What controls primary production in the Arctic Ocean? Results...

What controls primary production in the Arctic Ocean?Results from an intercomparison of five general circulationmodels with biogeochemistry

Ekaterina E. Popova,1 Andrew Yool,1 Andrew C. Coward,1 Frederic Dupont,2 Clara Deal,3

Scott Elliott,4 Elizabeth Hunke,4 Meibing Jin,3 Mike Steele,5 and Jinlun Zhang5

Received 10 March 2011; revised 26 August 2011; accepted 22 November 2011; published 31 January 2012.

[1] As a part of Arctic Ocean Intercomparison Project, results from five coupled physicaland biological ocean models were compared for the Arctic domain, defined here as northof 66.6°N. The global and regional (Arctic Ocean (AO)–only) models included in theintercomparison show similar features in terms of the distribution of present-day watercolumn–integrated primary production and are broadly in agreement with in situ andsatellite-derived data. However, the physical factors controlling this distribution differbetween the models. The intercomparison between models finds substantial variation in thedepth of winter mixing, one of the main mechanisms supplying inorganic nutrients overthe majority of the AO. Although all models manifest similar level of light limitation owingto general agreement on the ice distribution, the amount of nutrients available for planktonutilization is different between models. Thus the participating models disagree on afundamental question: which factor, light or nutrients, controls present-day Arcticproductivity. These differences between models may not be detrimental in determiningpresent-day AO primary production since both light and nutrient limitation are tightlycoupled to the presence of sea ice. Essentially, as long as at least one of the two limitingfactors is reproduced correctly, simulated total primary production will be close to thatobserved. However, if the retreat of Arctic sea ice continues into the future as expected, adecoupling between sea ice and nutrient limitation will occur, and the predictive capabilitiesof the models may potentially diminish unless more effort is spent on verifying themechanisms of nutrient supply. Our study once again emphasizes the importance of arealistic representation of ocean physics, in particular vertical mixing, as a necessaryfoundation for ecosystem modeling and predictions.

Citation: Popova, E. E., A. Yool, A. C. Coward, F. Dupont, C. Deal, S. Elliott, E. Hunke, M. Jin, M. Steele, and J. Zhang (2012),What controls primary production in the Arctic Ocean? Results from an intercomparison of five general circulation models withbiogeochemistry, J. Geophys. Res., 117, C00D12, doi:10.1029/2011JC007112.

1. Introduction

[2] Modeling of the Arctic Ocean biophysical interactionshas a long history starting probably from the work ofNiebauer and Smith [1989] and Smith and Niebauer [1993]on the impact of the mesoscale physical process on thephytoplankton dynamics within the framework of the two-dimensional numerical model. Since this pioneering work,a number of modeling studies of various complexity were

conducted in a variety of regions of the AO such as Chukchi[Walsh et al., 2004, 2005] and Barents [e.g., Wassmannet al., 2006; Ellingsen et al., 2008] seas. However, untilrecently, the Arctic Basin received little attention in eitherglobal- or even basin-scale ecosystem modeling studiesbecause of its status as a (relatively) low-productivity area.Nevertheless, the recent and ongoing retreat of Arctic seaice is exposing increasingly large surface areas of the basinto sunlight and thus promoting increased growth of phyto-plankton during summer months. A recent study reportingelevated Arctic primary production [Arrigo et al., 2008]greatly increased interest in modeling the present-day Arcticecosystem and its possible future under continuing climatechange. As a result, the first pan-Arctic biogeochemicalmodeling studies are now appearing [e.g., Lengaigne et al.,2009; Slagstad et al., 2011; Popova et al., 2010; Zhanget al., 2010; Jin et al., 2011; F. Dupont, Effect of sea-icealgae in a biophysical model of the Arctic Ocean, submittedto Journal of Geophysical Research, 2011] posing the

1National Oceanography Centre, Southampton, University ofSouthampton, Southampton, UK.

2RPN MRD, Environment Canada, Dorval, Quebec, Canada.3International Arctic Research Center, University of Alaska Fairbanks,

Fairbanks, Alaska, USA.4Los Alamos National Laboratory, Los Alamos, New Mexico, USA.5Polar Science Center, University of Washington, Seattle, Washington,

USA.

Copyright 2012 by the American Geophysical Union.0148-0227/12/2011JC007112

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 117, C00D12, doi:10.1029/2011JC007112, 2012

C00D12 1 of 16

http://dx.doi.org/10.1029/2011JC007112

question as to what extent different models agree on present-day AO primary production and on the factors that control it.Assessing model performance (against observations) andagreement (against one another) will greatly increase ourunderstanding of how much their future predictions mayvary, and why.[3] In this paper we present an intercomparison of 5 cou-

pled physical-biological ocean models for the Arctic domain,made within the framework of the Arctic Ocean ModelIntercomparison Project (AOMIP). The models examinedhave separate and distinct development histories for boththeir physics and biology, but overlap sufficiently in domainand functionality for a productive intercomparison. Sinceboth global and regional AO models participated in theintercomparison, the Arctic Ocean domain was defined hereas the area north of 66.6°N.[4] Ecosystem model intercomparisons in any geographi-

cal area face a number of challenges. For instance, differentecosystem models vary substantially in their structure, com-plexity, parameterizations of ecosystem processes and theirequilibration time. Furthermore, they are embedded intophysical models which vary in initial and boundary condi-tions, geographical domains (ranging from limited regions toglobal), resolution, numerical methods, sub-grid-scale para-meterizations, and external surface forcing. An extra chal-lenge in the Arctic domain is the overwhelming importanceof sea ice, models for which vary significantly in terms oftheir complexity, the processes represented and in their cou-pling to underlying ocean dynamics.[5] Taking into account the various difficulties mentioned

above, one may envisage two approaches for model inter-comparison. One is to create a common protocol in whichparticipating models are run under certain identical specifi-cations such as atmospheric forcing, equilibration procedureand boundary conditions. For example, such an approachwas successfully used in OCMIP (Ocean Carbon cycleModel Intercomparison Project) where, additionally, an ide-alized ecosystem model was a specified component of theprotocols [Najjar et al., 2007]. Although such an endeavorremoves many sources of incongruity between models andallows direct intercomparison of fundamental differencesbetween models, it requires a substantial investment ofresources from all participating groups since it necessitatesrerunning all the models under a common protocol. This isexpensive both in computational costs (especially, as inOCMIP, where equilibrium criteria were very specific), andin the effort required to accommodate the standard protocolwithin an existing model framework (e.g., coding, file-formatting, testing). An alternative approach is to considerexisting runs with all their underlying differences and attemptto quantify the skills of the models and, as a first step, toassess if the main features of the simulated ecosystemdynamics are driven by the same physical mechanisms.[6] The model intercomparison project described in this

paper follows the latter approach of using extant simulationswithout a common protocol for initialization and forcing.Five models of the Arctic Ocean (both from the regionalArctic or global modeling efforts) are compared with a focuson the water column primary production, inorganic macro-nutrients and physical factors that are shaping them. In allfive models the integration period is relatively short, andour analysis focuses on the present-day seasonal variability

rather than the models’ long-term behavior that determinesthe large-scale pan-Arctic distribution of nutrients.

2. Participating Models

2.1. NEMO

[7] The Nucleus for European Modeling of the Ocean(NEMO) model is composed of an ocean general circulationmodel, OPA [Madec, 2008], coupled with the Louvain-la-Neuve Ice Model v2, LIM2 [Timmermann et al., 2005]. Theversion of NEMO used here is v3.2 and has a horizontalresolution of 1/4°, and a vertical resolution of 64 levels (6 mthickness at the surface to 250 m at 6000 m; fractional bottomgrid boxes permit more realistic bathymetry). The LIM2 seaice submodel is based upon a viscous-plastic rheology[Hibler, 1979] and three layer (two ice; one snow) thermo-dynamics [Semtner, 1976], and includes a number of updatesto physical processes [Timmermann et al., 2005]. NEMOis forced at the surface using DFS4.1 fields developed bythe European DRAKKAR collaboration [DRAKKAR Group,2007]. This combines the CORE data set [Large and Yeager,2004], which provides precipitation and downward short-and long-wave radiation, with the ERA40 reanalysis, fromwhich 10 m wind and 2 m air humidity and temperature aresupplied. The latter are used with the bulk formulae proposedby Large and Yeager [2004] to compute air-sea and air-iceheat and freshwater fluxes. Frequency of the forcing fieldsis monthly for precipitation; daily for radiation; 6 h for tur-bulent variables. Vertical mixing is parameterized using Tur-bulent Kinetic Energy (TKE) scheme [Gaspar et al., 1990]with subsequent modifications by Madec [2008].[8] Biogeochemistry in NEMO is represented by the

plankton ecosystem model MEDUSA (Model of ecosystemDynamics, carbon Utilization, sequestration and Acidifica-tion) [Yool et al., 2011]. This is a size-based intermediatecomplexity model that divides the plankton community into“small” and “large” portions and which resolves the ele-mental cycles of nitrogen, silicon and iron. The “small”portion of the ecosystem is intended to represent the micro-bial loop of picophytoplankton and microzooplankton, whilethe “large” portion covers microphytoplankton (specificallydiatoms) and mesozooplankton. The intention of MEDUSAis to separately represent small, fast-growing phytoplanktonthat are kept in check by similarly fast-growing protistanzooplankton, and large, slower-growing phytoplankton thatare able to temporarily escape the control of slower-growingmetazoan zooplankton. The nonliving particulate detrituspool is similarly split between small, slow-sinking particlesthat are simulated explicitly, and large, fast-sinking particlesthat are represented only implicitly. See Yool et al. [2011] fora full description of MEDUSA.[9] The simulation of NEMO used here was originally

performed to produce a high resolution, global-scale hindcastof marine biogeochemistry during the past two decades.An initial physics-only spin-up simulated the period 1978to 1987 (10 years). The biogeochemistry was then coupledand MEDUSA was spun-up for the period 1988 to 2007(20 years). An analysis of the Arctic region in the resultingsimulation has previously been published by Popova et al.[2010]. Key properties of the model are summarized inTable 1.

