0

0

Latititude (oS)

0 0

200 200

400 400

600 600

800 800

1000 1000

1200 1200

ML

thic

knes

s(m

)

(f)Figure 5.5: Simulated zonal mean CFC-11 saturation degree (a), CFC-11 transfer velocity(10�2 m s�1; b), saturated CFC-11 concentration (pmol kg�1; c), ML potential temperature(�C; d), ML salinity (psu; e) and ML thickness (m; f) in September, 1990. The solid (dashed)lines represent Exp. 1 (Exp. 2).114

30

30

60

60

90

90

120

120

150

150

180

180

210

210

240

240

270

270

300

300

330

330

Longitude(oE)

-70 -70

-60 -60

-50 -50

Lat

itude

(o S)

1200

1200

200

200

400

800

2000

1200

200

400400

400

8001200 2000 2400

800

16002000

400

800

(a) Exp. 130

30

60

60

90

90

120

120

150

150

180

180

210

210

240

240

270

270

300

300

330

330

Longitude(oE)

-70 -70

-60 -60

-50 -50

Lat

itude

(o S)

200

200

400

400

800

400200

800

800

200

2400

200

200

4001200

800

400

(b) Exp. 2Figure 5.6: Simulated maximum ML thickness (m) in the SO in Exps. 1 (upper panel) and 2(lower panel) in year 1990. Shaded areas denote ML > 800 m, and the WOCE P19 sectionis marked at 272�E.

115

30

30

60

60

90

90

120

120

150

150

180

180

210

210

240

240

270

270

300

300

330

330

Longitude(oE)

-70 -70

-60 -60

-50 -50

Lat

itude

(o S)

36.79

36.92

37.07

37.11

36.37

36.92

36.7936

.79

37.01

37.11

37.17

37.1737.1137.14

36.61

(a) Exp. 130

30

60

60

90

90

120

120

150

150

180

180

210

210

240

240

270

270

300

300

330

330

Longitude(oE)

-70 -70

-60 -60

-50 -50

Lat

itude

(o S)

36.61

36

36.79

36.37

37.01

37.0136.61

37.0737.1737.14

36.92

36.37

36.79

37.01

37.07

36.3736.61

36.3

7

36.0

5

36.61

36.79

(b) Exp. 2Figure 5.7: Locations of the outcroping isopycnals when the ML is at maximum state in 1990in Exps. 1 (upper panel) and 2. The shaded areas denote where the �2 � 37:11 isopycnalsoutcrop to surface. The WOCE P19 section is shown at 272�E.

116

isopycnal outcrops to the surface in the Weddell Sea, and will consequently be ventilated bythe properties of the ML waters in this region. An additional source for the ventilation of thenear bottom �2 � 37:17 isopycnals in Exp. 1 is located west of the Antarctic Peninsula wherethe 37.17-isopycnal outcrops to the surface. Therefore, the source for the ventilation of the37.17-isopycnal will be the Weddell Sea and west of the Antarctic Peninsula in Exp. 1, andthe Weddell Sea only in Exp. 2.In the Paci�c sector of the SO in Exp. 1, the 37.14-isopycnal outcrops to the surface in theRoss Sea (at about 180�E) whereas the 37.11-isopycnal outcrops to the surface along most ofthe southern boundary. For comparison, no dense water masses with �2 > 36:92 outcrops tothe surface between 90�E and 280�E in Exp. 2. Therefore, the presence of CFC-11 in waterswith �2 > 36:92 in Exp. 2 can only be attributed to isopycnal transport and dispersive mixingfrom the source region east of 280�E and west of 90�E. West of the Antarctic Peninsula, the37.14-isopycnal outcrops to the surface in Exp. 2. In the Indian sector, the 37.11-isopycnaloutcrops to the surface between 64�S and 66�S in Exp. 1, whereas it is the 37.07-isopycnalthat outcrops to the surface in the similar area in Exp. 2.The in uence of the ventilation rate in the SO in Exps. 1 and 2 is illustrated in Fig. 5.8.It follows from this �gure that the CFC-11 inventories of the water masses with �2 > 36:92are higher in Exp. 1 than those in Exp. 2. The high CFC-11 inventory in the deepest watermasses in Exp. 1 con�rms that strong ventilation takes place on these isopycnals.To illustrate the meridional di�erence in the distribution of CFC-11, the observed andthe simulated CFC-11 concentrations along the WOCE P19 section at 272�E are shown inFig. 5.9. At the surface, largest di�erences are seen north of 10�S with too strong mixingin Exp. 1, too weak mixing in Exp. 2, and a fairly realistic mixing in Exp. 3. However,substantial di�erences are clearly seen in the distribution of CFC-11 in the bulk part of theintermediate and deep waters of the SO.In Exp. 1, the simulated invasion of CFC-11 in the SO is far deeper and more widespreadthan in the observations. In fact, there is almost no CFC-11 signal between 2400 m and 2600m in the observations. This in contrast to Exp. 1 where CFC-11 penetrate to the bottom atall latitudes south of 45�S. The vertical distribution is, however, greatly improved in Exps. 2and 3.To quantitatively evaluate the ocean storage of CFC-11, the simulated and observed in-ventory along the P19 section are shown in Fig. 5.10. All the simulations overestimate theCFC-11 inventory south of 60�S in the SO by a factor 2 or more, and then in particularExp. 1. Furthermore, the CFC-11 inventory decreases with reduced strength of the diapycnalmixing, as expected.To investigate the reason for the di�erences in the distribution of the observed and simu-117

0 10 20 30 40 50 60Inventory of CFC-11 (106 mol)

Mxd

33.22

34.26

35.04

35.62

36.05

36.37

36.61

36.79

36.92

37.01

37.07

37.11

37.14

37.17

37.20

σ 2

(a)

0 10 20 30 40 50 60Inventory of CFC-11 (106 mol)

Mxd

33.22

34.26

35.04

35.62

36.05

36.37

36.61

36.79

36.92

37.01

37.07

37.11

37.14

37.17

37.20

σ 2

(b)Figure 5.8: CFC-11 inventory (106 mol) on each isopycnal in year 1997 in the SO the in (a)Exp. 1 and (b) Exp. 2.lated CFC-11 concentrations along the WOCE P19 section, the climatological and the simu-lated density strati�cations are shown in Fig. 5.11. In Exp. 1, the simulated deep convectionreaches 1200 m south of 66�S, and here the 37.11-isopycnal outcrops to the ML. Therefore, bywater mass transfer, the 37.11-isopycnal will be ventilated by the ML water and the surfacewater CFC-11 signal will be entrained onto the layer. It also follows from the �gure thatthe 37.17-isopycnal can not be ventilated by the ML water locally. Therefore, the presenceof CFC-11 on the 37.17-isopycnal along the WOCE P19 section (cfr. Fig. 5.9b) is caused byisopycnal transport and dispersive mixing from other locations in the SO.In Exp. 2, the simulated ML thickness is about 500 m at the southern boundary, and herethe 37.01-isopycnal outcrops to the surface. Therefore, the 37.01 isopycnal can be ventilatedby the ML water at 272�E. Clearly, the 37.11, 37.14 and 37.17 isopycnals can not be ventilatedby ML water locally. The presence of CFC-11 on these layers (cfr. Fig. 5.9c) is therefore aresult of isopycnal transport and mixing. The situation for Exp. 3 falls between Exps. 1 and 2,but is closest to Exp. 2.It should be noted that the simulated deep convection south of 60�S is stronger in all ofthe experiments compared to that inferred from the climatological density strati�cation. Theoverestimated mixing depth forces the deepest isopycnals to outcrop to the surface and willsurely result in the overestimated CFC-11 uptake by the model, in accordance with Fig. 5.10.To identify the source regions for the ventilation of the 37.17-isopycnal in Exps. 1 and 2,118

