tracers and large-scale modelsweb.science.unsw.edu.au/~matthew/0160.pdftracers, such as those...
TRANSCRIPT
TRACERS AND LARGE-SCALE MODELS
M.H. England, Centre for Environmental Modellingand Predicition (CEMAP), School of Mathematics, TheUniversity of New South Wales, NSW 2052, Australia
Copyright ^ 2001 Academic Press
doi:10.1006/rwos.2001.0160
Introduction
Geochemical tracers are included in large-scalemodels for a variety of applications. These include(1) ocean model assessment, (2) the analysis of ob-served tracer Relds, (3) simulation of the ocean car-bon cycle, and (4) comparisons with paleoclimatemeasurements. This article reviews the use of tracersin large-scale ocean models, particularly in thecontext of ocean model assessment. Important in-formation can be derived from chemical tracersimulations, such as how well models capture deepocean ventilation. Tracer studies have revealed inad-equacies in the model representation of certainwater-mass formation processes, for example, con-vection, downslope Sows, and deep ocean currents.This helps identify the need for model improve-ments, including better parameterizations of mixingand bottom boundary currents. The simulation ofchemical tracers is important in model assessmentand as a tool for analyzing water-mass mixing andtransformation. For climate model validation,a cost-effective approach is to simulate naturalradiocarbon to assess long timescale processes, andchloroSuorocarbon compounds (CFCS) for decadalto interdecadal ocean ventilation. Idealized scalartracers, such as those tracking water-mass age andconcentration, are also an important component ofLarge-scale ocean modeling.
Large-Scale Models
Computational ocean models are a key tool inoceanographic and climate research. They can addto the knowledge gained from direct ocean measure-ments, which are often sparse in space and time.Ocean models are also incorporated into climatesimulations, as the oceans play a vital role in regula-ting our climate. Ongoing model development isfocused on improving existing simulations of theocean’s circulation, which involves the criticalassessment of how well models represent the realsystem. An important aspect of this is the repres-
entation of water-mass formation, since this con-trols the removal of surface waters to greater depth,as well as determining poleward heat and fresh-water transports. Model assessment relies on oceanmeasuring programs but is fundamentally limited incertain ways. For example, many dynamic processesthat are linked with water-mass formation, such asconvection, mixing, and deep currents, are extreme-ly difRcult to measure directly. Modelers can usetemperature}salinity (TS) to provide a ‘proxy’means for assessment of the water-mass formationprocesses operating in models. In addition, seawatercomprises dissolved chemical tracers, such as sili-cate, oxygen, phosphate, CFCs, and isotopes of car-bon. The ocean inSux of these tracers can be time-dependent and distinct for different locations, sothey can provide detailed information on the path-ways and rates of water-mass renewal beneath thesurface mixed layer. In this way, tracers can be usedto assess the simulated circulation in ocean models.
Geochemical tracers that have been used to assessocean models include tritium, CFCs natural andbomb-produced radiocarbon, and, to a lesser extent,oxygen, silicate, phosphate, isotopes of organic andinorganic carbon compounds, and certain noblegases (e.g., mantle helium and argon). The choice oftracer depends partly on the timescales of interest.For example, natural carbon isotopes, particularlyradiocarbon, are useful for examining old water-masses such as the North PaciRc and CircumpolarDeep Water. Anthropogenic tracers, such as tritiumand CFCs, are best suited for analyzing model venti-lation over decadal timescales, such as thermoclineventilation and the renewal of Antarctic Intermedi-ate Water. Factors determining which tracer is bestfor a given problem include the atmospheric historyor input function of the tracer, whether the tracer isinvolved in biological processes, and whether thetracer is easy to measure and indeed well observedin the region of interest. A summary of the maingeochemical tracers used in ocean model assessmentis given in Table 1.
Radiocarbon
Radiocarbon (14C) is created in the atmosphere bothnaturally and through human activities such as nu-clear weapons testing. Its entry into sea water isonly via air}sea gas exchange with the surfacemixed layer, from where its natural radioactive
0001
0002
0003
0004
TRACERS AND LARGE-SCALE MODELS 1
RWOS 0160 BINNY GSRS SATHYAJIT
decay leads to a gradual differentiation of content inwater masses. The longer a given water mass hasbeen out of contact with the atmosphere, the moredepleted the 14C content becomes in proportion tothe concentration of the main carbon isotope, 12C.The half-life of 14C is 5730 years, making it suitablefor differentiating ocean ventilation rates over longtimescales. In addition, bomb tests during the 1950sdramatically increased the atmospheric concentra-tion of 14C, and this can be used to study shortertimescale ventilation processes.