POPOVA ET AL.: PRIMARY PRODUCTION IN THE ARCTIC OCEAN C00D12C00D12

2 of 16

2.2. LANL

[10] The POP-CICE (Parallel Ocean Program–Sea ICE)model developed at Los Alamos National Laboratory (LANL)is coupled with an ice and ocean ecosystem model, describedbelow. The model is discretized for a nonuniform, generalcurvilinear grids in which the North Pole has been movedsmoothly into Greenland. For the experiments describedhere, we use a 320 � 384 global mesh. In the NorthernHemisphere the grid size ranges from 18 km (at high lati-tudes) to 85 km (at the equator). The vertical grid consists of40 levels that vary in thickness from 10 m at the surface to250 m below 2000 m depth. Atmospheric forcing datainclude six hourly air temperature, specific humidity, andwind velocity components from the Common Ocean Refer-ence Experiments (CORE) version 2 (1958–2007) [Largeand Yeager, 2009] along with monthly “normal year” pre-cipitation from version 1 [Large and Yeager, 2004]. Thesedata and version 2 of the Ocean Model IntercomparisonProject’s cloud climatology [Röske, 2001] are used to cal-culate radiation fields following the AOMIP protocol, andtotal runoff in the Arctic also is specified following theAOMIP protocol [Hunke and Holland, 2007]. The presentmodel run encompasses 1992–2007. The ocean component,

POP 2.0 [Smith and Gent, 2002], employs the anisotropicGent-McWilliams parameterization of Smith and Gent [2004]for lateral mixing. The K profile parameterization (KPP)[Large et al., 1994] provides vertical mixing of momentumand tracers. The sea ice component, CICE 4.0 [Hunke andLipscomb, 2008], includes Bitz and Lipscomb [1999] ther-modynamics, EVP dynamics [Hunke and Dukowicz, 1997,2002], and horizontal transport via incremental remapping[Lipscomb and Hunke, 2004]. The thickness distributionwithin each grid cell is represented using five thicknesscategories, each with 1 layer of snow atop 4 layers ice.[11] Biological model consists of (1) ice algal submodel

in a 3 cm layer at the bottom of each sea ice thickness cate-gory [Jin et al., 2006, 2007, 2009; Deal et al., 2011] and(2) pelagic ecosystem model with 24 variables [Moore et al.,2004]: nitrate, ammonium, iron, silicate, phosphate, threetypes of phytoplankton (explicit C, Fe, and chlorophyll poolsfor each phytoplankton group, and an explicit Si pool fordiatoms and CaCO3 pool for small phytoplankton totaling11 state variables), zooplankton, dissolved organic carbon,dissolved organic nitrogen, dissolved organic iron, dissolvedorganic phosphate, oxygen, dissolved inorganic carbon, andalkalinity. The two submodels were coupled through nutrient

Table 1. Summary of the Model Properties Directly Relevant to the Ecosystem Functioning

NEMO LANL UW UL OCCAM

Model GridModel domain global global north of 49°N north of 65°N globalAO minimum horizontal

resolution (km)�5 23 �5 55 111

AO maximum horizontalresolution (km)

�15 62 �30 55 111

AO minimum verticalresolution (m)

5 10 5 3.4 5

AO maximum verticalresolution (m)

250 250 600 210 208

Vertical MixingUML depth TKE KPP KPP TKE KPPAtmospheric forcing DFS4.1 Hunke and

Holland [2007]NCEP/NCARreanalysis

NCEP/NCAR reanalysis +temperature correction

for orography

NCEP [see Largeand Yeager, 2004]

Ice ModelIce physics LIM2 CICE 4.0 VP–LSR,

12 thicknesscategories

Hibler VP,one category

EVP model,two thicknesscategories

Ice biology none Deal et al. [2011] none Dupont (submittedmanuscript, 2011)

none

Ecosystem ModelInitialization of nutrients WOA2005 WOA2005 WOA2005 WOA2005 WOA2005Riverine input of nutrients relaxation of N and Si

toward climatologywithin 100 kmof shoreline

none none 4 mmol m�3 and monthlyclimatological river flow

[see Holloway et al., 2007]

none

Number of ecosystemstate variables

11 24 11 4 4

Explicit nutrient cycles N, Si, Fe N, Si, Fe, P, N, Si N N

Model EquilibrationPhysics only (years) 10 34 50 5 0Physics + biology (years) 20 0 30 5 188a

Period available for analysis 1991–2006 1992–2007 1978–2009 1960–2006 1958–2004

aOCCAM was simulated through the 1958–2004 forcing cycle four times. The first was for preindustrial spin-up, with the following three cyclesrepresenting the period 1864–2004 (inclusive), during which time anthropogenic changes to atmospheric pCO2 were simulated. The results presentedhere are taken from the final forcing cycle, at which point OCCAM has undergone more than 140 years of simulation since its initial condition.


3 of 16

and algal fluxes between the ice and ocean models [Jin et al.,2011].[12] Initial conditions for ocean temperature, salinity and

biological variables (NO3, Si) are from the gridded WorldOcean Atlas (WOA2005) [Garcia et al., 2006]. Other oceanbiological model initial conditions are taken from the globalmodel simulation by Moore et al. [2004]. Sea ice initialconditions for the physical model are from an earlier run(1958–2006) [Hunke and Bitz, 2009] without the biologicalmodel. Initial conditions of nutrients at the bottom ice are setto sea surface values. The current version of the model doesnot consider riverine input of nutrients.

2.3. UW Model

[13] The UW model is the Biology/Ice/Ocean Modelingand Assimilation System (BIOMAS). BIOMAS is a coupled3-D model consisting of three model elements: a sea icemodel, an ocean circulation model, and a biological model[Zhang et al., 2010]. The sea ice model is the multicategorythickness and enthalpy distribution (TED) sea ice model[Hibler, 1980; Zhang and Rothrock, 2001, 2003]. It employsa teardrop viscous-plastic rheology [Zhang and Rothrock,2005], a mechanical redistribution function for ice ridging[Thorndike et al., 1975; Hibler, 1980], and a LSR (LineSuccessive Relaxation) dynamics model to solve the icemomentum equation [Zhang and Hibler, 1997]. The oceanmodel is based on the Parallel Ocean Program (POP) devel-oped at Los Alamos National Laboratory [Smith et al., 1992;Dukowicz and Smith, 1994]. The POP ocean model has beenmodified so that open boundary conditions can be specified[Zhang and Steele, 2007]. This modification allows themodel to be nested to a global ice/ocean model of Zhang[2005]. Open boundary conditions are specified along themodel’s southern boundary with sea surface height and oceanvelocity, temperature, and salinity from the global model.Nitrate and silicate are restored to monthly climatology datafrom WOA2005 [Garcia et al., 2006] along the southernboundary. Ocean mixing follows the K profile parameteri-zation, as discussed in the work of Zhang and Steele [2007],while ice-ocean heat flux parameterizations were discussedin the work of Hibler [1980] and Hibler and Bryan [1987].[14] The biological model is based on the work of Zhang

et al. [2010]. It incorporates two phytoplankton compo-nents (diatoms; flagellates), three zooplankton components(microzooplankton; copepods; predator zooplankton), dis-solved organic nitrogen, detrital particulate organic nitrogen,particulate organic silica, nitrate, ammonium, and silicate.[15] The model domain covers the Northern Hemisphere

north of 49°N. Model grid configuration is based on a gen-eralized curvilinear coordinate system with the “North Pole”of the model grid being displaced in Greenland by a linearorthogonal transformation. Mean horizontal resolution ofboth ice and ocean models is about 22 km. The ocean mod-el’s vertical dimension has 30 levels of increasing thicknesswith depth, starting with six 5 m thick levels in the upper30 m and then increasing gradually toward the bottom. Dailymean NCEP/NCAR reanalysis data are used as atmosphericforcing; that is, 10 m surface winds, 2 m surface air temper-ature (SAT), specific humidity, precipitation, evaporation,downwelling long-wave radiation, sea level pressure, andcloud fraction. Cloud fraction and SAT are used to calculatedownwelling short-wave radiation following Parkinson and

Washington [1979]. More model information can be foundin the work of Zhang et al. [2010] and in Table 1.

2.4. UL Model

[16] The ocean model is an extension of the one used byHolloway and Sou [2002] and Steiner et al. [2003] and tookpart in the international intercomparison project for the ice-ocean models of the Arctic [Holloway et al., 2007; Hakkinenet al., 2007]. It is a z level model on a 55 km grid (rotatedgrid, limited to the Pan-Arctic ocean), with 40 levels down to4350 m with the first layer being 3.4 m thick, and includes aparameterization of the “Neptune effect” [Holloway, 1987;Zou and Holloway, 1994] which represents the interactionsbetween oceanic eddies and the bathymetry. This aimsat reproducing certain aspects of the AO general circulationwithout resorting to eddy-resolving resolutions. The modelalso includes the tidal dissipation parameterization ofHolloway and Proshutinsky [2007]. The atmospheric forcingcomes from the NCEP/NCAR reanalysis in the form of daily2 m air temperature, wind velocity, 2 m surface humidity andradiations, with air temperature corrections discussed in thework of Dupont (submitted manuscript, 2011).[17] The biological model is a simple NPZD [Lima

et al., 2002] model with four compartments for nitrate-phytoplankton-zooplankton-detritus (no further subdivisionfor species, size or nutrient) and was modified to the polarenvironment for temperature and light dependence (Dupont,submitted manuscript, 2011). It is able to capture a deepchlorophyll maximum if the phytoplankton sinking rateis fixed to �0.1 m d�1, value retained hereafter. Anotherimportant addition was a sea-ice algae component based onthe work of Lavoie et al. [2005], fit into the same 4 com-partments described above. As such, the sea-ice NPZDmodel retains the same dynamics as for the pelagic model andassumes that all the biology occurs in a thin ice-ocean inter-face of 5 cm. Detritus is exported away from this interface assoon as it is produced, and nutrients are replenished viavertical turbulence flux. During ice growth, it is assumed thatthe phytoplankton present in the water will not be trapped inthe ice matrix but will accumulate at the ice base, in the ice-ocean interface. As such it is assumed that this source ofphytoplanktonic cells will give rise to the ice algae.