3.21.63.2

1.6

0.10.03

0.015

0.1

0.4

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

0

0

Latitude (oS)

0 0

1000 1000

2000 2000

3000 3000

4000 4000

5000 5000

Dep

th(m

)0.015 0.03 0.05 0.1 0.2 0.4 0.8 1.6 3.2

(a) Observed CFC-113.2

1.60.2

0.40.050.1

0.20.8

0.015

1.6

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

0

0

Latitude (oS)

0 0

1000 1000

2000 2000

3000 3000

4000 4000

5000 5000

Dep

th(m

)

0.015 0.03 0.05 0.1 0.2 0.4 0.8 1.6 3.2

(b) CFC-11 in Exp. 11.6

0.053.21.6 0.8

0.050.015

0.2

0.0150.2

0.0150.030.05

0.2

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

0

0

Latitude (oS)

0 0

1000 1000

2000 2000

3000 3000

4000 4000

5000 5000

Dep

th(m

)

0.015 0.03 0.05 0.1 0.2 0.4 0.8 1.6 3.2

(c) CFC-11 in Exp. 21.6

0.053.2

1.6

0.10.4

0.03

0.05

0.1

0.2

0.8

0.0150.8

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

0

0

Latitude (oS)

0 0

1000 1000

2000 2000

3000 3000

4000 4000

5000 5000

Dep

th(m

)

0.015 0.03 0.05 0.1 0.2 0.4 0.8 1.6 3.2

(d) CFC-11 in Exp. 3Figure 5.9: Observed (a) and simulated (b, c, d) CFC-11 (pmol kg �1) along WOCE P19section at 272�E in the Southern Paci�c in early 1993. The cut-o� value of the CFC-11concentration is 0.01 pmol kg�1 in all panels. Observed CFC-11 from P19c and P19s onhttp://whpo.ucsd.edu/data/onetime/pac�c/p19/.-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

0

0

Latitude (oS)

0 0

2 2

4 4

6 6

8 8

CF

C-1

1In

vent

ory

(um

olm

-2)

Figure 5.10: Observed and simulated CFC-11 inventories (�mol m�2) along WOCE P19.The thick solid line, the dashed line, the thin solid line and the dash-dotted line represent theinventories of CFC-11 in the observation, and in Exps. 1, 2 and 3, respectively.119

37.01

37.0737.11

37.1437.17

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

0

0

Latitude (oS)

0 0

1000 1000

2000 2000

3000 3000

4000 4000

5000 5000

Dep

th(m

) 36.7936.6136.37

36.0535.62

36.92

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

0

0

Latitude (oS)

0 0

1000 1000

2000 2000

3000 3000

4000 4000

5000 5000

Dep

th(m

)

(a) Climatological density strati�cation ac-cording to Levitus and Boyer (1994) and Lev-itus et al. (1994)37.07

37.1137.14

37.01

36.92

36.79

36.61

36.37

37.17

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

0

0

Latitude (oS)

0 0

1000 1000

2000 2000

3000 3000

4000 4000

5000 5000

Dep

th(m

)(b) Exp. 1

36.79

36.61

36.92

37.01

37.0737.1137.14

37.17

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

0

0

Latitude (oS)

0 0

1000 1000

2000 2000

3000 3000

4000 4000

5000 5000

Dep

th(m

)

(c) Exp. 236.79

36.61

36.92

37.01

37.0737.11

37.1437.17

-60

-60

-50

-50

-40

-40

-30

-30

-20

-20

-10

-10

0

0

Latitude (oS)

0 0

1000 1000

2000 2000

3000 3000

4000 4000

5000 5000

Dep

th(m

)

(d) Exp. 3Figure 5.11: Positions of (a) climatological isopycnals from Levitus and Boyer (1994) andLevitus et al. (1994), and (b, c, d) simulated isopycnals in September of CFC-11 year 1993 inExps. 1, 2 and 3 along the WOCE P19 section.120

60 120 180 240 300Longitude(oE)

-70

-60

-50

-40

Lat

itude

(o S)

0.015 0.03 0.05 0.1 0.2 0.4 0.8 1.6 3.2

(a) Exp. 1 60 120 180 240 300Longitude(oE)

-70

-60

-50

-40

Lat

itude

(o S)

0.015 0.03 0.05 0.1 0.2 0.4 0.8 1.6 3.2

(b) Exp. 2Figure 5.12: Simulated CFC-11 (pmol kg�1) on the 37.17-isopycnal in (a) Exp. 1 and (b)Exp. 2. Cut-o� value is 0.01 pmol kg�1. The WOCE P19 section is shown at 272�E.60 120 180 240 300

Longitude(oE)

-70

-60

-50

-40

Lat

itude

(o S)

0.015 0.03 0.05 0.1 0.2 0.4 0.8 1.6 3.2

(a) Exp. 1 60 120 180 240 300Longitude(oE)

-70

-60

-50

-40

Lat

itude

(o S)

0.015 0.03 0.05 0.1 0.2 0.4 0.8 1.6 3.2

(b) Exp. 2Figure 5.13: As Fig. 5.12, but for the 37.14-isopycnal.the isopycnal distribution of CFC-11 is displayed in Fig. 5.12. The CFC-11 distribution clearlyindicates that the source of the ventilation is located in the Weddell Sea, the Ross Sea andwest of the Antarctic Peninsula in Exp. 1. In Exp. 2, the ventilation source is located in theWeddell Sea and to some extent west of the Antarctic Peninsula. Therefore, the presence ofCFC-11 at the bottom along the WOCE P19 section at 272�E is caused by isopycnal transportand mixing from east.Likewise, the CFC-11 distribution on the 37.14-isopycnal is displayed in Fig. 5.13. InExp. 1, the 37.14-isopycnal is heavily ventilated in the Weddell Sea, west of the AntarcticPeninsula and in the India Ocean sector. In contrast, the 37.14-isopycnal is only ventilated inthe Weddell Sea and west of the Antarctic Peninsula in Exp. 2. The latter �nding con�rmsthe absence of the CFC-11 on this isopycnal along the WOCE P19 section in Exp. 2 (seeFig. 5.9c). 121