The distribution of natural radiocarbon providesthe best measure of deep ocean ventilation ratesover century timescales and beyond. Modern an-thropogenic tracers (e.g., CFCs, bomb-producedtracers) only resolve the decadal to interdecadaltimescale. Radiocarbon simulations, therefore, pro-vide the most practical test of model ventilation inthe deep Indian and PaciRc Oceans, where renewaltimescales are of the order of hundreds to thousandsof years.
The 14C content of the ocean is expressed as thedeviation of the 14C/12C ratio (in parts per thousand&) from a reference atmospheric ratio. The nota-tion used for this difference is *14C, which is nor-mally negative since the 14C isotope decays in theocean. The more negative the value of *14C, themore time has elapsed since the water mass was lastin contact with the atmosphere. The *14C ratio isthe quantity normally simulated in ocean models,not the absolute concentration of 14C. This meansthat biological conversion processes can be largelyignored because they affect 12C and 14C compoundsin the same manner. In addition, because *14C simu-lations do not predict gas transfer on the basis ofactual 12CO2 and 14CO2 concentration levels, iso-topic fractionation between gaseous and dissolvedCO2 phases can be neglected. This means that theprebomb cycle of *14C can be forced towardsa time-independent atmospheric value of zero, typi-cally at a rate that is wind-speed dependent andcalibrated to match observed global mean gas ex-change rates. Interior *14C is transported by themodel circulation, convection and mixing processes,as well as undergoing natural radioactive decay witha half-life of 5730 years. When simulating bomb-produced 14C uptake, it is simply a matter of in-creasing the atmospheric reference value of *14Cfrom zero to the estimated atmospheric concentra-tions resulting from nuclear bomb testing during the1950s and 1960s.
An example of natural *14C simulations is shownin Figure 1, comparing the measured *14C in thewestern PaciRc during 1974 with that simulated intwo natural radiocarbon simulations. Experiment
B is from a robust diagnostic experiment whereinthe interior model TS are continuously restored to-wards climatological data. Experiment P is identicalto experiment B, except that there is no interiorrestoring of TS, so they are free to drift towardsunrealistic values. Experiment P is therefore prog-nostic in its treatment of water masses; they form atthe surface due to convection and vertical motions,and spread laterally due to interior currents andmixing. In contrast, case B has realistic values of TS,although they may be generated by the artiRcialinterior restoring terms. Which of the two modelsimulations determines the most realistic interiorventilation rates, can be distinguished by usingradiocarbon.
Apparent in the upper kilometer of the observa-tions is the penetration of high concentrations ofradiocarbon due to bomb-produced levels of the14C isotope. This could be simulated by running thenatural *14C simulations from 1950 onward, usingthe appropriate bomb-loaded atmospheric *14C con-centrations. Below 1 km, however, all of the ob-served *14C is derived from the prebombatmosphere. Clearly apparent is the relative successof the prognostic experiment (P) in simulating theventilation of North PaciRc Deep Water in contrastto the robust-diagnostic experiment (B). Similar re-sults have been obtained in the Atlantic Ocean;namely, that experiment P is superior to the robust-diagnostic case. This result is related to the wayconvection and vertical motions are suppressed inthe diagnostic integration because water masses arecreated by the interior restoring of TS to observa-tions; they need not form at the sea surface. Indeed,water-mass formation is largely suppressed by thestable stratiRed water column, which is artiRciallymaintained by the internal restoring terms. Withreduced convection and vertical motion, insufRcient14C is injected into the deep ocean, exposing a prob-lem with the robust-diagnostic technique. One ofthe key conclusions here is that a robust-diagnosticsimulation } in spite of a more realistic model TS} is inferior to a prognostic run in representingimportant ocean ventilation processes. This Rndingalso points directly to geochemical tracers as animportant component of ocean model validation ef-forts, since a correct TS Reld can theoretically sup-port spurious circulation patterns.