2.5. OCCAM

[18] OCCAM (Ocean Circulation and Climate AdvancedModeling project) is a global, primitive equation, finite dif-ference ocean general circulation model. The version ofOCCAM used here has a horizontal resolution of typically1°, and a vertical resolution of 66 levels (5 m thickness at thesurface to 200 m at depth; fractional bottom grid boxes per-mit more realistic bathymetry). OCCAM is organized intotwo horizontal grids to avoid a North Pole singularity, withgrid 2 encompassing the North Atlantic and the ArcticOcean, and grid 1 including the rest of the global ocean. Thesea ice model comprises sea ice dynamics with elastic-viscous-plastic rheology (EVP) [e.g., Hunke, 2001], and icethermodynamics derived from Semtner [1976] with 2 layersfor sea ice and one for snow. The sea ice thermodynamicsincludes lateral and bottom ice melting and accounts forsnow-ice formation. Sea ice is embedded into the upperoceanic layer conserving volume in the sea ice–ocean sys-tem. The dynamical coupling between sea ice and ocean is


4 of 16

done on each baroclinic oceanic time step via the quadraticdrag law [McPhee, 1984]. The details of the OCCAM modelare in the work of Aksenov et al. [2010]. OCCAM includesa K profile parameterization (KPP) mixed layer (see Largeet al. [1997] with modifications described in the work ofPopova et al. [2006]), Gent-McWilliams eddy parameteri-zation and an advection scheme that is fourth-order accurate(a modified split-quick scheme). Surface fluxes of heat,freshwater and momentum are not specified directly, butare instead calculated using empirical formulae and NCEP-derived atmospheric boundary layer quantities (monthly forprecipitation; daily for radiation; 6 h for turbulent variables).See Marsh et al. [2005] for a more complete description ofOCCAM’s physics.[19] OCCAM’s biogeochemical model is a nitrogen-

based nutrient-phytoplankton-zooplankton-detritus model;see Yool et al. [2011] for a complete description of OCCAM’sbiogeochemistry. The simulation of OCCAM used here wasoriginally performed to investigate the invasion of the oceanby anthropogenic CO2, and has been previously published inthe work of Yool et al. [2009, 2010] and Saba et al. [2010].Using multiple passes of the 1958–2004 NCEP forcing data,the simulation comprised one preindustrial spin-up cyclefollowed by three further cycles to represent the period 1864to 2004. During the latter three cycles, atmospheric pCO2

was increased following the historical record. However,since the model plankton ecosystem is unaffected by changesin the carbon cycle, the output used here is drawn from theinitial spin-up phase.

2.6. Summary of Participating Models

[20] As described above, the five models participating inthis intercomparison vary greatly in their numerics, para-meterizations, geographical domain, grid resolution, initial/boundary conditions and in the complexity of their sea iceand ecosystem submodels. What is probably even more sig-nificant is that the models vary in the history of applicationsfor which they were developed, and this typically determinesthe level of sophistication in their descriptions of particularprocesses. While it is outside of the scope of this paper toattempt to analyze the consequences of these differences forecosystem dynamics, some features of particular importancefor the intercomparison project are worth noting.[21] Three out of five participating models are global

(NEMO, OCCAM and LANL) and two are regional (UL andUW), developed specifically for the Arctic Ocean. The AO isstrongly coupled by advective processes to adjacent regions,and in the former cases its solution greatly depends on thequality of the simulation in the upstream regions (northernNorth Pacific and Atlantic), while in the latter cases thesolution is significantly dictated by the quality of informationat the inflow boundaries. Global models provide inflowinformation that is dynamically consistent with the interior ofthe AO. However, they tend to drift away from the obser-vationally derived initial conditions with the increasingduration of the run. Such drift is usually most pronounced inthe biogeochemical properties which have a short (annual)equilibration timescale in the upper ocean. Out of the threeglobal models, OCCAM has the longest unconstrained run(180 years), while NEMO and LANL are both simulated formuch shorter periods of about 30 years and are less likelyto experience significant drift. Regional models are more

heavily constrained by their observation-based boundaryconditions, although in case of the UW model this constraintis imposed on the biogeochemistry only, since physicalboundary conditions are taken from a lower-resolution globalmodel.[22] Model resolution varies significantly from �10 km

(NEMO, UW) to �100 km (OCCAM), with LANL and ULusing an intermediate resolution of �50 km. Although reso-lution significantly impacts the realism of the AO circulationfeatures, especially advection from the upstream areas, noneof the models is capable of resolving the Rossby radiusof deformation which varies between �1 km on the shelvesand �10 km in the deep AO (G. Nurser, personal commu-nication, 2010). Thus, none of the models is capable ofresolving mesoscale and other localized mechanisms of thenutrient supply such as eddy-induced, ice-edge and shelf-break upwelling and topographic effects. Nonetheless, thesignificance of nutrient supply by small-scale mechanismsfor large-scale biogeochemical budgets in the AO is notclear given the current state of observational and modelingunderstanding of AO dynamics.[23] Because of the role they play in determining upper

ocean structure and the availability of light and nutrients,upper mixed layer (UML) dynamics are of key importancefor the plankton ecosystem. They are modeled using a Kprofile parameterization (KPP) in the OCCAM, LANL andUW models, a Turbulent Kinetic Energy (TKE) schemein the UL and NEMO models. Although various theoreticalarguments have been put forward to promote the lattermodel, there appears to be no strong evidence that moreadvanced representations of turbulence, such as the TKEscheme, provides consistently better results in global orbasin-scale models than simpler schemes such as KPP. Forinstance, in a global analysis of UML dynamics in OCCAM,Popova et al. [2006] found that the KPP scheme had a ten-dency to systematically overestimate winter mixing depthwhile estimating summer UML depths that were too shallowcompared to observations. However, modifications to KPPparameters were sufficient to improve the representation ofthe UML for the realistic behavior of biological tracers.[24] The ecosystem dynamics represented in participating

models varies substantially, ranging from UL’s simple,four component nutrient-phytoplankton-zooplankton-detritus(NPZD) model, to the complex, 24 component planktonfunctional type (PFT) model employed in LANL. However,in spite of numerous apparent differences, the hierarchy ofthe ecosystem models employed and the range of theirapplicability is clear.[25] The simplest model in the intercomparison is that used

in the UL model, a four state variable ecosystem that is basedon the nitrogen cycle only, and which includes single phy-toplankton, zooplankton and detritus components.[26] OCCAM uses a very similar NPZD structure based

around the nitrogen cycle (but with different governingequations). These allow OCCAM to directly simulate carboncycle dynamics, and permit additional aspects such as non-Redfield stoichiometry.[27] The UW model separates 11 state variables and takes

a significant step toward increased ecosystem complexityby adding silicate as a second nutrient and splitting phyto-plankton into two functional types. The resulting modelidentifies diatoms (limited by nitrogen and silicon) and


5 of 16

dinoflagellates (limited by nitrogen only), and splits zoo-plankton into three state variables to achieve realistic grazingof the two phytoplankton types. Such an increase in com-plexity allows the reproduction of difference between theecosystem regimes of the Pacific and Atlantic inflows; theformer is silicic acid replete and diatom dominated, while thelatter is more silicic acid limited [e.g., Popova et al., 2010].Although modeling this difference in plankton communitycomposition may not be of primary importance in realisti-cally capturing the large-scale features of total annual pri-mary production, it may be essential for accuratelyrepresenting seasonal ecosystem dynamics, the export ofparticulate organic matter out of the photic zone and the roleof higher trophic levels. Additionally, the UW modelincludes dissolved organic matter (DOM). Although the roleof DOM in controlling spatial and temporal features of thepresent-day AO primary production is not yet clear [e.g.,Alling et al., 2010], its importance in the AO carbon cycle isexpected to increase in the future as permafrost melting andelevated riverine discharge increase its supply to the AO[e.g., Frey and McClelland, 2009]. As such, it may beadvantageous for future AO models to consider DOM as astate variable at the early stages of their development.[28] Similarly to the UW model, NEMO (11 state vari-

ables) includes the silicon cycle and splits both phyto-plankton and zooplankton variables into separate, size-basedcommunities. However, unlike the UW, it also considerschlorophyll-a (Chl-a) as a state variable independent ofphytoplankton biomass, though it does not include DOM, inlarge part because it was developed for use at global ratherthan AO scale. Addition of the Chl-a as independent statevariables is essential to describe formation of the subsurfaceChl-a maximum and its substantial contribution to AO pri-mary production. NEMO also includes a simplified repre-sentation of the iron cycle, though this has little significancefor the AO and was introduced because of its importance inthe other regions of the global ocean, such as the SouthernOcean.[29] Finally, the most complex ecosystem model is that

employed by LANL (24 state variables) [Moore et al., 2004].In addition to the nitrogen, silicon and iron cycles included inNEMO, this introduces the phosphorus cycle; it includesthree types of phytoplankton with independent Chl-a statevariables; and it includes DOM. Unlike NEMO, whichdistinguishes fast-growing microzooplankton from slower-growing mesozooplankton, this model has only a single typeof zooplankton. Though it was added for its role at the globalscale, a potential advantage of including phosphorus limita-tion in the AO is that low-salinity (<26 PSU) shelf waters fedby Arctic rivers are relatively rich in dissolved inorganicnitrogen and silicic acid but poor in phosphate [see Sakshaug,2004, and references therein]. However, at present the LANLmodel does not include the riverine input of nutrients, so thepresence of the phosphorus cycle is not likely to introduceecosystem dynamics that are absent in the other models.

3. Results

[30] We focus our analysis here on the comparison of thetotal annual, vertically integrated primary production andthe physical factors that regulate it in the AO. We have spe-cifically chosen to examine year 1998 since this falls in the

overlap between the runs of participating model (see Table 1),it is sufficiently forward in time from run initialization to beat least partially equilibrated and avoid strong drift, and itdoes not experience the (unrepresentative) low ice extent of2006 and 2007.