5.4 Discussion and ConclusionSensitivity of the diapycnal mixing on the oceanic uptake of anthropogenic CFC-11 and theventilation in the SO (here de�ned as the region between 69�S and 45�S) are investigatedwith a near global isopycnic coordinate ocean model. Three simulations have been performedwith a prescribed diapycnal mixing Kd (m2 s�1) of 2� 10�7=N (Exp. 1), vanishing diapycnalmixing (Exp. 2), and 5 � 10�8=N (Exp. 3; N (s�1) is the Brunt-V�ais�al�a frequency). InExp. 2, transport and mixing of all of the model variables occur along the isopycnal layers asthe isopycnic interfaces are truly material surfaces for Kd = 0.The performed model simulations indicate that the observed vertical distribution of CFC-11 along 272�E (Fig. 5.9a) in the SO is caused by two processes: Local ventilation of thesurface waters leading to elevated CFC-11 concentrations over the uppermost 1000-2000 mof the water column, and westward-directed isopycnic transport and mixing from deeplyventilated waters in the Weddell Sea region.In the model, the deepest ventilation takes place as convective mixing to more than 2000 min the Weddell Sea region (Fig. 5.6b) with subsequent transport and mixing along downward-sloping isopycnals (Figs. 5.11c and 5.12b), yielding CFC-11 signals in the bottom waters ofthe SO north to about 55�S. The simulated and observed CFC-11 distributions show the samelarge-scale features (see Figs. 5.9a and 5.9c), although the CFC-enriched bottom water in theobservation is possibly a result of a combination of deep convective mixing in the WeddellSea and geostrophically-driven density currents formed under the Weddell Sea ice cap (e.g.,Killworth, 1977).The presented model experiments clearly show that the uptake of CFC-11 in the SOand the subsequent spreading of the CFC-enriched water masses are sensitive to the applieddiapycnal mixing. In fact, at the end of 1997, the simulated net ocean uptake of CFC-11 inExp. 2 is 25% below that of Exp. 1. Of this reduction, the decreased uptake of CFC-11 in theSO accounts for 80% of the total di�erence. It is further shown that weak and even vanishingdiapycnal mixing greatly improve the simulated CFC-11 distribution in the SO. Apparently,Exp. 3 with Kd = 5 � 10�8=N produced a vertical CFC-11 distribution that resembles theobserved distribution quite well.The performed triple-experiment also highlights the sensitivity of the SO surface waterventilation to the distribution of the simulated ML. Rather small di�erences in the maximumML depth may easily generate huge di�erences in the temporal-spatial properties of theventilated waters. One of the major challenges in climate modeling is therefore to simulatethe ventilation of the SO in a realistic manner to avoid, for instance, excessive ocean uptake ofCO2 and heat. This is a tremendous challenge for coupled atmosphere-sea ice-ocean models122

as proper representation of the ML dynamics in the region depend on the surface buoyancy(i.e., the heat and fresh water) and momentum forcing, the presence and the actual simulatedseasonal cycle of sea ice, the release of ice bergs from the massive Antarctic ice shelves andAntarctic melt water in general. Furthermore, the vertical density strati�cation in the SOis weak, so small changes in the simulated hydrography in the region may easily result inunrealistic ventilation rates of the ML properties.The presented sensitivity study shows that if CFCs are included in coupled climate models,�rst order evaluation of decadal-scale ventilation processes can be based on a single transectlike the WOCE P19 transect along 272�E. The inclusion of CFCs in coupled climate models isstraight forward and computationally cheap, and could be used as an e�cient and illuminatingtest-bed for evaluating the ocean uptake of heat and CO2 on decadal time scales.5.5 AcknowledgementsThe study has been supported by the G. C. Rieber Foundations, Norsk Hydro as, the Norwe-gian Research Council through the RegClim and the Programme of Supercomputing projects,and the project GOSAC (ENV4-CT97-0495) under the EC Environment and Climate Pro-gramme. The authors would also like to acknowledge the smooth accessibility of the WOCEobservations through the WOCE Data Assembly Center at the Scripps Institution of Oceanog-raphy (http://whpo.ucsd.edu/).

123

ReferencesAnderson, L. G., & Jones, E. P. (1991). The transport of CO2 into Actic and Antarctic Seas: Similarities anddi�erences in the driving processes. J. Mar. Systems, 2, 81{95.Arakawa, A., & Lamb, V. (1977). Computational design of the basic processes of the UCLAGeneral CirculationModel. Methods Comput. Phys., 17, 174-265.Bleck, R., Rooth, C., Hu, D., & Smith, L. T. (1992). Salinity-driven thermohaline transients in a wind-and thermohaline-forced isopycnic coordinate model of the North Atlantic. J. Phys. Oceanogr., 22,1486-1515.Dixon, K., Bullister, J., Gammon, R., & Stou�er, R. (1996). Examining a coupled climate model using CFC-11as an ocean tracer. Geophys. Res. Letters, 23, 1957-1960.Dutay, J., Bullister, J., Doney, S., Orr, J., Najjar, R., Caldeira, K., Campin, J., Drange, H., Follows, M., Gao,Y., Gruber, N., Hecht, M., Ishida, A., Joos, F., Lindsay, K., Madec, G., Marier-Reimer, E., Marshall,J., Matear, R., Monfray, P., Plattner, G., Sarmiento, J., Schlitzer, R., Slater, R., Totterdell, I., Weirig,M., Yamanaka, Y., & Yool, A. (2002). Evaluation of ocean model ventilation with CFC-11: comparisonof 13 global ocean models. Ocean Modelling, 4 (2), 89-120.Gargett, A. (1984). Vertical eddy in the ocean interior. J. Marine. Res., 42 (2), 359-393.Gaspar, P. (1988). Modeling the seasonal cycle of the upper ocean. J. Phys. Oceanogr., 18, 161{180.Killworth, P. D. (1977). Mixing on the Weddell Sea continential slope. Deep-Sea Res., 24, 427-448.Ledwell, J. R., Montgomery, E., Polzin, K., Laurent, L., Schmitt, R., & Toole, J. (2000). Evidence for enhancedmixing over rough topography in the abyssal ocean. Nature, 403, 179-182.Ledwell, J. R., & Watson, A. J. (1993). Evidence for slow mixing across the pycnocline from an open oceantracer-release experiment. Nature, 364, 701-703.Levitus, S., & Boyer, T. P. (1994). World Ocean Atlas 1994 Volume 4: Temperature. NOAA Atlas NESDIS4. Washington, D.C., USA.Levitus, S., Burgett, R., & Boyer, T. P. (1994). World Ocean Atlas 1994 Volume 3: Salinity. NOAA AtlasNESDIS 3. Washington, D.C., USA.McDougall, T., & Dewar, W. (1998). Vertical mixing, cabbeling and thermobaricity in layered models. J.Phys. Oceanogr., 1458-1480.Muench, R., Padman, L., Howard, S., & Fahrbach, E. (2002). Upper ocean diapycnal mixing in the north-western Weddell Sea. Deep Sea Res., 49, 4843-4861.Oberhuber, J. M. (1988). An atlas based on the 'COADS' data set: The budgets of heat, buoyancy andturbulent kinetic energy at the surface of the global ocean (Tech. Rep. No. 15). Hamburg, Germany:Max-Planck-Inst. f}ur Meteorol. 124

Oberhuber, J. M. (1993). Simulation of the Atlantic Circulation with a coupled Sea Ice - Mixed Layer -Isopycnal General Circulation Model, Part I: Model Description. J. Phys. Oceanogr., 23 (5), 808-829.Orr, J., Maier-Reimer, E., Mikolajewicz, U., Monfray, P., Sarmiento, J., Toggweiler, J., Taylor, N., Palmer,J., Gruber, N., Sabine, C., Quere, C., Key, R., & Boutin, J. (2001). Estimates of anthropogenic carbonuptake from four three-dimensional global ocean models. Global Biogeochem. Cycles, 15 (1), 43-60.Orsi, A., Johnson, G., & Bullister, J. (1999). Circulation, mixing, and production of Antarctic Bottom Water.Prog. Oceanogr., 43, 55-109.Robitaille, D., & Weaver, A. (1995). Validation of sub-grid-scale mixing schemes using CFCs in a global oceanmodel. Geophys. Res. Letters, 22 (21), 2917-2920.Sloyan, B., & Rintoul, S. (2000). Estimates of area-averaged diapycnal uxes from basin-scale budgets. J.Phys. Oceanogr., 30, 2320-2341.Smethie Jr, W. M., Fine, R., Putzka, A., & Jones, E. (2000). Tracing the ow of north atlantic deep waterusing chloro uorocarbons. J. Geophys. Res., 105 (C6), 14297-14323.Spencer, R. (1993). Global Oceanic Precipitation from the MSU during 1979-91 and comparisons to otherClimatologies. J. Climate, 6, 1301-1326.Sun, S. (1997). Compressibility e�ects in Miami Isopycnic Coordinate Ocean Model. Unpublished doctoraldissertation, Univ. of Miami.Toole, J. M., Polzin, K. L., & Schmitt, R. W. (1994). Estimates of Diapycnal Mixing in the Abyssal Ocean.Science, 264, 1120{1123.Walker, S., Weiss, R., & Salameh, P. (2000). Reconstructed histories of the annual mean atmospheric molefractions for halocarbons CFC-11, CFC-12, CFC-113, and carbon tetrachloride. J. Geophys. Res.,105 (C6), 14285-14296.Wanninkhof, R. (1992). Relationship Between Wind Speed and Gas Exchange Over the Ocean. J. Geophys.Res., 97 (C5), 7373-7382.Woodru�, S., Slutz, R., & Jenne, R. (1987). A comprehensive ocean-atmosphere data set. Bull. Amer. Meteor.Soc., 68, 1239-1250.