Tritium
Tritium (3H) is a radioactive isotope (or radionucl-ide) of hydrogen that was produced by atmosphericnuclear bomb testing in the 1950s and 1960s in anamount greatly exceeding its natural abundance. Be-
0005
0006
0007
0008
0009
2 TRACERS AND LARGE-SCALE MODELS
RWOS 0160 BINNY GSRS SATHYAJIT
cause of its time-dependent atmospheric history it isclassed as a ‘transient’ tracer. It decays into 3Hewith a half-life of 12.43 years. Its input functioninto the world oceans is rather complex as it de-pends on rainfall, air moisture, geographic location,and river input. Input from marine air-masses de-pends on latitude and rainfall/water vapor, the latit-ude dependence being weighted heavily towardsthose regions where nuclear bomb testing occurred(403}503N). Air masses that move out over the seafrom land have higher inputs for a given latitudethan do marine air masses, because gas exchangeover land is not as effective in removing tritium as itis over the ocean. Tritium concentrations are aboutfour times higher in continental air as comparedwith marine air. Lastly, river input of tritium issigniRcant since catchment areas over land can con-centrate tritium-borne rainfall into single entrypoints into the ocean. Tritium has been measuredand analyzed in several studies of oceanic circula-tion, particularly in the North Atlantic.
The main uncertainty in modeling ocean tritiumuptake is in the estimation of the source function.Although global input climatologies have been con-structed, the complex processes inSuencing tritiumentry into the oceans renders these estimatesa source of possible model error. However, thesource function errors are typically smaller thanthose due to model circulation errors, so validationstudies are still possible.
A model}observation comparison of tritium ona north}south section in the western Atlantic isshown in Figure 2. Apparent in the observations isthe southward spreading of tritium-laden North At-lantic Deep Water (NADW) at around2000}4000m depth. In the model simulation,a stronger than observed signal of downslope Sow-ing tritium is seen south of the Greenland}Ice-land}Scotland Ridge. This is probably because themodel has spuriously deep convective mixed layers,with near-uniform tritium concentrations of 4.0 TUreaching 3000m at 603N. In the observed section,noticeable vertical gradients of tritium are seen inthis region. As such, the model simulation of tritiumhas revealed a problem with excessive ventilationand overturn of surface waters in the North Atlan-tic. This could compromise the model’s ability topredict climate change or CO2 uptake in the region.
Chloro]uorocarbons
The industrial release of chloroSuorocarbons (CFCs)began in the early 1930s, and accelerated greatlyduring the next three decades. Despite the discoverythat CFCs destroy stratospheric ozone, the atmo-
spheric concentration of these gases has continuedto rise until very recently. Even though virtually allproduction and release occurs in the NorthernHemisphere, rapid mixing rates in the lower atmo-sphere and the chemical stability of CFCs ensurerelatively uniform distributions of these gases overthe troposphere. Their input functions are thereforesigniRcant over the entire ocean, unlike somebomb-produced tracers (e.g., tritium) whose atmo-spheric concentrations favour the Northern Hemi-sphere.
Because CFCs are weakly soluble in sea water,they dissolve at the sea surface like most otheratmospheric gases. However, relatively low solubil-ity and negligible biological interaction ensure thatthe ocean uptake of CFC is only a small componentof the global cycling of these gases (unlike, say,CO2). Upper level traces of CFC get redistributedvertically by convection and subduction, and thenhorizontally by interior ocean currents and mixing.Recently ventilated waters are therefore character-ized by comparatively high concentrations of theseanthropogenic gases.
Unlike the majority of ocean tracers, CFCs arenot known to be inSuenced by biological processesand are very stable compounds; they therefore serveas relatively unambiguous tracers of the present dayocean circulation over decadal to interdecadal time-scales. Direct measurements of the concentration ofchloroSuorocarbons (CFCs) in ocean are nowroutinely used to add information to conventionalhydrographic surveys. With relatively well-knownatmospheric histories and solubility properties in seawater, CFCs can be readily incorporated into gen-eral circulation models of the ocean.
An example of a CFC simulation is given in Fig-ure 3, which shows observed and modeled CFC-11along a transect in the South Atlantic near253W}353W during 1989. The observations wereonly taken as far south as 553S, no CFC data cantherefore be presented south of this latitude in theobserved panel. The dashed contours included in thediagrams are locally referenced density surfaces inclimatological observations as well as in two modelexperiments.