3.1. Total Primary Production

[31] Figures 1a–1e show total annual, vertically integratedprimary production for year 1998 for all five models, and thatestimated using satellite-derived observations (Figure 1f)[Pabi et al., 2008]. Note that model-observation comparisonshould be undertaken with care here for several reasons.First, satellite-derived primary production is based on anempirical model which includes a number of assumptions(e.g., the vertical structure of primary production) and whichhas been subject to only limited verification for the Arctic.Second, problems associated with quantifying chlorophyll-a(Chl-a) from remotely sensed ocean color are particularlyacute in the Arctic relative to other parts of the World Ocean.These problems include restriction of observations to waterswith sea ice concentrations of less than 10%; signal con-tamination by sea ice itself; the frequent occurrence of Arcticfog in areas coincident with maximum Chl-a concentrations[e.g., Perrette et al., 2010]; and signal contamination bycolored dissolved organic matter (CDOM) in areas affectedby riverine input. In addition, the vertical distribution ofChl-a in the AO is characterized by a strong subsurfacemaximum that occurs below the upper mixed layer (UML)and which cannot be detected remotely [e.g., Hill and Cota,2005]. Pabi et al. [2008] assume an exponential decline ofChl-a below the UML which may have introduced substan-tial errors in the depth-integrated estimates. For comparison,the synthesis of in situ data [cf. Carmack et al., 2006,Figure 13] suggests that satellite-derived values of pro-ductivity at the Pacific and Atlantic inflows are generallyunderestimated, while values along the Siberian andCanadian coastline are overestimated. Nevertheless, sincesatellite-derived estimates of primary production remain theonly source of high-resolution spatiotemporal information,we retain the Pabi et al. [2008] data here.[32] As can be seen from Figure 1, all models are in general

agreement on the main features of the spatial distribution ofprimary production across the AO. The highest observed andmodeled values occur in the Norwegian Sea and on theinflow shelves to the Arctic (Chukchi and Barents seas)[Carmack et al., 2006]. These regions are where the influ-ence of the high-nutrient environments of the North Pacificand North Atlantic is the greatest. The central Arctic, with itspermanent ice cover, manifests the lowest values, whileproductivity of the seasonally ice-free shelves shows inter-mediate values in good agreement with the in situ data[Carmack et al., 2006]. However, in spite of general agree-ment on the large-scale distribution patterns, there are sub-stantial regional variations between the models. Thus, theChukchi Sea, which has the highest observed productivity inthe Arctic Ocean (reaching 400 g C m�2 yr�1) [Sakshaug,2004; Carmack et al., 2006] is only moderately productive(less than 100 g C m�2 yr�1) in the OCCAM and UL whilethe rest of the models appear to reproduce this “hot spot” ofproductivity relatively well. Similarly, OCCAM substan-tially underestimates the productivity of the Norwegian andBarents Seas with values of 70–100 g C m�2 yr�1 compared


6 of 16

with observed values that reach 150–200 g C m�2 yr�1.Given these patterns of observed and modeled primary pro-duction, a number of critical questions occur. What drivessimilarities and differences between the model estimates ofArctic production? Primary production in the AO is deter-mined by a complex interplay between light and nutrientavailability and stratification. Are similar features betweenmodels (and observations) driven by similar underlyingphysical processes? In sections 3.2–3.4 we investigate themain physical drivers that influence the magnitude and dis-tribution of ocean primary production in the Arctic.

3.2. Short-Wave Radiation and Ice Concentration

[33] The AO manifests extreme seasonality in light regimefrom permanent darkness during winter to continuous sun-light in summer, though even then solar elevation remainslow. Light availability is also strongly influenced by thepresence of sea ice, which (especially with snow cover)reduces the irradiance reach the ocean to a fraction of itsvalue immediately above the ice. However, sea ice concen-tration is often less than 100% during the summer months,with regions experiencing coverage of 80–90% even in areas

of multiannual ice [Sakshaug, 2004]. Thus, in consideringimpact of light on large-scale distribution of primaryproduction, two characteristics of prime importance are:downwelling short-wave radiation at the surface of the oceanand sea ice concentration.[34] The modeled and observed (HadISST) [Rayner et al.,

2003] monthly mean sea ice concentrations for September(the month of the minimal ice extent) are shown in Figure 2.Although the correlation between observed and modeled datais high (correlation coefficient are �0.7 for UW, LANL andNEMO and �0.6 for UL and OCCAM) there are substantialdisagreements in respect to features of spatial variability. Forinstance, four models (NEMO, LANL, UW and UL) sys-tematically underestimate sea ice concentration in the centralAO by 20–30%, and by 50% in certain areas (e.g., NEMOand UL show pronounces minimum in the Laptev sector withconcentrations below 50% compared to observed values of90%). This underestimation of sea ice concentration leads toan equivalent overestimation of short-wave radiation in theseregions and, therefore, also the availability of photosynthet-ically active radiation (PAR) for phytoplankton. The finalmodel, OCCAM, does not follow this pattern and generally

Figure 1. Mean annual water column primary production (in g C m�2 yr�1) for (a) NEMO, (b) LANL,(c) UW, (d) UL, (e) OCCAM, and (f) satellite-derived estimates of Pabi et al. [2008].


7 of 16

overestimates sea ice extent and manifests much higherconcentration, extending permanent cover onto the shelves ofthe Beaufort and Laptev Seas, and into Baffin Bay contrary toobservations. Because of a number of significant differencesin both the ice models and surface forcing fields, it is difficultto isolate the reason (or reasons) for the reduced CentralArctic sea ice cover simulated by the four models. At thesame time an alternative algorithm [e.g., Cavalieri et al.,1996] of sea ice estimation gives much lower ice concentra-tions in summer (Figure 3) possibly due to retrievals of sur-face melt ponds looking like open water. The differencesbetween the two algorithms are substantial and their assess-ment is outside of the scope of this paper. We can onlycomment that the bias may not be significantly detrimentalfor the simulation of physics in the upper and intermediateocean, but may cause the light regime to be unrealisticallyfavorable for primary production and that it requires furtherinvestigation.

3.3. Depth of the Upper Mixed Layer

[35] Although the impact of sea ice on productivity is mostimmediately evident through its control of the penetration of

Figure 2. Monthly mean ice concentration in September for (a) NEMO, (b) LANL, (c) UW, (d) UL,(e) OCCAM, and (f) observations (HadISST) [see Rayner et al., 2003].

Figure 3. Monthly mean ice concentration for Septemberfrom Nimbus-7 SMMR and DMSP SSM/I passive micro-wave data [Cavalieri et al., 1996].


8 of 16

solar radiation into the ocean, ice influence is not restricted tothe ambient light regime. Sea ice also affects vertical strati-fication via salt rejection in winter and fresh water input insummer and by presenting a barrier to wind-driven mixing ofthe water column. In addition by contributing fresh waterthrough melting during spring and summer time it acts tostrengthen water column stratification which inhibits nutrientresupply from below and provides an additional constraint onprimary production [Carmack et al., 2006].[36] Winter mixing is one of the two main mechanisms

that supply nutrients into the photic zone of the AO andwhich, in doing so, set up a limit available for the con-sumption by phytoplankton. The second mechanism is hor-izontal advection of more nutrient-rich seawater, mostprominently via the Pacific and Atlantic inflows to the AO.In addition, there is a number of secondary mechanismsthat can influence nutrient supply and primary production ona local scale and episodically. These include severe stormsand internal waves that erode the halocline, enhanced tidalmixing in the areas of rough topography, wind-driven shelfbreak and ice edge upwellings and the turbulent wake

behind banks and cyclonic eddies. Since the horizontal res-olution of the models employed in this study is insufficientto permit localized nutrient supply mechanisms, some small-scale or episodic “hot spots” of productivity are beyond ouranalysis and we have instead focused on winter mixing asthe main mechanism controlling basin-scale patterns ofnutrient supply.[37] The maximum depth of the UML during the year

(based on monthly mean values) from the models and WOAclimatology [Locarnini et al., 2006; Antonov et al., 2006](using variable density criterion) is shown in Figure 4.Owing to the very stable stratification of the AO, deep wintermixing (in excess of 300 m) occurs only in the Atlanticinflow waters in the southeast Greenland and southwestBarents sectors (see Figure 4f). Two additional features ofimportance to ecosystem productivity are apparent fromobservational mixed layer depth: (1) winter mixing rarelyexceeds 80 m outside of the Atlantic inflow, and on averageis only 40 m and (2) mixing does not penetrate deeper than20 m throughout the year on the Siberian shelves which areaffected by significant riverine inputs of fresh water.

Figure 4. Maximum depth of UML during the year on the basis of monthly averaged values (m; note non-linear color scale) for (a) NEMO, (b) LANL, (c) UW, (d) UL, (e) OCCAM, and (f) WOA climatology.


9 of 16

[38] All participating models reproduce the occurrence ofthe deep mixing in the Atlantic inflow, although its magni-tude and spatial extent vary. It is most intense (deeper than1000 m) in the LANL and UW models, and shallower in theNEMO, UL and OCCAM models (500–700 m). As theDIN profile is relatively homogeneous in these areas below700 m, the difference in the depth of winter convectionshould not have a significant effect on the winter nutrientsupply in the models. However, difference in mixing in thecentral AO between the models is of substantial importance.Thus, NEMO and OCCAM both show good agreement withobservations in confining deep mixing to Atlantic inflowwaters, while LANL and UL both show overestimate mixingin the central AO with values up to 300 m over significantareas. UL also shows overestimated values although to alesser degree than LANL and UW, generally reaching 90–110 m over the central AO. All models find mixing overthe Siberian shelves to be more shallow than in other areasreflecting the influence of the freshwater input by Siberianrivers. However, this effect appears to be less pronounced inthe LANL, OCCAM and UW models, where mixing in the

East Siberian Sea penetrates down to the bottom (50–70 m)over large areas of the shelf.