125

126

Chapter 6Paper V: Simulating transport ofradionuclides in the NorthAtlantic-Arctic areas

127

128

Simulating transport of radionuclides in the North Atlantic-Arctic regionYongqi Gao, Helge Drange�, Mats Bentsen and Ola M. Johannessen�Nansen Environmental and Remote Sensing Center, Bergen, Norway�Also at Geophysical Institute, University of Bergen, Norway(submitted to J. Environ. Radioactivity, 2002)

129

AbstractThe spatial and temporal distributions of the anthropogenic radionuclides 137Cs and 90Sr,originating from nuclear bomb testing and the Sella�eld reprocessing plants in the Irish Sea,are simulated using a global version of the Miami Isopycnic Coordinate Ocean Model (MI-COM). The physical model is forced with daily atmospheric re-analyses �elds for the period1950 to present. It is shown that the radionuclides from the Sella�eld discharge reach theBarents Sea region after 4-5 years, in accordance with observations. The simulation providesa detailed distribution and evolution of the radionuclides over the integration time. For theAtlantic waters o� the coast of Norway and in the southern Barents Sea, the atmosphericfallout dominates over the Sella�eld release up to the mid 1960s and from the early 1990s,whereas Sella�eld is the main source for the two radionuclides in the 1970s and 1980s. It isfurthermore argued that model systems like the one presented here can be used for futureprediction of radioactive contaminations in the Nordic Seas and the Arctic Ocean, for instanceunder various global warming scenarios.

130

6.1 IntroductionA scienti�cally based description, understanding, quanti�cation and potential prediction ofthe spatial distribution, temporal evolution and biogeochemical consequences of human-generated contamination of the environment require integrated use of observations and nu-merical modeling. Here a 3-dimensional numerical ocean general circulation model (OGCM)is used to demonstrate the usefulness, and also to point towards the limitations, of simulatingpathways and levels of trace compounds in the ocean environment. The trace compoundsused in the study are the anthropogenic radionuclides 137Cs and 90Sr originating from nuclearbomb testing and the Sella�eld release to the Irish Sea.The model system used in this work has been validated against observed chloro uorocar-bon (CCl3F and CCl2F2) distributions in the North Atlantic Ocean for the period 1930 topresent (Gao et al., 2002), focussing on multi-annual to decadal scale ventilation and watermass transformation processes in the region. In addition, mass transports in and out of theNordic Seas region have been validated based on observations and observational-based esti-mates for the last 40 years (Nilsen et al., 2002). Both studies show that the model system isable to describe most of the key observed quantities in a fairly realistic manner. It is there-fore believed that the model system is a potentially useful tool in simulating the advectivetransport and dispersive mixing of radionuclides.The objective of the study is to use 137Cs and 90Sr to simulate the transport and dispersionof coastal pollution from Europe to the Arctic Ocean (e.g., Dahlgaard, 1995; Nies et al., 1998),and to assess the relative contribution from the nuclear bomb testing and the Sella�eld releasein the region. The Chernobyl release in 1986 is not included in the present integration.The simulation di�ers from the large scale OGCM simulation by Nies et al. (1998) in thefollowing ways: A fully on-line instead of o�-line integration procedure is adopted; the modelis forced with synoptic (i.e., daily) atmospheric �elds instead of climatological (i.e., monthlymean) forcing �elds; a truly global OGCM with a horizontally stretched grid system is usedinstead of a regional North Atlantic-Arctic model set-up; both atmospheric fallout and theSella�eld discharge are simulated in contrast to the Sella�eld release only; and the integrationtime period is 1950{1999 compared to 1965{1995.The paper is organized as follows: A description of the model system is provided inSection 2. In Section 3, a brief overview of the ocean circulation in the Atlantic-ArcticOcean is given, together with the observed distribution of the 137Cs and 90Sr radionuclides.The radionuclides simulation is described in Section 4. The obtained results are given inSection 5, and the paper is summarized in Section 6.131

6.2 Model descriptionThe model system applied in this study is the OGCM MICOM (Bleck et al., 1992), fullycoupled to a sea-ice module consisting of the Hibler (1979) rheology in the implementation ofHarder (1996), and the thermodynamics of Drange and Simonsen (1996).The model has 23 layers with �xed potential densities, and an uppermost mixed layer(ML) with temporal and spatial varying density. The speci�ed potential densities of the sub-surface layers were chosen to ensure a realistic representation of the major water masses inthe North Atlantic-Nordic Sea region. The densities of the isopycnic layers (in �0-units) are24.12, 24.70, 25.28, 25.77, 26.18, 26.52, 26.80, 27.03, 27.22, 27.38, 27.52, 27.63, 27.71, 27.77,27.82, 27.86, 27.90, 27.94, 27.98, 28.01, 28.04, 28.07 and 28.10. In the horizontal, the model iscon�gured with a local orthogonal grid mesh with one pole over North America and one poleover western part of Asia (Bentsen et al., 1999), yielding a grid-spacing of about 90{120 kmin the North Atlantic-Nordic Seas region.The vertically homogeneous ML utilises the Gaspar (1988) bulk parameterization for thedissipation of turbulent kinetic energy, and has temperature, salinity and layer thickness as theprognostic variables. In the isopycnic layers below the ML, temperature and layer thicknessare the prognostic variables, whereas salinity is diagnostically determined by means of thesimpli�ed equation of state of (Friedrich & Levitus, 1972). The bathymetry is computed asthe arithmetic-mean value based on the ETOPO-5 data base (Data Announcement 88-MGG-02, Digital relief of the Surface of the Earth, NOAA, National Geophysical Data Center,Boulder, Colorado, 1988).The continuity, momentum and tracer equations are discretised on an Arakawa C-gridstencil (Arakawa & Lamb, 1977). The di�usive velocities (di�usivities divided by the size ofthe grid cell) for layer interface di�usion, momentum dissipation, and tracer dispersion are0.015 m s�1, 0.015 m s�1 and 0.005 m s�1, respectively. A ux corrected transport scheme(Zalesak, 1979; Smolarkiewicz & Clark, 1986) is used to advect the model layer thickness andthe tracer quantities.The diapycnal mixing coe�cient Kd (m2 s�1) is parameterized according to the Gargett(1984) expression Kd = 3 � 10�7=N , where N (s�1) is the Brunt-V�ais�al�a frequency. Thenumerical implementation of the diapycnal mixing follows the scheme of McDougall andDewar (1998).The model was initialized by the January Levitus and Boyer (1994) and Levitus et al.(1994) climatological temperature and salinity �elds, respectively, a 2 m thick sea-ice coverbased on climatological sea-ice extent, and an ocean at rest. The model was �rst spin-upfor 180 years with monthly-mean atmospheric forcing �elds derived from the NCEP/NCAR132