The two numerical experiments shown are identi-cal global ocean models, are only one case (ISOP)adopts simpliRed along-density surface mixing;whereas the other (GM) employs mixing that morerealistically parameterizes the effects of subgridscale eddies on the lager-scale transport of tracers.In the ISOP case extensive surface outcropping ofdeep density layers occurs in the Southern Oceanand Weddell Sea. In contrast, the observed deepocean is much denser than in ISOP and deep ocean
0010
0011
0012
0013
0014
0015
0016
TRACERS AND LARGE-SCALE MODELS 3
RWOS 0160 BINNY GSRS SATHYAJIT
density layers do not outcrop at the surface. So inthis model case there is an unrealistic pathway forpenetration of CFC into the ocean interior. Thistranslates to a strong vertical Sux of CFC to rela-tively deep waters. In the GM case deep oceandensity is greater, and deep-ocean density layers donot outcrop over the open ocean but instead onlyover the shelf in the southern Weddell Sea. Thisleads to less overturn of CFC-rich surface waters. Inaddition, a clear signal of CFC-depleted deep wateris seen at 503S in the GM model run. This corres-ponds to the upwelling of old Circumpolar DeepWater, as also noted in the observations. From thisCFC model comparison, it is clear that the moresophisticated mixing scheme produces a superiorsimulation of Southern Ocean water-mass circula-tion.
Ocean Carbon Cycle Tracers
Models of the oceanic carbon cycle embedded inthree-dimensional ocean models range in complexityfrom simple nonbiological CO2 perturbation studiesthrough to quite complex biogeochemical tracercycles. The simpler variety of oceanic CO2 uptakemodel assumes a perturbation approach, namelythat the preindustrial ocean carbon cycle continueswithout being affected by anthropogenic inSuences,such as increasing levels of atmospheric CO2 or theresulting changes in ocean circulation. This enablesthese models to ignore biological processes, therebygreatly simplifying the seawater chemistry includedin the model. Such simulations typically rely onother tracers (e.g., tritium and radiocarbon) to vali-date the simulated circulation Relds before includinganthropogenic CO2 uptake.
More complex models incorporating biologicalcycles typically include the production of biogenicorganic matter and nutrients at the surface, trans-port of biogenic material to the deep waters, andremineralization. This normally involves the oceanmodel carrying a number of chemical tracers asprognostic variables, including carbon compounds(such as total dissolved inorganic carbon and cal-cium carbonate), as well as phosphate, dissolvedoxygen, particulate organic matter, and silicate. Of-ten radiocarbon and CFCs also included for modelvalidation purposes.
Unfortunately most natural biogeochemicaltracers are not very useful for the purpose of large-scale model assessment. This is because they areaffected by both biological and circulation pro-cesses. A poor simulation of such a tracer mightindicate either deRciencies in the model circulationRelds or incorrect parameterization of a biological
process, or both. In addition, biological cycles aredifRcult to model explicitly. As such, for validationpurposes most ocean carbon cycle models rely oninert tracers (e.g., CFCs) or ones whose biologicalconversion processes can be ignored (e.g., *14C).
Some natural biogeochemical tracers can, how-ever, be used to assess model circulation Relds. Forexample, the simulation of deep phosphate, silicate,and oxygen in ocean carbon cycle models primarilyreveals the ocean’s conveyor belt at depth. Youngerwaters have low concentrations of phosphate andsilicate, and are enriched in oxygen and 14C, where-as older waters show higher concentrations inphosphate and silicate, and are oxygen-depleted.This can be used to assess model water-masses inmuch the same way as in observational oceanogra-phy. For example, Figure 4 shows the global distri-bution of POH4"(PO4#O2/177!1.95) at 3000mdepth in an ocean circulation model. The quantityPOH4 can be regarded as a physical tracer at depthbecause the effects of remineralization have beencancelled out by combining POH4 and O2 in thisway. The value 1.95 is arbitrarily subtracted inorder to obtain similar numbers to those for phos-phate, and O2 is scaled by 177 due to the RedReldratio for PO4 and O2. The distribution of POH4 at3000 m depth reveals the mixing effects between thesource waters of NADW (POH4"0.7) and WeddellSea Bottom Water (WSBW; POH4"1.8). In much ofthe PaciRc Ocean, for example, POH4 is almost uni-form at values between 1.35 and 1.4, indicatinga model mixing ratio of 30 : 70 between NADW andWSBW.