3.4. Inorganic Nutrients

[39] Nitrogen appears to be the primary limiting nutrient inthe Arctic as it is the first to become exhausted in bloomevents [see Tremblay and Gagnon, 2009, and referencestherein]. All five participating models use nitrogen as theirbase “currency” although NEMO, LANL and UW alsoinclude silicic acid in order to distinguish between diatomand nondiatom phytoplankton. In addition UW and LANLinclude ammonium as a state variable, however, as ammo-nium concentrations are usually small, we consider nitrate asa representative of dissolved inorganic nutrient (DIN) in allmodels. All models are similar in the way they initialize DIN(WOA05) [Garcia et al., 2006]; however, they vary inboundary conditions. Two regional models (UW and UL)specify monthly WOA05 on the open boundaries while threeglobal models crucially depend on the quality of their solu-tion in upstream areas (northern North Pacific and Atlantic)for correct representation of the nutrient supply through theBering Strait and Atlantic inflow. Additionally, while the

Figure 5. Maximum DIN concentrations (in mmol N m�3) on the basis of monthly averaged values for(a) NEMO, (b) LANL, (c) UW, (d) UL, (e) OCCAM, and (f) WOA climatology.


10 of 16

rest of the models ignore riverine inputs of nutrients,NEMO includes a simplified representation in which theconcentrations of N and Si are relaxed toward the WOAclimatology within 100 km of shoreline. This approach aimsto take account of poorly known concentrations of nutrientsin rivers and their transformation in unresolved near-coastalareas.[40] Here we focus our analysis of modeled nutrient

dynamics on two metrics: the maximum and minimumsurface nutrient concentrations during the year (based onmonthly mean model output). The maximum value usuallyoccurs at the end of the winter season and is strongly drivenby physical processes such as winter mixing and horizontaladvection of seawater with high nutrient concentrations fromthe northern North Pacific and Atlantic basins. The minimumvalue of DIN typically occurs in summer months and reflectsthe local conditions for primary production. Where phyto-plankton have a relatively long growing season because ofearly ice melt, nutrient concentrations generally fall belowtheir half-saturation constants for uptake by phytoplankton,

and nutrient limitation regulates production. Where mini-mum nutrient concentrations are still relatively high, thisis indicative of light availability setting an upper limit onannual primary production [e.g., Popova et al., 2010]. Max-imum DIN concentrations for all models and from the WOAclimatology are shown in Figure 5. The WOA climatologysuffers from substantial undersampling in the AO (e.g.,characteristic pole-oriented pattern typical of extrapolationinto unsampled areas). Figure 6, showing total number ofobservations used in WOA climatology for surface DIN(Figure 6a) in comparison to temperature (Figure 6b), clearlydemonstrates magnitude of the undersampling. Neverthelessthree main features are apparent in the spatial distribution:(1) nutrient concentrations are greatest in the areas of Pacificand Atlantic inflow; (2) typical values in the central AO arelow and �1 mmol N m�3; and (3) Siberian shelves show thelowest values over the AO basin which are comparable orbelow the typical half-saturation values for DIN uptake (0.5–1 mmol N m�3). The model which is probably the closestto the observed pattern is NEMO (Figure 5a) although itunderestimates the already low nutrient concentrations in thecentral basin by about a factor of 2. In addition it alsounderestimates nutrient concentration at the Pacific andAtlantic inflows by 2–3 mmol N m�3 (10–20%). Underesti-mation of nutrients in the central Arctic results in NEMOfrom excessive productivity under sea ice, the concentrationsof which are underestimated by the model (see section 3.2and Figure 1). We can speculate that improvement of iceconcentration in NEMO will improve the agreement withobserved nutrients. Slight underestimation of the DIN at thePacific and Atlantic inflow is driven by model performancein upstream areas and will require further verification whichis beyond the scope of this paper.[41] UW and LANL both show excellent agreement

with the observed DIN at the Pacific and Atlantic inflows.However, in very similar ways, both also suffer from twoproblems. The first is overestimation of nutrients in thecentral AO (with values in excess of 10 mmol N m�3),which generally follows the corresponding pattern of over-estimated winter mixing (cf. Figure 4). The second problemis extremely high nutrient concentrations (in excess of15 mmol N m�3) on the Siberian shelves, and in particular inthe East Siberian and Laptev seas, where observed values ofDIN appear to be very low. This overestimation appears toarise from excessive mixing on the Siberian shelves. Giventheir shallow depth (e.g., average depth of the East SiberiaSea is only 45 m), a few mixing events during the year whichpenetrate to the bottom layer of the model are sufficient toprovide entrainment of high deep nutrient concentrations.This is contrary to observations that show year-round stablelow-salinity layer on the top of the water column whichefficiently prevents entrainment of the high near-bottomnutrients [Nitishinsky et al., 2007].[42] The UL model shows elevated DIN concentrations

at the Atlantic inflow although its pattern (which closelyfollows that of winter mixing; cf. Figure 4) does not agreewell with the observations (Figure 5f), and is generally con-fined to the shelf break of the Norwegian and GreenlandSeas. The model also underestimates DIN concentrationnorth of the Bering Strait where model boundary conditionsrequire further verification. Similarly to UW and LANLmodels, UL also overestimates mixing in the central AO

Figure 6. Total number of observations included intothe annual WOA climatology per one degree grid area for(a) DIN and (b) temperature.


11 of 16

leading to concomitant overestimation of nutrients, althoughthis problem is not as pronounced as in the two formermodels.[43] OCCAM’s distribution of DIN (Figure 5e) does not

follow the general tendency of DIN in four other models tomimic the dynamics of the maximum winter mixing. Themodel shows overestimated DIN concentrations in excess of15 mmol N m�3 over the majority of the AO with maximumvalues at the Pacific inflow despite its realistic spatial distri-bution of the UML depth. An explanation for this pattern liesin part in OCCAM’s overestimation of sea ice extent (seeFigure 2e and section 3.2). As perennial sea ice cover inthis model extends too far south onto the Canadian andSiberian shelves and leaves only small portion of the south-ern Chukchi Sea briefly open to the sunlight, and essentiallysuppresses productivity in all areas other than in the Atlanticinflow region. Thus nutrient-rich, low-salinity water pene-trating from the Bering Strait and residing in the upper watercolumn [Carmack et al., 2006] retains its high nutrient con-centrations contrary to the observations which show quicknutrient utilization downstream owing to the highly produc-tive waters of the Chukchi and Beaufort sectors. Analysis of

OCCAM equilibration dynamics during the spin-up phase(not shown) finds that it takes 10 to 20 years for the nutrient-rich waters to penetrate into the AO through the Beaufortgyre up to the Lomonosov ridge and establish a persistent andhigh nutrient environment over the majority of the AO.[44] Minimum nutrient concentrations in the models

and observations (WOA2005) are shown in Figure 7. Here,again, we must take care in interpreting details of the spatialdistribution from climatological data in the heavily under-sampled AO (Figure 6). Nevertheless the climatologyunambiguously points toward the fact that, over the majorityof the AO, DIN is drawn down to concentrations belowits half-saturation level at some stage during the summer,with the exception of areas immediately north of the BeringStrait and Atlantic inflow. As seen previously, NEMO isthe only model which shows nutrient exhaustion over themajority of the AO. Nevertheless, it underestimates residualnutrients at the Atlantic inflow implying that total AO regimeis more oligotrophic than is actually observed. The rest of themodels overestimate unutilized nutrients to a various degree(especially OCCAM), although LANL, UW and UL model

Figure 7. Minimum DIN concentration (in mmol N m�3) on the basis of monthly averaged values for(a) NEMO, (b) LANL, (c) UW, (d) UL, (e) OCCAM, and (f) WOA climatology.


12 of 16

agree on summer oligotrophic conditions on the Beaufortshelves.[45] Model skill in reproducing annual minimum nutrient

concentrations in the AO is probably the most crucial testto demonstrate if the realistic distribution of the primaryproduction is achieved for the right reasons and drivenby the right physical mechanisms. In the AO, where bothnutrient and light limitation of primary production is tightlycontrolled by the distribution of ice, it is easy to fall into atrap of correctly modeling only one of two factors with asatisfactory consequence for the distribution of the totalproductivity. Thus, in our intercomparison experiment:OCCAM finds that the present-day (1998) AO is totallylimited by light; LANL and UL generally agree on about 2/3of the total area being limited by light (minimum DIN con-centration above half-saturation level) and 1/3 by the nutri-ents; UW points toward 1/3 of the total OA area being limitedby light; and NEMO finds nutrient limitation of the entireocean except a small area immediately adjacent to the BeringStrait.

4. Potential Role of the Ice Algae

[46] One immediate caveat to our analysis is that it hasfocused on total annual AO primary production, which isdominated by water column primary production in theseasonally ice-free AO where productivity is highest. Thisapproach ignores the role played by attached ice algae. Thesehave the ability to grow at low-light levels [Kirst andWiencke, 1995] and provide food in autumn and spring,when phytoplankton in the water column are scarce [Bluhmand Gradinger, 2008; Schnack-Schiel, 2003]. Further, bothobservations and model results indicate that upon ice melt,release of substantial amounts of ice algae biomass cancontribute to phytoplankton seeding and nutrition of zoo-plankton and benthos [see Gradinger, 2009, and referencestherein; Jin et al., 2007, 2009]. On the basis of estimates fromobservations, more than 50% of primary production in thecentral AO covered by perennial ice where water columnproduction is low is attributed to ice algae [Gosselin et al.,1997], while on Arctic shelf seas, the percentage of marineprimary production by ice algae is estimated to be between 4and 25%, depending upon location [Legendre et al., 1992].Only the LANL and ULmodels in this intercomparison studyincludes ice algae and ice algal production. Pan-Arctic resultsfor simulated primary production within sea ice from thesemodels can be found in the work of Jin et al. [2011] andDupont (submitted manuscript, 2011). Although importanceof ice algae as an early and concentrated food source forhigher trophic levels is clear, assessment of the necessity oftheir inclusion in large-scale models aimed at lower trophiclevels is outside of the scope of this paper.