reanalysis (Kalnay et al., 1996). In this spin-up, both sea surface salinity (SSS) and seasurface temperature (SST) was relaxed towards monthly mean surface climatology. Therelaxation was carried out by applying uxes of heat and salt proportional to the SSS andSST di�erences between model and climatology, respectively, with an e-folding time scale of30 days for a ML of 50 m or less, decreasing linearly with thicker ML depths. The spin-up wasthen continued with daily NCEP/NCAR forcing, repeating the period 1974-1978 twice. Onlysalinity relaxation was applied in these two integrations, and the deduced salinity adjustment ux was stored from the latter integration to produce seasonal averaged restoring uxesfor salt. The period 1974-1978 was chosen because of the relatively neutral North AtlanticOscillation (NAO; Hurrell, 1995) conditions of these years (see below). Finally, the modelwas integrated with daily forcing for the period 1948-1999 with no relaxation but with thediagnosed restoring uxes for salt.6.2.1 The tracer moduleThe transport and mixing of the radionuclides follow the equation (Bleck et al., 1992)@@t (@p@sC) +rs � (~v @p@sC)| {z }advection + @@s( _s@p@sC)| {z }mass transfer = rs � (� @p@srsC)| {z }di�usion + �C|{z}source/sink� �C|{z}decay (6.1)Here p is the interface pressure of the isopycnals, C is the concentrations of the tracers, ~v isthe isopycnal velocity, s is the vertical coordinate, � is the di�usive velocity for the tracers,and � denotes the decay of C. Both 137Cs and 90Sr are soluble in sea water and can beconsidered as passive tracers with a half-life of 30.1 and 29 years, respectively.6.3 Observed distribution of water masses and radionuclidesThe Ocean circulation in the region of interest is complex and is characterized by warmand saline surface Atlantic Water (AW) owing through the Denmark Strait, passing theridge between Iceland and the Faroes, and entering east of the Faroes (Hansen & �sterhus,2000). Observational based estimates of the transport of AW through these openings are 1Sv (Hansen & �sterhus, 2000), 3.3 Sv (Hansen & �sterhus, 2000) and 4.3 Sv (Turrell et al.,2002), respectively (1 Sv = 106 m3 s�1). The Atlantic Water (AW) continues northward alongthe Norwegian coast as the Norwegian Atlantic Current (NAC). One branch of the AW entersthe Barents Sea (2.3 Sv; Ingvaldsen et al., 2002), whereas the remaining AW continues to theFram Strait as the West Spitsbergen Current (WSC). Here a fraction of the AW subductsand enters the Arctic Ocean, whereas the rest recirculates and ows southward.133

Greenland

20 cm s-1

Norw

ay

Figure 6.1: The simulated mean circulation �eld in the upper 100 m of the water column forthe period 1950 to 1999.The mean simulated surface current �eld from our model study in the Atlantic-Arcticregion is displayed in Fig. 6.1. The circulation �eld is, in general, in accordance with observa-tions. The most pronounced exception is the too northenly separation of the Gulf Stream o�the North American coast. This is a classical problem for OGCMs; a properly resolved GulfStream requires a horizontal resolution one order of magnitude higher than in the presentintegration (Bleck et al., 1995).Furthermore, the general surface ocean circulation in the Arctic Ocean is dominated by theBeaufort Gyre and the Transpolar Drift (TPD). As already stated, volume exchanges betweenthe Arctic Ocean and the North Atlantic takes place via the Barents Sea and through theFram Strait. The cold and fresh Polar Water leaves the Arctic with the East GreenlandCurrent (EGC) on the western side of the Fram Strait. This water continues southward alongthe coast of Greenland. Most of the polar water ows through the Denmark Strait and entersthe sub-polar gyre. A branch of the cold and fresh Arctic water ows eastward north of theDenmark Strait and is trapped in the cyclonically directed circulation in the Nordic Seas.The pathway of the soluble radionuclides 137Cs and 90Sr from Sella�eld to the Arctic hasbeen summarized by Kershaw and Baxter (1995) (see Fig. 6.2): Initially the tracer is carriednorthward from the Irish Sea as a plume-like structure via the North Channel, and it then ows along the coast of Scotland into the North Sea. The tracer is transported northward134

Figure 6.2: Estimated pathway and transit time (yr) of the Sella�eld signal, adapted fromDahlgaard (1995).before branching o� northern Norway: One branch passes eastwards into the Barents Sea;the other passes through the Fram Strait with the WSC. The main surface water return owoccurs with the EGC. In case the tracer is mixed into sub-surface waters, the return ow willalso include the Denmark Strait over ow, the over ow across the Iceland-Faroe Ridge andthe Faroe-Bank Channel over ow (Hansen & �sterhus, 2000).Figure 6.2 also provides estimated transit times for the Sella�eld release. It follows that thesignal enters the Barents Sea after about 4 years. In the Fram Strait, the transit time is about5 years for the AW and 6-10 years for the southward owing Arctic waters. After additionaltwo years, the Sella�eld signal passes the Denmark Strait, and continues cyclonically alongthe periphery of the sub-polar gyre. It is further speculated that a part of the signal from the135

1950 1960 1970 1980 1990 2000Year

1000

2000

3000

4000

5000

6000

Cs-

137

(10

12B

qy-1

)

100

200

300

400

500

600

700

Sr-

90

(10

12B

qy-1

)

(a) 1950 1960 1970 1980 1990 2000Year

0

200

400

600

800

1000

1200

1400

Atm

os.F

allo

ut(B

qm

-2y-1

)

(b)Figure 6.3: The time evolution of 137Cs (bars) and 90Sr (line) from (a) the Sella�eld release(unit: 1012 Bq y�1) and (b) the atmospheric fallout (unit: Bq m�2 y�1) over Denmark. (S.Nilsen, pers. comm., 2001).sub-polar gyre re-enters the northward owing AW after a total transit time of 14-17 years.6.4 The radionuclides simulationNuclear bomb testing, the Chernobyl accident and release from the European reprocessingplants Sella�eld (the UK) and Cap de La Hague (France) are the most important sources forthe radionuclides in the North Atlantic-Arctic region (Nies et al., 1998). The time evolutionof 137Cs and 90Sr from the Sella�eld release, from the nuclear bomb testing and from thechernobyl accident in 1986 over Denmark (S. Nilsen, pers. comm., 2001) is shown in Fig. 6.3.The atmospheric fallout provides the present background concentration of the surface andsub-surface waters of the region, while the Sella�eld discharge is one of the major sources ofradioactive contamination in the Arctic Ocean since the seventies (Strand et al., 1996). TheSella�eld signal has been observed over the last two decades, making model-evaluation of inparticular 137Cs distributions possible.The latitudinal distribution of the atmospheric 90Sr deposition (UNSCEAR, 1982, seeFig. 6.4) makes it feasible to construct the latitudional distribution of the atmospheric falloutof 137Cs, with the fallout of 137Cs a factor 1.6 higher than that of 90Sr (amap98, 1998). Atotal of four tracers are included in the simulation; the atmospheric fallout of 137Cs and 90Sr,and the Sella�eld discharge of 137Cs and 90Sr. The Chernobyl release in 1986, seen as theisolated peak in the atmospheric fallout in Fig. 6.3b, is not included in the present integration.The initial �elds of all of the radionuclides are set to zero. The simulation starts in 1950 andends in 1999. 136