Other Chemical Tracers
So far we have looked at radiocarbon, tritium,chloroSuorocarbons, and standard biogeochemicaltracers in analyzing circulation and water-masses inocean models. There are enough measurements ofthese chemical tracers to enable meaningful oceanmodel assessment. There are several other chemicaltracers that are present in sea water and have thepotential for use in ocean modeling studies. Exam-ples include argon, mantle helium, and cesium, aswell as radioactive isotopes of common tracers suchas silicon and oxygen. However, because ofmeasurement difRculty } sometimes requiring large-volume seawater samples } there are generally notmany observations of these tracers. Mantle heliumis an exception in this context, with several transectsmade as part of the World Ocean Circulation Ex-periment.
The injection of mantle helium into the deep seacan be used to trace abyssal to middepth ocean
0017
0018
0019
0020
0021
0022
4 TRACERS AND LARGE-SCALE MODELS
RWOS 0160 BINNY GSRS SATHYAJIT
circulation. Helium produced in hydrothermal Suidsin seaSoor volcanism has a distinct isotopic signa-ture: it is highly enriched in 3He relative to 4Hecompared with atmospheric helium. Model simula-tions assume mantle helium injection along themid-ocean ridge axes with different local spreadingrates (Figure 5A). The forcing Relds are based onobservational studies of helium sources in mantlevolcanism and some less direct estimates. Sourcefunctions of mantle helium are still under investiga-tion, so the tracer is not ideal for direct modelvalidation studies. Also, deep mantle simulationsoften ignore the atmospheric source of 3He asa radioactive by-product of bomb-produced tritium,so simulations are then only valid away from re-cently ventilated waters, such as in the deep PaciRcOcean.
Results from a mantle helium simulation in thePaciRc Ocean are shown in Figure 5. In comparisonwith observed 3He, the model simulats a maximumin mantle helium that is too shallow and depleted in3He. Overall, insufRcient mantle helium is apparentin the model section, a result of excessive loss of3He through abyssal to surface upwelling in thePaciRc. Once upwelled into the surface layer, themantle helium is lost by air}sea exchange to theatmosphere. This rapid Sux of mantle-laden seawater from the abyssal PaciRc into the upper oceansuggests the model removes too much PaciRc DeepWater directly through equatorial upwelling. Itshould be noted, however, that uncertainties in theinput source function for mantle helium remain, sothe model-observed differences could in part be dueto the assumed Sux of 3He at the PaciRc OceanseaSoor.
Large-scale Ocean model simulations have alsobeen carried out using a number of natural andartiRcial radioisotope tracers. Examples include ar-gon-39, silicon-32, krypton-85, thorium-230, andprotactinium-231. However, there are relatively fewmeasurements of these tracers rendering model vali-dation over largescales nearly impossible. ArtiRcialradiotracers leaked from nuclear reactors, such as137Cs from the Chernobyl power station accident,can also be used to assess circulation models ofnearby regional seas. Examples include the BlackSea and the Kara Sea, as well as parts of the ArcticOcean.
In addition to the tracers described above, large-scale model simulations have also been performedwith stable isotopes, such as d18O and d13C. Directoceanic measurements of these tracers are quitesparse and their concentrations are complicated byfractionation, so they are generally not used forocean model assessment. They tend to be used to
study the role of fractionation in determining large-scale ocean distributions. In addition, these tracersare important in paleoceanographic studies; they aremeasured in deep-sea cores and coral debris andused to infer past climate conditions, includingoceanic TS. As such, their simulation in large-scalemodels aids their interpretation in paleoclimatestudies.
Idealized Tracers
Apart from incorporating chemical tracers intoocean simulations for model validation or carboncycle studies, specialized nonchemical or ‘idealized’tracers can be conRgured for other purposes. Exam-ples include tracers of water-mass age, tracers tag-ging the spread of predetermined source waters, andparticle (or ‘Lagrangian’ trajectory) tracers. The realworld analogue of these tracers might be difRcult toestimate, yet the model simulation can provide pos-sible insight into ocean ventilation rates and water-mass spreading pathways.
An example is given in Figure 6, which showswater-mass age near 2000m depth in an oceanmodel. Here, the age is deRned as the time taken forsurface waters to reach each location, averaged forall source water types contributing to the water-mass mixture. So, on average, it takes about1000}1200 years for surface waters to ventilate thedeep North PaciRc in this model. This time-scalecompares favorably with estimates using radiocar-bon. Source waters for the deep PaciRc comprisecomponents of Antarctic Bottom Water and NorthAtlantic Deep Water. Contrasting the PaciRc Ocean,relatively rapid ventilation rates are seen in the At-lantic, with NADW Sowing in a deep westernboundary current into the Southern Ocean. Thiswater mass of relatively young age can be seen toSow into the Indian Ocean, giving that deep basinmore rapid ventilation rates than the remote NorthPaciRc.