5. Discussion and Conclusions

[47] In this study we have compared Arctic Ocean (AO)primary production, and the physical factors controlling it,in five general circulation models with coupled biogeo-chemistry. All participating model reproduce broadly similarpatterns of total annual AO primary production with thehighest values at the Atlantic and Pacific inflows and lowvalues in the central Arctic, in agreement with both satellite-

derived and in situ data. Nevertheless, the physical factorscontrolling this distribution in the five models are different.[48] Production in the Arctic Ocean is known to be

colimited by availability of light and nutrients. Our analysisfinds that sea ice extent and concentration are broadly similaramong the models, and provide comparable levels of lightlimitation. However, substantial differences in the depthof winter mixing between the participating models providesignificantly different amounts of nutrients available for thephytoplankton to utilize during the spring and summer pro-ductive season. The nutrient climatology from the WorldOcean Atlas shows that nitrate concentrations fall below thehalf-saturation level over at least 70% of the area of the AO.[49] The annual minimum nutrient concentration varies

substantially between the participating models from near-complete utilization across the AO in some models to year-round nutrient replete conditions in others. A root cause ofthe unrealistically high nutrient concentrations found in someof the models is excessive vertical mixing, and this shifts thebalance between the relative importance of nutrient and lightlimitation in the AO. These differences between modelsmight not be detrimental in determining present-day AOprimary production since both light and nutrient limitationare tightly coupled to the presence of ice. Essentially, as longas at least one of the two limiting factors is reproduced cor-rectly, simulated total primary production will be close to thatobserved. However, if the retreat of the Arctic sea ice con-tinues into the future as expected, a decoupling between seaice and nutrient limitation will occur and the future predictivecapabilities of the models will potentially diminish.[50] Consequently, in finding a high sensitivity of the

Arctic Ocean primary production to vertical mixing, andechoing earlier studies for other regions of the World Ocean,our study emphasizes the importance of its realistic repre-sentation in ocean GCMs.

6. Recommendation for Future AO ModelingStudies

[51] 1. Our study represents only a first step toward AOecosystem model intercomparison and as such, it has onlyfocused on total water column primary production. However,it is “export” rather than total primary production that con-trols removal of the organic carbon into the deep ocean andout of the contact with the atmosphere [e.g., Eppley andPeterson, 1979]. Thus, ecosystem model intercomparisonexperiments should in future address a split between new andregenerated fractions of primary production and their poten-tial future changes as a way to assess model uncertainty inpredicting AO impact on the atmospheric CO2.[52] 2. This study focuses on the magnitude of AO primary

production in the pelagic plankton ecosystem and largelyignores the role of attached sea ice algae. These occur on theunderside of sea ice, and provide a concentrated food sourcefor metazoan grazers that in turn feed fish, seabirds andmarine mammals [Bluhm and Gradinger, 2008; Schnack-Schiel, 2003]. Because sea ice algae are adapted for growthat the low-light levels that occur under sea ice [Kirst andWiencke, 1995], they bloom earlier in the year than phyto-plankton, and serve as a food source for both pelagic andbenthic communities [Gradinger, 2009]. Furthermore, icealgae are released into the water column upon sea ice melt


13 of 16

and those that are not immediately consumed by pelagicgrazers, sink to the seafloor where they provide a source offood for benthic organisms. However, in this study only theLANL and UL models explicitly represent sea ice algae.Thus, more complete considerations of higher trophic levelsand biogeochemical cycling in the AO ecosystem will needto include parameterizations or explicit representation of seaice algae.[53] 3. Sea ice plays an overwhelmingly important role in

the AO. Its impact on productivity is perhaps most evidentthrough its direct control of solar irradiance and thus the lightavailable for photosynthesis [e.g., Sakshaug, 2004]. How-ever, the indirect impact of sea ice on stratification, and thusnutrient supply, is likely to play a more significant role inthe future, as the AO changes under continuous ice retreat[e.g., Carmack et al., 2006; Carmack and Wassmann, 2006;Tremblay and Gagnon, 2009]. Thus, we caution against theassumption that more ice-free Arctic may be anticipated toexhibit increased productivity on grounds of elevated lightavailability. It may actually respond quite differently ongrounds of nutrient availability.[54] 4. Circulation, mixing and freshwater balance play a

pivotal role in maintaining the present-day distribution of thenutrients in Arctic surface waters on decadal timescales, andare probably more important in controlling plankton pro-ductivity than in any other ecological domain of the worldocean [e.g., Carmack and McLaughlin, 2011; Tremblay andGagnon, 2009; Popova et al., 2010]. Thus, the fate of theArctic’s ecosystems is tightly coupled to the fate of its uniquehalocline. Although understanding and modeling of theAO circulation and water mass transformation made asignificant progress over the last few decades, expansion ofthis knowledge into the biogeochemical cycling is still at itsinfancy. Thus, more studies linking ocean dynamics withbiogeochemical tracers are needed in order to better under-stand the present and predict the future changes of AOproductivity.[55] 5. Our study shows that there is a substantial dis-

agreement between the models themselves and models andobservations in respect to a fundamental question: whichfactor, light or nutrients, controls present-day Arctic pro-ductivity. Our inability to answer this question with certaintyreduces our capability for future Arctic predictions. Thus,care should be taken in carrying out forecasting of the AOecosystem dynamics in its future transition to the seasonallyice free ocean until there is a sufficient confidence in modelsability to predict the present-day state of the AO ecosystems.[56] 6. There is a growing realization of the limitations

of our knowledge of AO ecosystems owing to severity oftheir undersampling, exacerbated by the transient regimeexperienced by the whole ocean-sea-ice-ecosystem [e.g.,Wassmann, 2011]. However, although our knowledge andobservational information may not be sufficient for under-standing biophysical interactions in all of their complexity,there is enough accumulated information to verify our mod-els in respect to their fundamental properties (e.g., produc-tivity). As we have shown in the present study, there is asubstantial scope for reducing model uncertainty in predic-tion of reasonably well observed and understood propertiesof the AO. Essentially, incomplete observational informationshould not serve as a reason to abandon the developmentof coupled physical and biological models of the AO.

[57] Acknowledgments. We acknowledge the great role of theAOMIP program in providing a stimulating forum and logistical supportfor this study. The contribution of the Polar Science Center, University ofWashington, was supported by the NSF’s Office of Polar Programs. Thecontribution of the National Oceanography Centre was supported by NaturalEnvironment Research Council. The contribution of the International ArcticResearch Center, University of Alaska, was supported by the DOE EPSCoRProgram, grant DE-FG02-08ER46502.

ReferencesAksenov, Y., S. Bacon, A. C. Coward, and A. J. G. Nurser (2010), TheNorth Atlantic inflow to the Arctic Ocean: High-resolution model study,J. Mar. Syst., 79, 1–22, doi:10.1016/j.jmarsys.2009.05.003.

Alling, V., et al. (2010), Nonconservative behavior of dissolved organiccarbon across the Laptev and East Siberian seas, Global Biogeochem.Cycles, 24, GB4033, doi:10.1029/2010GB003834.

Antonov, J. I., R. A. Locarnini, T. P. Boyer, A. V. Mishonov, and H. E.Garcia (2006), World Ocean Atlas 2005, vol. 2, Salinity, NOAA AtlasNESDIS 62, edited by S. Levitus, 182 pp., U.S. Gov. Print. Off.,Washington, D. C.

Arrigo, K. R., G. van Dijken, and S. Pabi (2008), Impact of a shrinkingArctic ice cover on marine primary production, Geophys. Res. Lett., 35,L19603, doi:10.1029/2008GL035028.

Bitz, C. M., and W. H. Lipscomb (1999), An energy-conserving thermo-dynamic sea ice model for climate study, J. Geophys. Res., 104,15,669–15,677, doi:10.1029/1999JC900100.

Bluhm, B., and R. Gradinger (2008), Regional variability in food availabil-ity for arctic marine mammals, Ecol. Appl., 18, S77–S96, doi:10.1890/06-0562.1.

Carmack, E., and F. McLaughlin (2011), Towards recognition of physicaland geochemical change in Subarctic and Arctic Seas, Prog. Oceanogr.,90, 90–104, doi:10.1016/j.pocean.2011.02.007.

Carmack, E., and P. Wassmann (2006), Food webs and physical-biologicalcoupling on pan-Arctic shelves: Unifying concepts and comprehensiveperspectives, Prog. Oceanogr., 71, 446–477, doi:10.1016/j.pocean.2006.10.004.

Carmack, E., D. Barber, J. H. Christensen, R. Macdonald, B. Rudels, andE. Sakshaug (2006), Climate variability and physical forcing of the foodwebs and the carbon budget on pan-Arctic shelves, Prog. Oceanogr., 71,145–181, doi:10.1016/j.pocean.2006.10.005.

Cavalieri, D., C. Parkinson, P. Gloersen, and H. J. Zwally (1996), Sea iceconcentrations from Nimbus-7 SMMR and DMSP SSM/I passive micro-wave data, http://nsidc.org/data/nsidc-0051.html, Natl. Snow and IceData Cent., Boulder, Colo. [Updated 2008.]

Deal, C., M. Jin, S. Elliott, E. Hunke, M. Maltrud, and N. Jeffery (2011),Large-scale modeling of primary production and ice algal biomass withinarctic sea ice in 1992, J. Geophys. Res., 116, C07004, doi:10.1029/2010JC006409.

DRAKKAR Group (2007), Eddy-permitting ocean circulation hindcasts ofpast decades, CLIVAR Exch., 12(3), 8–10.

Dukowicz, J. K., and R. D. Smith (1994), Implicit free-surface method forthe Bryan-Cox-Semtner ocean model, J. Geophys. Res., 99, 7991–8014,doi:10.1029/93JC03455.

Ellingsen, I. H., P. Dalpadado, D. Slagstad, and H. Loeng (2008), Impact ofclimatic change on the biological production in the Barents Sea, Clim.Change, 87, 155–175, doi:10.1007/s10584-007-9369-6.

Eppley, R. W., and B. J. Peterson (1979), Particulate organic matterflux and planktonic new production in the deep ocean, Nature, 282,677–680, doi:10.1038/282677a0.

Frey, K. E., and J. W. McClelland (2009), Impact of permafrost degrada-tion on arctic river biogeochemistry, Hydrol. Process., 23, 169–182,doi:10.1002/hyp.7196.