60-70N 30-40N 0-10N 20-30S 50-60S 80-90SLatitude Band

0

0.5

1

1.5

2

2.5

3

3.5

4

Dep

ositi

on

Den

sity

(kB

q/m

2 )

Figure 6.4: The atmospheric deposition of 90Sr (kBq m�2) vs. latitude (UNSCEAR, 1982).6.5 ResultsBefore extending northward with the NAC, the simulated Sella�eld radionuclides circulatecyclonically in the North Sea (see Figs. 6.1 and 6.5a). One branch enters the Barents Sea,and one branch heads towards the Fram Strait following the WSC. The simulated pathway ofthe Sella�eld release to the Barents Sea is in general agreement with observations (Kershaw& Baxter, 1995, and Fig. 6.2). In the Barents Sea, the Sella�eld signal spreads towardsnorth-east, and one branch enters the northern Kara Sea. The simulated Sella�eld signal haspassed the Denmark Strait by the EGC in year 1975 and is found in the Labrador Sea in theearly 1980s. It should be noted that the simulated Sella�eld signal does not extend eastwardof the Kara Sea, instead, the signal propagates westward following the Canadian Archipelagoin the 1980s. It is also seen that the surface 137Cs concentration is quickly decreasing in thelate 1990s.Figures 6.6 and 6.7 display the evolution of the vertically integrated inventory of 137Csand the ratio of the vertically integrated inventory from Sella�eld to that from atmosphericfallout, respectively. Generally, the inventory in the central Arctic Ocean is low due to therather weak intrusion of the radioactive signal from the AW. In the Barents Sea and west ofNorway, the Sella�eld contribution plays the dominant role in the inventory in the 1970s and1980s. In the central Nordic Seas, the Sella�eld source dominates the inventory from the mid1980s.The distribution of the Sella�eld signal is closely associated with the simulated circulation�eld in the region. In Fig. 6.8, the circulation anomalies associated with years with high (1957,137

61, 73, 75, 76, 81, 83, 84, 89, 90, 92, 93, 94, 95, 99) and low (1958, 60, 62, 63, 64, 65, 66, 68, 69,70, 71, 77, 79, 87, 96) North Atlantic Oscillation index (NAO; Hurrell, 1995) are displayed.The high and low NAO anomalies are associated with the most profound atmospheric wintertime anomalies in the region (Visbeck et al., 2002), and therefore illustrate the degree of long-term (i.e., multi-annual) variability in the surface circulation �eld. It should be mentionedthat short-term variability occurring on time scales of days to months will far exceed thevariability depicted in Fig. 6.8. The robust long-term changes, however, are expected tofollow the changes in Fig. 6.8.For high NAO-forcing (i.e., for fairly persistent and strong westerlies in the North At-lantic), there is an enhanced northward transport of water east of the Faroes and southwardtransport through the Denmark Strait (Fig. 6.8a). This result is in general accordance withobservational based estimates (Nilsen et al., 2002). There are also enhanced surface trans-ports into the Barents Sea and of Arctic waters out of the Arctic Ocean through the FramStrait. For years with low NAO-forcing, the anomalies oppositely mirror those of the highNAO-forcing closely (Fig. 6.8a).The temporal variability of the simulated volume transports over the uppermost 100 m ofthe water column for some of the major openings in the Nordic Seas-Arctic region is providedin Fig. 6.9. The trend seen in these panels are in general agreement with the graduallyincreasing NAO-index over the time period from the early 1960s to the mid 1990s (Hurrell,1995), and points toward the need for using synoptic rather than climatological forcing �eldsfor tracer simulations.To qualitatively and quantitatively evaluate the simulated transport and mixing of theSella�eld discharge, time series of the observed and simulated surface 137Cs concentrationeast of Scotland (57.0-57.5�N, 1.5-2.0�W), west of Norway (59-61�N, 3.5-5.0�E) and in thesouth-western Barents Sea region (71-72�N, 20-30�E) are shown in Fig. 6.10. The simulated90Sr signal is also shown in the �gure. Superimposed on the panels, there is a histogram ofthe annual mean Sella�eld discharge rates.Clearly, the atmospheric fallout dominates the surface 137Cs and 90Sr distributions in mid1960s in the three regions. However, from the late 1970s to the late 1980s, the Sella�eld dis-charge heavily dominates the surface concentrations. For the last 5-10 years of the simulation,the surface concentrations of 137Cs and 90Sr are mainly governed by the atmospheric fallout.The temporal and spatial evolution of the surface 137Cs concentrations are in broad agree-ment with observations (Fig. 6.10a, c, e). However, the simulated surface 137Cs concentrationsare generally lower than the observed values. For instance, the maximum concentration in theBarents Sea is about 75% of the observed value. This �nding, at least up to the Chernobylaccident in 1986, may indicate that the applied model resolution is too coarse. It is also likely138

Greenland Norw

ay

(a) December 1970Greenland N

orway

(b) December 1975Greenland N

orway

(c) December 1980Greenland N

orway

(d) December 1985Greenland N

orway

(e) December 1990Greenland N

orway

(f) December 1995Greenland N

orway

(g) December 1999105.0

95.085.075.065.055.045.035.025.015.0

5.02.00.80.0

137Cs (Bq m-3)

(h) Concentration scaleFigure 6.5: Simulated surface distribution of 137Cs (Bq m�3) in December 1970 to December1999 caused by the Sella�eld release. Concentration scale in (h), with a cut-o� value of 1 Bqm�3.139

Greenland Norw

ay

(a) December 1970Greenland N

orway

(b) December 1975Greenland N

orway

(c) December 1980Greenland N

orway

(d) December 1985Greenland N

orway

(e) December 1990Greenland N

orway

(f) December 1995Greenland N

orway

(g) December 199932000.0

16000.0

10000.0

8000.0

6000.0

4000.0

2000.0

Inventory of 137Cs (Bq m-2)

(h) Inventory scaleFigure 6.6: Simulated vertically integrated inventory of 137Cs (Bq m�2) in December 1970 toDecember 1999 caused by the Sella�eld release and atmospheric fallout. Inventory scale in(h), with a cut-o� value of 2000 Bq m�2.140

Greenland Norw

ay

(a) December 1970Greenland N

orway

(b) December 1975Greenland N

orway

(c) December 1980Greenland N

orway

(d) December 1985Greenland N

orway

(e) December 1990Greenland N

orway

(f) December 1995Greenland N

orway

(g) December 199910.0

5.0

1.0

0.5

0.1

0.0

Ratio

(h) Ratio scaleFigure 6.7: Simulated ratio of the vertically integrated inventory of the Sella�eld release tothat of the atmospheric fallout in December 1970 to December 1999. Ratio scale in (h), witha cut-o� value of 0.1.141

Greenland

3 cm s-1

Norw

ay

(a) Surface ow anomalies for high NAOGreenland

3 cm s-1

Norw

ay

(b) Surface ow aomalies for low NAOFigure 6.8: The simulated surface circulation anomalies for high (a) and low (b) NAO indexes.142

1955 1960 1965 1970 1975 1980 1985 1990 19951.5

2

2.5

3

Vol

um

eF

lux

(Sv)

(a) Faroe-Scotland 1955 1960 1965 1970 1975 1980 1985 1990 19951

1.5

2

2.5

Vol

um

eF

lux

(Sv)

(b) Barents opening1955 1960 1965 1970 1975 1980 1985 1990 1995

1.5

2

2.5

3

Vol

um

eF

lux

(Sv)

(c) Fram Strait 1955 1960 1965 1970 1975 1980 1985 1990 19951.5

2

2.5

3

Vol

um

eF

lux

(Sv)

(d) Denmark StraitFigure 6.9: Simulated time series of the volume transports (in Sv) over the upper 100 m for(a) the northward transport through the Faroe-Shetland Chanel, (b) the northward transportthrough the Barents Opening, (c) the southward transport through the Fram Strait, and(d) the southward transport through the Denmark Strait. The thick solid lines indicate theaveraged surface transport over the period from 1950 to 1999 (2.2 Sv, 1.8 Sv, 2.1 Sv and 2.2Sv, respectively).