Uncertainties
Incorporating a chemical tracer into an ocean modelrequires knowledge of the tracer’s source functionand a faithful representation of its seawater chem-istry. Some of these issues have been discussedabove. For some tracers, the seawater chemistry iswell known whereas the input function is a sourceof uncertainty (e.g., tritium, mantle helium). Forothers, a complex seawater chemistry necessitatesinclusion of other geochemical tracers and biolo-gical effects (e.g., CO2, O2, phosphate). In this sec-tion uncertainties in simulating chemical tracer
0023
0024
0025
0026
0027
0028
TRACERS AND LARGE-SCALE MODELS 5
RWOS 0160 BINNY GSRS SATHYAJIT
uptake, particularly in the context of large-scalemodel assessment are discussed.
Source function uncertainties remain for bomb-produced tritium and mantle helium. Further studiesin these areas are required. For CFCs and 14C, un-certainties in the input function relate more to theparameterization of air}sea gas exchange, since at-mospheric concentrations are relatively well known.Uncertainty in air}sea gas Suxes is related to thespeciRcation of the gas transfer speed or piston velo-city. Several researchers have suggested bulk windspeed-dependent parameterizations for the gastransfer speed. However, the estimated value canvary by about a factor to two depending on thechosen formulation. This translates into uncertaintyin the model simulation of gas uptake, most notablyin slowly equilibrating gases such as CO2.
Tracer simulations are sometimes calculated in an‘off-line’ model in which the steady-state Relds of TSand velocity from an ocean model are used to ad-vect and mix the tracer. This is potentially useful inmodels of high resolution where the computationalcost of an ‘on-line’ tracer model can be prohibitive.Although computational savings are made with suchan approach, there are important factors that mustbe considered. For example, the tracer transporteffects of convective overturn and internal variabil-ity (e.g., eddies) must be considered, including theirfull seasonal cycle. The ‘off-line’ tracer model needsto include all physical transport processes of the‘on-line’ model.
Finally, it is of interest to consider how tracersimulations should be performed within coupled cli-mate models. If the interest of the study is to simu-late, for example, the oceanic carbon cycle duringanthropogenic climate change, the air}sea Sux ofcarbon should be simulated using the model-derivedwind speed, sea ice, TS, and so on. This is becausethese variables are likely to change during the cli-mate change run, affecting the simulated carbonuptake by the ocean. Indeed, the very goal of suchstudies is to examine how changed climatic condi-tions will alter the oceanic uptake of CO2. If, how-ever, the goal of the study is to validate the oceanmodel used within the climate system, an additionalsimulation needs to be performed. This secondmodel run should employ observed wind speed, seaice, and TS to estimate the gas Sux. This is becausecoupled models can simulate erroneous wind speedsand sea ice, which can bias the gas Suxes in someway. Then, compensating errors in model circula-tion can result in reasonable simulation of tracers.Using Sux estimates derived directly from observa-tions provides a less ambiguous test of climatemodel performance.
Summary
A number of geochemical tracers can be used toassess the circulation of water-mass formation inocean models. Tracers that have been in this contextinclude tritium, CFCs, natural and bomb-producedradiocarbon, and, to a much lesser extent, mantlehelium, stable isotopes of oxygen and carbon, andother less-measured radionuclides. Substantiallymore information can be derived from tracer experi-ments than that obtainable from TS alone. Forexample, natural radiocarbon can be used to assessocean models with regard to long time-scale circula-tion processes (centennial and beyond), such asdeep-water ventilation. Anthropogenic tracers suchas tritium, CFCs, and bomb-produced 14C are wellsuited for analysing model ventilation over interan-nual to decadal time-scales. It should be noted,however, that the ocean inSux of tritium and bomb14C is less accurately known than that for CFCs.As such, CFCs are presently the best availabletracers for decadal to interdecadal ocean modelvalidation.
The simulation of chemical tracers in ocean andclimate model assessment is highly recommended.A cost-effective approach in large-scale models is tosimulate natural 14C to assess long time-scale pro-cesses, and CFCs for decadal to interdecadal oceanventilation.