Garcia, H. E., R. A. Locarnini, T. P. Boyer, and J. I. Antonov (2006),WorldOcean Atlas 2005, vol. 4, Nutrients (Phosphate, Nitrate, Silicate), NOAAAtlas NESDIS 64, edited by S. Levitus, 396 pp., U.S. Gov. Print. Off.,Washington, D. C.

Gaspar, P., Y. Grégoris, and J.-M. Lefevre (1990), A simple eddy kineticenergy model for simulations of the oceanic vertical mixing tests atstation papa and long-term upper ocean study site, J. Geophys. Res., 95,16,179–16,193, doi:10.1029/JC095iC09p16179.

Gosselin, M., M. Levasseur, P. A. Wheeler, R. A. Horner, and B. C. Booth(1997), New measurements of phytoplankton and ice algal production inthe Arctic Ocean, Deep Sea Res., Part II, 44, 1623–1644, doi:10.1016/S0967-0645(97)00054-4.

Gradinger, R. (2009), Sea-ice algae: Major contributors to primary produc-tion and algal biomass in the Chukchi and Beaufort Seas during May/June 2002, Deep Sea Res., Part II, 56, 1201–1212, doi:10.1016/j.dsr2.2008.10.016.


14 of 16

Hakkinen, S., F. Dupont, F. Kauker, M. Karcher, and J. Zhang (2007),Model simulation of Greenland Sea upper ocean variability, J. Geophys.Res., 112, C06S90, doi:10.1029/2006JC003687.

Hibler, W. D. (1979), A dynamic thermodynamic sea ice model, J. Phys.Oceanogr., 9, 815–846, doi:10.1175/1520-0485(1979)009<0815:ADTSIM>2.0.CO;2.

Hibler, W. D. (1980), Modeling a variable thickness sea ice cover,Mon. Weather Rev., 108, 1943–1973, doi:10.1175/1520-0493(1980)108<1943:MAVTSI>2.0.CO;2.

Hibler, W. D., and K. Bryan (1987), A diagnostic ice ocean model, J. Phys.Oceanogr., 17, 987–1015, doi:10.1175/1520-0485(1987)017<0987:ADIM>2.0.CO;2.

Hill, V., and G. Cota (2005), Spatial patterns of primary production on theshelf, slope 25 and basin of the Western Arctic in 2002, Deep Sea Res.,Part II, 52, 3344–3354, doi:10.1016/j.dsr2.2005.10.001.

Holloway, G. (1987), Systematic forcing of large-scale geophysical flowsby eddy-topography interaction, J. Fluid Mech., 184, 463–476,doi:10.1017/S0022112087002970.

Holloway, G., and A. Proshutinsky (2007), The role of tides in Arcticocean/ice climate, J. Geophys. Res., 112, C04S06, doi:10.1029/2006JC003643.

Holloway, G., and T. Sou (2002), Has Arctic sea ice rapidly thinned?,J. Clim., 15, 1691–1701, doi:10.1175/1520-0442(2002)015<1691:HASIRT>2.0.CO;2.

Holloway, G., et al. (2007), Water properties and circulation in Arctic Oceanmodels, J. Geophys. Res., 112, C04S03, doi:10.1029/2006JC003642.

Hunke, E. C. (2001), Viscous plastic dynamics with the EVP model:Linearization issues, J. Comput. Phys., 170, 18–38, doi:10.1006/jcph.2001.6710.

Hunke, E. C., and C. M. Bitz (2009), Age characteristics in a multidecadalArctic sea ice simulation, J. Geophys. Res., 114, C08013, doi:10.1029/2008JC005186.

Hunke, E. C., and J. K. Dukowicz (1997), An elastic-viscous-plastic modelfor sea ice dynamics, J. Phys. Oceanogr., 27, 1849–1867, doi:10.1175/1520-0485(1997)027<1849:AEVPMF>2.0.CO;2.

Hunke, E. C., and J. K. Dukowicz (2002), The elastic-viscous-plastic seaice dynamics model in general orthogonal curvilinear coordinates on asphere—Effect of metric terms, Mon. Weather Rev., 130, 1848–1865,doi:10.1175/1520-0493(2002)130<1848:TEVPSI>2.0.CO;2.

Hunke, E. C., and M. M. Holland (2007), Global atmospheric forcing datafor Arctic ice-ocean modeling, J. Geophys. Res., 112, C04S14,doi:10.1029/2006JC003640.

Hunke, E. C., and W. H. Lipscomb (2008), CICE: The Los Alamos Sea IceModel: Documentation and software user’s manual, version 4.0, Tech.Rep. LA-CC-06-012, Los Alamos Natl. Lab., Los Alamos, N. M.

Jin, M., C. J. Deal, J. Wang, K. H. Shin, N. Tanaka, T. E. Whitledge, S. H.Lee, and R. R. Gradinger (2006), Controls of the landfast ice-ocean eco-system offshore Barrow, Alaska, Ann. Glaciol., 44, 63–72, doi:10.3189/172756406781811709.

Jin, M., C. Deal, J. Wang, V. Alexander, R. Gradinger, S. Saitoh, T. Iida,Z. Wan, and P. Stabeno (2007), Ice-associated phytoplankton bloomsin the southeastern Bering Sea, Geophys. Res. Lett., 34, L06612,doi:10.1029/2006GL028849.

Jin, M., C. J. Deal, J. Wang, and C. P. McRoy (2009), Response of lowertrophic level production to long-term climate change in the southeasternBering Sea, J. Geophys. Res., 114, C04010, doi:10.1029/2008JC005105.

Jin, M., C. Deal, S. H. Lee, S. Elliott, E. Hunke, M. Maltrud, and N. Jeffery(2011), Investigation of Arctic sea ice and ocean primary production forthe period 1992 to 2007 using a 3-D global ice-ocean ecosystem model,Deep Sea Res., Part II, doi:10.1016/j.dsr2.2011.06.003, in press.

Kirst, G. O., and C. Wiencke (1995), Ecophysiology of polar algae,J. Phycol., 31, 181–199, doi:10.1111/j.0022-3646.1995.00181.x.

Large, W., and S. Yeager (2004), Diurnal to decadal global forcing forocean and sea-ice models: The data sets and flux climatologies, NCARTech. Note NCAR/TN-460+STR, Natl. Cent. for Atmos. Res., Boulder,Colo.

Large, W., and S. Yeager (2009), The global climatology of an interannu-ally varying air-sea flux data set, Clim. Dyn., 33, 341–364, doi:10.1007/s00382-008-0441-3.

Large, W. G., J. C. McWilliams, and S. C. Doney (1994), Oceanic verticalmixing: A review and a model with a nonlocal boundary layer parameter-ization, Rev. Geophys., 32, 363–403, doi:10.1029/94RG01872.

Large, W. G., G. Danabasoglu, and S. C. Doney (1997), Sensitivity tosurface forcing and boundary layer mixing in a global ocean model:Annual-mean climatology, J. Phys. Oceanogr., 27, 2418–2447,doi:10.1175/1520-0485(1997)027<2418:STSFAB>2.0.CO;2.

Lavoie, D., K. Denman, and C. Michel (2005), Modeling ice algal growthand decline in a seasonally ice-covered region of the Artic (Resolute

Passage, Canadian Archipelago), J. Geophys. Res., 110, C11009,doi:10.1029/2005JC002922.

Legendre, L., S. F. Ackley, G. S. Dieckmann, B. Gullicksen, R. Horner,T. Hoshiai, I. A. Melnikov, W. S. Reeburgh, M. Spindler, and C. W.Sullivan (1992), Ecology of sea ice biota: Part 2, Global Significance,Polar Biol., 12, 429–444.

Lengaigne, M., G. Madec, L. Bopp, C. Menkes, O. Aumont, and P. Cadule(2009), Biophysical feedbacks in the Arctic Ocean using an Earth systemmodel, Geophys. Res. Lett., 36, L21602, doi:10.1029/2009GL040145.

Lima, I. D., D. B. Olson, and S. C. Doney (2002), Intrinsic dynamicsand stability properties of size structured pelagic ecosystem models,J. Plankton Res., 24, 533–556, doi:10.1093/plankt/24.6.533.

Lipscomb, W. H., and E. C. Hunke (2004), Modeling sea ice transportusing incremental remapping, Mon. Weather Rev., 132, 1341–1354,doi:10.1175/1520-0493(2004)132<1341:MSITUI>2.0.CO;2.

Locarnini, R. A., A. V. Mishonov, J. I. Antonov, T. P. Boyer, and H. E.Garcia (2006), World Ocean Atlas 2005, vol. 1, Temperature, NOAAAtlas NESDIS 61, edited by S. Levitus, 182 pp., U.S. Gov. Print. Off.,Washington, D. C.

Madec, G. (2008), NEMO Reference Manual, Ocean Dynamic Component:NEMO-OPA, Tech. Rep. 27, Inst. Pierre Simon Laplace, Paris.

Marsh, R., B. A. de Cuevas, A. C. Coward, H. L. Bryden, and M. Alvarez(2005), Thermohaline circulation at three key sections of the North Atlan-tic over 1985–2002, Geophys. Res. Lett., 32, L10604, doi:10.1029/2004GL022281.

McPhee, M. G. (1984), Drift Velocity During the Drift-Station Phase ofMIZEX 83, MIZEX Bull. Ser., vol. 4, pp. 1–11, U.S. Army Cold Reg.Res. and Eng. Lab., Hanover, N. H.

Moore, J. K., S. C. Doney, and K. Lindsay (2004), Upper ocean ecosystemdynamics and iron cycling in a global three-dimensional model, GlobalBiogeochem. Cycles, 18, GB4028, doi:10.1029/2004GB002220.

Najjar, R. G., et al. (2007), Impact of circulation on export production, dis-solved organic matter, and dissolved oxygen in the ocean: Results fromPhase II of the Ocean Carbon-cycle Model Intercomparison Project(OCMIP-2), Global Biogeochem. Cycles, 21, GB3007, doi:10.1029/2006GB002857.