143

that reduced diapycnal and isopycnal mixing rates of a factor 2-4 would maintain higher peaktracer concentrations (cfr. Gao et al., 2002).Based on the simulated 137Cs concentration �elds, the time for the Sella�eld release toreach eastern Scotland is 2 years, 4 years for western Norway and 4-5 years for south-westernBarents Sea. The transit time is here de�ned as the di�erence in time when the maximumconcentration occurs in the speci�c region and when the maximum Sella�eld release occurred.The simulated transit times are in accodance with earlier studies (Kershaw & Baxter, 1995;Livingston et al., 1984; and Fig. 6.2).6.6 Discussion and conclusionThe quality or realism of simulated tracer distributions is by necessity constrained by thesimulated water mass transport and transformation rates. The classical way to validate anOGCM is to perform some base-line integrations, and thereafter compare the simulated �elds(temperature, salinity, velocity, and derived parameters thereof) with observations. Based onthis comparison, model parameters can be tuned, parameterization schemes changed or thenumerics improved. In addition, inherent model de�ciencies can be uncovered.Unfortunately, the temporal-spatial distribution of hydrodynamic and dynamic in situocean observations are sparse. In addition, apparently realistic simulated hydrography canlead to erroneous current �elds as demonstrated by e.g. Toggweiler et al. (1989). For thisreason, validation of OGCMs based on well-documented tracer distributions with well-de�nedsource functions (e.g., chloro uorocarbons and natural and bomb-produced radiocarbon), inaddition to the conventional hydrographic and transport validations, represent a powerfuland cost-e�cient OGCM validation strategy (England & Holloway, 1998; England & Maier-Reimer, 2001).A well-tested and validated numerical model system can be used to provide guidance ofthe short-term (days) to long-term (decadal) transport and spreading of trace compounds forprescribed source functions. This can be done by performing a series of integrations withthe OGCM forced with atmospheric �elds provided by an atmospheric forecast system withpredictable skill of typically 4-6 days, by forcing a series of OGCM simulations with combina-tions of observed reanalyses �elds available for the last 50 years (e.g. from the NCAR/NCEPre-analyses project, Kalnay et al., 1996) or, for a global warming scenario, by adopting forc-ing �elds from a fully coupled climate model. For all types of experiments, the ensemblemean pathway and concentration level, together with statistics describing the spreading ofthe ensemble members, yield information about the likely spatial-temporal evolution of thetracers. These results are valuable as input for optimizing the observational strategy and to144

assess transit times and possible transport routes of the tracers.In addition, the circulation �eld and water mass transformation rates from an OGCM canbe used to identify possible source locations of various trace compounds based on isolatedobservations. The use of such an inverse modeling approach is a powerful tool for guiding anobservational network to monitor the origin and the spatial-temporal evolution of the tracers.In the model study of Nies et al. (1998), monthly mean climatological forcing �elds wereused, yielding a simulated transport time from Sealla�eld to the Barents Sea of 6 years.In the simulation presented here, the corresponding time is 4-5 years, or approximately thetransit time based on observations. The di�erence highlights the importance of using themore energetic synoptic forcing �elds to drive the model. In addition, synoptic forcing �eldswill, in general, result in year-to-year variations in the simulated current �elds (Figs. 6.8and 6.9), and thereby in uence the simulated speed and pathway of the tracers. For instance,the spreading of the Sella�eld signal in the Arctic Ocean in this study is quite di�erent fromthat in Nies et al. (1998). In the latter study, peak values of the Sella�eld signal were obtainedin 1985 and 1990 at the North Pole, whereas mainly found on the Atlantic side of the Pole inthis study (Fig. 6.5). It is hard to assess which model realisation is the most realistic. It iswell known, however, that the sea ice and surface water circulation in the Arctic Ocean arehighly determined by the natural variability modes of the atmospheric circulation, notablythe NAO and the closely associated Arctic Oscillation (AO; Mysak, 2001). It is therefore ofparticular importance to use realistic atmospheric forcing �elds when simulating transportand mixing of tracers in the Arctic Ocean.The synoptic-forced integration presented here shows some of the features that can bededuced from an OGCM tracer simulation. It is, as an example, interesting to note that inthe AW o� the coast of Norway, the atmospheric fallout of 137Cs and 90Sr dominates overthe Sella�eld release from the early 1990s and onwards (see Fig. 6.7f and g). The degree ofrealism of this �nding highly depends on the quality of the simulated transport and mixingprocesses, and the representativeness of the applied time history of the Sealla�eld releaseand the atmospheric fallout. It is also interesting to note how the radionuclides signal �rstbuilds up and thereafter diminishes in the Nordic Seas-Arctic Ocean region. This shift ismost clearly seen in the 1980s when a substantial fraction of the radionuclides is transportedsouthward across the Greenland-Scotland Ridge, before continuing southward in the AtlanticOcean as a part of the geostrophocally-controlled Deep Western Boundary Current o� thecoast North America (particularly Figs. 6.6e-g).145

1955 1960 1965 1970 1975 1980 1985 1990 1995Year

0

100

200

300

400

500

600

137 C

s(B

qm

-3)

0

1000

2000

3000

4000

5000

6000

Rel

ease

(1012

Bq

)

(a) 137Cs east of Scotland (57.0-57.5�N, 1.5-2.0�W) 1955 1960 1965 1970 1975 1980 1985 1990 1995Year

0

5

10

15

20

25

90S

r(B

qm

-3)

0

100

200

300

400

500

600

700

Rel

ease

(1012

Bq

)

(b) As a), but for 90Sr1955 1960 1965 1970 1975 1980 1985 1990 1995

Year

0

20

40

60

80

100

120

140

137 C

s(B

qm

-3)

0

1000

2000

3000

4000

5000

6000

Rel

ease

(1012

Bq

)

(c) 137Cs west of Norway (59-61�N, 3.5-5.0�E) 1955 1960 1965 1970 1975 1980 1985 1990 1995Year

0

5

10

15

20

90S

r(B

qm

-3)

0

100

200

300

400

500

600

700

Rel

ease

(1012

Bq

)

(d) As a), but for 90Sr1955 1960 1965 1970 1975 1980 1985 1990 1995

Year

0

5

10

15

20

25

30

35

40

45

137 C

s(B

qm

-3)

0

1000

2000

3000

4000

5000

6000

Rel

ease

(1012

Bq

)

(e) [137Cs in the Barents Sea (71-72�N, 20-30�E) 1955 1960 1965 1970 1975 1980 1985 1990 1995Year

0

2.5

5

7.5

10

90S

r(B

qm

-3)

0

100

200

300

400

500

600

700

800

Rel

ease

(1012

Bq

)

(f) As a), but for 90SrFigure 6.10: Time series of the imposed Sella�eld release rates of (1012 Bq y�1; bars), theobserved 137Cs surfce concentration (Bq m�3) from Kershaw and Baxter (1995) in circles, andthe simulated surface concentration for atmsopheric fallout (dashed lines) and the sum of theSella�eld release and the atmospheric fallout (solid lines) for East of Scotland (upper row),West of Norway (mid row) and in the South-West Barents Sea region (lower row). 137Cs tothe left and 90Sr to the right. Note the di�erent concentration scales on the panels.146

AcknowledgementsThis study is supported by the EU-project Simulation Scenarios for Potential RadioactiveSpreading in the 21 Century from Rivers and External Sources in the Russian Arctic CoastalZone (RADARC; ICA2-CT-2000-10037). The authors are grateful to Dr. Sven Nielsen forproviding time histories of the radionuclides. The model development has been supportedby the Research Council of Norway through the RegClim project, KlimaProg's \Spissforskn-ingsmidler", and the Programme of Supercomputing. The authors are grateful to L. H.Petterson, NERSC, for administration of the RADARC project. Support from the G. C.Rieber Foundations is highly acknowledged.