See also
Air+Sea Gas Exchange (60). Carbon Cycle (272).CFC’s in the Ocean (166). Chemistry of Hydrother-mal Vent Fauna (101). Cosmogenic Isotopes (179).General Circulation Models (394). General Pro-cesses in Ocean Circulation (107). Modelling theOcean Carbon System (173). Nuclear Fuel Rep-rocessing and Related Discharges (169). Radioac-tive Wastes (57). Radiocarbon (162). ThermohalineCirculation (111). Volcanic Helium (165). WaterTypes and Water Masses (108). 161. 408.
Further ReadingBroecker WS and Peng, TH (1982) Tracers in the Sea.
New York: Eldigio pressBullister JL (1989) ChloroSuorocarbons as time-depen-
dent tracers in the ocean. Oceanography 2: 12}17.England MH (1995) Using chloroSuorocarbons to assess
ocean climate models. Geophysical Research Letters22: 3051}3054.
England MH and Maier-Reimer E (2000) Using chemicaltracers to assess ocean models. Review of Geophysics(in press).
Libers SM (1992) An Introduction to Marine Bi-ogeochemistry. New York: John Wiley.
0029
0030
0031
0032
0033
0034
6 TRACERS AND LARGE-SCALE MODELS
RWOS 0160 BINNY GSRS SATHYAJIT
Maier-Reimer E (1993) Design of a 3D biogeochemicaltracer model for the ocean. In: Anderson D and Will-ebrand J (eds) Modelling Oceanic Climate Interactions,NATO, ASI Series, Vol. 111, Berlin: Springer-Verlag.415}464.
Sarmiento JL (1983) A simulation of bomb tritium entryinto the Atlantic Ocean. Journal of Physical Oceanog-raphy 13: 1924}1939.
Toggweiler JR, Dixon K and Bryan K (1989) Simulationsof radiocarbon in a coarse-resolution world oceanmodel. Journal of Geophysical Research 94:8217}8264.
TRACERS AND LARGE-SCALE MODELS 7
RWOS 0160 BINNY GSRS SATHYAJIT
Table 1 The principal chemical tracers used to assess large-scale ocean circulation models
Tracer Chemical formula Main source(s) Properties Applications
Natural radiocarbon 14C Natural isotope of 12C 5730 year half-life Long time-scaleventilation
Bomb radiocarbon 14C Nuclear bomb-testing 5730 year half-life Transient and decadalTritium 3H Bomb-produced
radionuclide12.43 year half-life Transient (favors NH)
Chlorofluorocarbons CClNFM Refrigerants, foams,solvents
Stable, inert Transient and decadal
Helium-3 3He Seafloor volcanism,3H by-product
Stable Deep water flows
Chemical formulae indicate the modeled isotope, compound or ion. Chlorofluorocarbons cover a variety of species (CFC-11 (CCI3F)and CFC-12 (CCI2F2) are the most common). NH refers to the Northern Hemisphere. Natural radioactive isotopes are created bycosmic rays in the atmosphere, then radioactively decay once dissolved in sea water.
a0160tbl0001
8 TRACERS AND LARGE-SCALE MODELS
RWOS 0160 BINNY GSRS SATHYAJIT
(C)
_ 40 _ 40
_ 240_ 220
Longitude
_ 200_ 180
_ 160
_ 60_ 80_ 100_ 120
_ 140
0
500
1000
2000
Dep
th
3000
4000
5000
90˚N 60 30 EQ 30 60 90˚S
(A)
0
500
1000
2000 Dep
th
3000
4000
500090˚N
160 12080 8040
_40_160
_160
_120
_240
_200_180
_220
00
_80
60 30 EQ 30 60 90˚S
W. PACIFICGEOSECSOBSERVATIONS
(B)
_ 40_ 60_ 80
_ 100_ 120
_ 160
_ 180_ 200
_ 140
_ 220_ 260_ 240_ 280_ 300
0
500
1000
2000 Dep
th
3000
4000
5000
90˚N 60 30 EQ 30 60 90˚S
W. PACIFICC14B
EXPERIMENT
_ 140
W. PACIFICP
EXPERIMENT
Figure 1 Western Pacific sections of *14C (&) in observations(A) and in two natural radiocarbon experiments. (B) ExperimentB is from a robust diagnostic experiment wherein the interiormodel TS are continuously restored towards climatological data.(C) Experiment P is identical to experiment B, except that thereis no interior restoring of TS.