Niebauer, H. J., and W. O. Smith Jr. (1989), A numerical model of meso-scale physical-biological interactions in the Fram Strait marginal ice zone,J. Geophys. Res., 94, 16,151–16,175, doi:10.1029/JC094iC11p16151.

Nitishinsky, M., L. G. Anderson, and J. A. Hölemann (2007), Inorganiccarbon and nutrient fluxes on the Arctic Shelf, Cont. Shelf Res., 27,1584–1599, doi:10.1016/j.csr.2007.01.019.

Pabi, S., G. L. van Dijken, and K. R. Arrigo (2008), Primary productionin the Arctic Ocean, 1998–2006, J. Geophys. Res., 113, C08005,doi:10.1029/2007JC004578.

Parkinson, C. L., and W. M. Washington (1979), Large-scale numerical-model of sea ice, J. Geophys. Res., 84, 311–337, doi:10.1029/JC084iC01p00311.

Perrette, M., A. Yool, G. D. Quartly, and E. E. Popova (2010), Near-ubiquity of ice-edge blooms in the Arctic, Biogeosciences, 8, 515–524,doi:10.5194/bg-8-515-2011.

Popova, E. E., A. C. Coward, G. A. Nurser, B. de Cuevas, M. J. R. Fasham,and T. R. Anderson (2006), Mechanisms controlling primary and newproduction in a global ecosystem model—Part I: Validation of the biolog-ical simulation, Ocean Sci., 2, 249–266.

Popova, E. E., A. Yool, A. C. Coward, Y. K. Aksenov, S. G. Alderson,B. A. de Cuevas, and T. R. Anderson (2010), Control of primary pro-duction in the Arctic by nutrients and light: Insights from a high resolu-tion ocean general circulation model, Biogeosciences, 7, 3569–3591,doi:10.5194/bg-7-3569-2010.

Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland, L. V. Alexander,D. P. Rowell, E. C. Kent, and A. Kaplan (2003), Global analyses of seasurface temperature, sea ice, and night marine air temperature since thelate nineteenth century, J. Geophys. Res., 108(D14), 4407, doi:10.1029/2002JD002670.

Röske, F. (2001), An atlas of surface fluxes based on the ECMWFreanalysis—A climatological data set to force global ocean general circu-lation models, Tech. Rep. 323, Max-Planck-Inst. fur Meteorol., Hamburg,Germany.

Saba, V., et al. (2010), Challenges of modeling depth-integrated marine pri-mary production over multiple decades: A case study at BATS and HOT,Global Biogeochem. Cycles, 24, GB3020, doi:10.1029/2009GB003655.

Sakshaug, E. (2004), Primary and secondary production in the Arctic Seas,in The Organic Carbon Cycle in the Arctic Ocean, edited by R. Steinand R. Macdonald, pp. 57–81, Springer, Berlin, doi:10.1007/978-3-642-18912-8_3.

Schnack-Schiel, S. B. (2003), The macrobiology of sea ice, in Sea Ice—An Introduction to Its Physics, Chemistry, Biology and Geology, edited


15 of 16

by D. N. Thomas and G. S. Dieckmann, chap. 7, pp. 211–239, BlackwellSci., London.

Semtner, A. J. (1976), A model for the thermodynamic growth of sea icein numerical investigation of climate, J. Phys. Oceanogr., 6, 376–389,doi:10.1175/1520-0485(1976)006<0379:AMFTTG>2.0.CO;2.

Slagstad, D., I. H. Ellingsen, and P. Wassmann (2011), Evaluating primaryand secondary production in an Arctic Ocean void of summer sea ice: Anexperimental simulation approach, Prog. Oceanogr., 90, 117–131,doi:10.1016/j.pocean.2011.02.009.

Smith, R., and P. Gent (2002), Reference manual for the Parallel OceanProgram (POP), Tech. Rep. LAUR-02-2484, Los Alamos Natl. Lab.,Los Alamos, N. M.

Smith, R., and P. R. Gent (2004), Anisotropic Gent-McWilliams parameter-ization for ocean models, J. Phys. Oceanogr., 34, 2541–2564,doi:10.1175/JPO2613.1.

Smith, R., J. K. Dukowicz, and R. C. Malone (1992), Parallel oceangeneral-circulation modeling, Physica D, 60, 38–61, doi:10.1016/0167-2789(92)90225-C.

Smith, W. O., Jr., and H. J. Niebauer (1993), Interactions between biologi-cal and physical processes in Arctic Seas: Investigations using numericalmodels, Rev. Geophys., 31, 189–209, doi:10.1029/93RG00318.

Steiner, N., G. Holloway, and T. Sou (2003), Estimation of arctic wind-speeds and stresses with impacts on ocean-ice snow modeling, J. Mar.Syst., 39, 129–151, doi:10.1016/S0924-7963(02)00276-2.

Thorndike, A. S., D. A. Rothrock, G. A. Maykut, and R. Colony (1975),The thickness distribution of sea ice, J. Geophys. Res., 80, 4501–4513,doi:10.1029/JC080i033p04501.

Timmermann, R., H. Goosse, G. Madec, T. Fichefet, C. Ethe, andV. Duliere (2005), On the representation of high latitude processes inthe ORCA-LIM global coupled sea ice–ocean model, Ocean Modell., 8,175–201, doi:10.1016/j.ocemod.2003.12.009.

Tremblay, J.-E., and J. Gagnon (2009), The effect of irradiance and nutrientsupply on the productivity of Arctic waters: A perspective on climatechange, in Influence of Climate Change on the Changing Arctic andSub-Arctic Conditions, edited by J. C. J. Nihpul and A. G. Kostianoy,pp. 73–94, Springer, New York.

Walsh, J. J., D. A. Dieterle, W. Maslowski, and T. E. Whitledge (2004),Decadal shifts in biophysical forcing of Arctic marine food webs:Numerical consequences, J. Geophys. Res., 109, C05031, doi:10.1029/2003JC001945.

Walsh, J. J., et al. (2005), A numerical model of seasonal primary pro-duction within the Chukchi/Beaufort Seas, Deep Sea Res., Part I, 52,3541–3576, doi:10.1016/j.dsr2.2005.09.009.

Wassmann, P. (2011), Arctic marine ecosystems in an era of rapid climatechange, Prog. Oceanogr., 90, 1–17, doi:10.1016/j.pocean.2011.02.002.

Wassmann, P., D. Slagstad, C. W. Riser, and M. Reigstad (2006), Model-ling the ecosystem dynamics of the Barents Sea including the marginalice zone II. Carbon flux and interannual variability, J. Mar. Syst., 59,1–24, doi:10.1016/j.jmarsys.2005.05.006.

Yool, A., J. G. Shepherd, H. L. Bryden, and A. Oschlies (2009), Low effi-ciency of nutrient translocation for enhancing oceanic uptake of carbondioxide, J. Geophys. Res., 114, C08009, doi:10.1029/2008JC004792.

Yool, A., A. Oschlies, A. J. G. Nurser, and N. Gruber (2010), A model-based assessment of the TrOCA approach for estimating anthropogeniccarbon in the ocean, Biogeosciences, 7, 723–751, doi:10.5194/bg-7-723-2010.

Yool, A., E. E. Popova, and T. R. Anderson (2011), Medusa-1.0: A newintermediate complexity plankton ecosystem model for the global domain,Geosci. Model Dev., 4, 381–417, doi:10.5194/gmd-4-381-2011.

Zhang, J. (2005), Warming of the arctic ice-ocean system is faster thanthe global average since the 1960s, Geophys. Res. Lett., 32, L19602,doi:10.1029/2005GL024216.

Zhang, J., and W. D. Hibler (1997), On an efficient numerical methodfor modeling sea ice dynamics, J. Geophys. Res., 102, 8691–8702,doi:10.1029/96JC03744.

Zhang, J., and D. A. Rothrock (2001), A thickness and enthalpy distributionsea-ice model, J. Phys. Oceanogr., 31, 2986–3001, doi:10.1175/1520-0485(2001)031<2986:ATAEDS>2.0.CO;2.

Zhang, J., and D. A. Rothrock (2003), Modeling global sea ice with athickness and enthalpy distribution model in generalized curvilinear coor-dinates, Mon. Weather Rev., 131, 845–861, doi:10.1175/1520-0493(2003)131<0845:MGSIWA>2.0.CO;2.

Zhang, J., and D. A. Rothrock (2005), The effect of sea-ice rheology innumerical investigations of climate, J. Geophys. Res., 110, C08014,doi:10.1029/2004JC002599.

Zhang, J., and M. Steele (2007), The effect of vertical mixing on theAtlantic water layer circulation in the Arctic Ocean, J. Geophys. Res.,112, C04S04, doi:10.1029/2006JC003732.

Zhang, J., Y. H. Spitz, M. Steele, C. Ashjian, R. Campbell, L. Berline, andP. Matrai (2010), Modeling the impact of declining sea ice on the Arcticmarine planktonic ecosystem, J. Geophys. Res., 115, C10015, doi:10.1029/2009JC005387.

Zou, J., and G. Holloway (1994), Entropy maximization tendency in topo-graphic turbulence, J. Fluid Mech., 263, 361–374, doi:10.1017/S0022112094004155.

A. C. Coward, E. E. Popova, and A. Yool, National OceanographyCentre, Southhampton, University of Southampton, Waterfront Campus,European Way, Southampton SO14 3ZH, UK. ([email protected])C. Deal and M. Jin, International Arctic Research Center, University of

Alaska Fairbanks, Fairbanks, AK 99775, USA.F. Dupont, RPN MRD, Environment Canada, 2121 Trans Canada Hwy.,

Dorval, QC H9P 1J3, Canada.S. Elliott and E. Hunke, Los Alamos National Laboratory, Mail Stop

D413, Los Alamos, NM 87545, USA.M. Steele and J. Zhang, Polar Science Center, University of Washington,

1013 NE 40th St., Seattle, WA 98105, USA.


16 of 16

What controls primary production in the Arctic Ocean? Results...

Documents

Transcript of What controls primary production in the Arctic Ocean? Results...