147

ReferencesAMAP assessment report: Arctic pollution issues. (1998). Oslo, Norway: Arctic Monitoring and AssessmentProgramme (AMAP).Arakawa, A., & Lamb, V. (1977). Computational design of the basic processes of the UCLAGeneral CirculationModel. Methods Comput. Phys., 17, 174-265.Bentsen, M., Evensen, G., Drange, H., & Jenkins, A. D. (1999). Coordinate Transformation on a Sphere UsingConformal Mapping. Mon. Wea. Rev., 127, 2733{2740.Bleck, R., Dean, S., Keefe, M. O., & Sawdey, A. (1995). A Comparison of Data-Parallel and Message-PassingVersions of the Miami Isopycnic Coordinate Ocean Model. Parallel Computing, 21, 1695-1720.Bleck, R., Rooth, C., Hu, D., & Smith, L. T. (1992). Salinity-driven thermohaline transients in a wind-and thermohaline-forced isopycnic coordinate model of the North Atlantic. J. Phys. Oceanogr., 22,1486-1515.Dahlgaard, H. (1995). Transfer of European Coastal Pollution to the Arctic: Radioactive Tracers. MarinePollution Bulletin, 31, 3-7.Drange, H., & Simonsen, K. (1996). Formulation of air-sea uxes in the ESOP2 version of MICOM (Tech.Rep. No. 125). Edv. Griegsv. 3A, N-5059 Solheimsviken, Norway: Nansen Environmental and RemoteSensing Center.England, M., & Holloway, G. (1998). Simulations of CFC content and water mass age in the deep NorthAtlantic. J. Geophys. Res., 103 (C8), 15885-15901.England, M., & Maier-Reimer, E. (2001). Using Chemical Tracers to Assess Ocean Models. Rev. of Geophys.,39 (1), 29-70.Friedrich, H., & Levitus, S. (1972). An approximation to the equation of state for sea water, suitable fornumerical ocean models. J. Phys. Oceanogr., 2, 514-517.Gao, Y., Drange, H., & Bension, M. (2002). E�ects of diapycnal and isopycnal mixing on the ventilation ofCFCs in the North Atlantic in an isopycnic coordinate OGCM. (re-submitted to Tellus)Gargett, A. (1984). Vertical eddy in the ocean interior. J. Marine. Res., 42 (2), 359-393.Gaspar, P. (1988). Modeling the seasonal cycle of the upper ocean. J. Phys. Oceanogr., 18, 161{180.Hansen, B., & �sterhus, S. (2000). North Atlantic-Nordic Seas exchanges. Prog. Oceanogr., 45 (2), 109-208.Harder, M. (1996). Dynamik, Rauhigkeit und Alter des Meereises in der Arktis. Unpublished doctoraldissertation, Alfred-Wegner-Institut f�ur Polar- ind Meeresforschung, Bremerhaven, Germany.Hibler, W. (1979). A dynamic thermodynamic sea ice model. J. Phys. Oceanogr., 9, 815-846.148

Hurrell, W. (1995). Decadal trends in the north atlantic oscillation: Regional temperatures and precipitation.Science, 269, 676-679.Ingvaldsen, R., Loeng, H., & Asplin, L. (2002). Variability in the Atlantic in ow to the Barents Sea based ona one-year time series from moored current meters. Continental Shelf Research, 22, 505-519.Kalnay, E., et al.. (1996). The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77 (3),437{471.Kershaw, P., & Baxter, A. (1995). The transfer of reprocessing wastes from north-west Europe to the Arctic.Deep Sea Res., 42, 1413-1448.Levitus, S., & Boyer, T. P. (1994). World Ocean Atlas 1994 Volume 4: Temperature. NOAA Atlas NESDIS4. Washington, D.C., USA.Levitus, S., Burgett, R., & Boyer, T. P. (1994). World Ocean Atlas 1994 Volume 3: Salinity. NOAA AtlasNESDIS 3. Washington, D.C., USA.Livingston, H., Kupferman, S., Bowen, V., & Moore, R. (1984). Vertical pro�le of arti�cial radionuclideconcentrations in the central Arctic Ocean. Geichimica et Cosmochimica Acta, 48, 2195-2203.McDougall, T., & Dewar, W. (1998). Vertical mixing, cabbeling and thermobaricity in layered models. J.Phys. Oceanogr., 1458-1480.Mysak, L. A. (2001). Patterns of Arctic Circulation. Science, 293, 1269-1270.Nies, H., Harms, H., Karcher, M., Dethle�, D., Bahe, C., Kuhlmann, G., Oberhuber, J., Backhaus, J., Kleine,E., Loewe, P., Matishov, D., Stepanov, A., & Vasiliev, O. (1998). Anthropogenic Radioactivity in theNordic Seas and the Arctic Ocean - Results of a Joint Project. Deutsche Hydrographische Zeitschrift,50 (4), 313-343.Nilsen, J. E. �., Gao, Y., Drange, H., Furevik, T., & Bentsen, M. (2002, nov). Simulated North Atlantic{NordicSeas water mass exchanges in an isopycnic coordinate OGCM. (submitted to Geophys. Res. Letters)Smolarkiewicz, P. K., & Clark, T. L. (1986). The multidimensional positive de�nite advection transportalgorithm: Further development and applications. J. Comp. Phys., 67, 396-438.Strand, P., Aarkrog, A., Bewers, J., Tsaturov, Z., & Magnusson, S. (1996). Radioactive Contamination ofthe Arctic Marine Environment. In Radionuclides in the Oceans - Inputs and Inventories (p. 95-119).Inst. de Protection et de Surete Nucleaire, France.Toggweiler, J., Dixon, K., & Bryan, K. (1989). Simulations of radiocarbon in a coarse-resolution world oceanmodel, 1, Steady state prebomb distributions. J. Geophys. Res., 94, 8217-8242.Turrell, W. R., Hansen, B., Hughes, S., & �sterhus, S. (2002). Hydrographic variability during the decade ofthe 1990s in the Northeast Atlantic and southern Norwegian Sea. In ICES Symposium on HydrobiologicalVariability in the ICES Area, 1990-99. (ICES Mar. Sci. Symp., ICES.)UNSCEAR. (1982). Ionizing Radiation: Sources and Biological E�ects. In Unscear 1982 report to the generalassembly with scienti�c annexes. United Nations, New York: United Nations Scienti�c Committee onthe E�ects of Atomic Radiation.Visbeck, M., Chassignet, R., Delworth, T., Dickson, B., & Krahmann, G. (2002). The Ocean's Response toNorth Atlantic Oscillation Variability. In J. Hurrell, Y. Kushnir, G. Ottersen, & M. Visbeck (Eds.),The North Atlantic Oscillation. AGU monograph. (in press)Zalesak, S. (1979). Fully multidimensional ux-corrected transport algorithms for uids. J. Comp. Physics,31, 335{362. 149