a0160fig0001
10 TRACERS AND LARGE-SCALE MODELS
RWOS 0160 BINNY GSRS SATHYAJIT
80˚N604020EQ20406080˚S6000
5000
4000
3000
2000
1000
01000
800
600
400
200
0
Dep
th (
m)
Dep
th (
m)
(A)6000
5000
4000
3000
2000
1000
01000
800
600
400
200
00.8 0.8
0.4 0.6
0.2
2
0.4
0.2
46
88
6
51
35
5420.20.4
0.60.8
0.6 0.4 0.20.4
0.2
2
0.60.2
2
0.81
0.8
6 5
3 4
7
21
0.6
0.4
0
(B) 80˚N604020EQ20406080˚S
6000
5000
4000
3000
2000
1000
01000
800
600
400
200
0
Dep
th (
m)
Dep
th (
m)
6000
5000
4000
3000
2000
1000
01000
800
600
400
200
02
0.2
8 6
0.2
0.6 0.4
0.2
0.4
0.2
2
0.6
0.20.8 10.8 653 4
2
1
00
10.80.6
0.4
1 0.8 0.6 0.4 0.23
29 6 5
812
95
0.81
0.60.4
23
4
4 45
6
6 7
5
12
0.2
0.60.2
0.41
3 4
3
3
4
Figure 2 Tritium concentrations along a north}south section in the western Atlantic. (A) Measurements made in 1972; (B) a modelsimulation. Units are TU, (1 TU corresponds to a tritium/H2O ratio of 10~18).
a0160fig0002
TRACERS AND LARGE-SCALE MODELS 11
RWOS 0160 BINNY GSRS SATHYAJIT
1.41.4 1.4 1.5 1.6
1.3
1.5
0.7
0.8
0.91.0
1.1
1.41.7
90
90
60
60
30
30
0
Depth (m) 3000 Time (A) 1700
25˚E 25˚E85˚E 95˚W 35˚W145˚E 155˚W
˚N
˚S
Figure 4 Global distribution of POH4"PO4#O2/177!1.95 at 3000 m depth in an ocean model.a0160fig0004
12 TRACERS AND LARGE-SCALE MODELS
RWOS 0160 BINNY GSRS SATHYAJIT
25E 25E85E 95W 35W145E 155W
90
90
60
60
30
30
0
(A)
(B)
0
1
2
3
4
5
6
(C)
....... 0 _ 5 cm / y _1
------ 5 _ 10 cm / y _1
˚N
˚S
10 _15 cm / y _1
15 _20 cm / y _1
0
04 48 16 12
12
0
04
8 8 12
20
16
2016
80
12
24
0
Dep
th (
km)
1
2
3
4
5
6
90 60 30 EQ 30 60 90˚N˚S
Dep
th (
km)
4
Figure 5 (A) Source function of mantle helium assuming 3He injection along the midocean ridge axes with different localspreading rates. (B) Modeled section (Pacific Ocean, mantle source only). (C) Observed Pacific Ocean section of 3He(mantle#tritium by-product).
a0160fig0005
TRACERS AND LARGE-SCALE MODELS 13
RWOS 0160 BINNY GSRS SATHYAJIT
80˚N
60˚N
40˚N
20˚N
EQ
20˚S
40˚S
60˚S
80˚S
60˚E 120˚E 180˚E 120˚W 60˚W 0
Longitude
100
1200
1000
0
600
500
400
400
300
600
500 700
800
9001000
1100
0
200
400
300
400
Latit
ude
>
Figure 6 Water-mass age (years) near 2000 m depth in an ocean model. The age is defined as the time taken for surface watersto reach each location, averaged for all source water types contributing to the water-mass mixture. Waters younger than 100 yearsare stippled.
a0160fig0006
14 TRACERS AND LARGE-SCALE MODELS
RWOS 0160 BINNY GSRS SATHYAJIT
Figure 3 Vertical section of CFC-11 in the South Atlantic Ocean extending from the Weddell Sea northwards. (A) Observed, (B) ina model with simple isopycnal mixing, and (C) in a model including parametrized eddy mixing effects.
a0160fig0003
TRACERS AND LARGE-SCALE MODELS 9
RWOS 0160 BINNY GSRS SATHYAJIT