Tracers have always been an important adjunct to Tracers ... · PDF fileI5 Geochemical Tracers...

I5GeochemicalTracers and OceanCirculation

W. S. Broecker

15.1 Introduction

Tracers have always been an important adjunct tophysical oceanography. The distribution of dissolvedoxygen and to some extent of the nutrients, nitrate,phosphate, and silica) played a very important role in'defining the major water masses of the ocean [see Sver-drup, Johnson, and Fleming (1942) for a review of thissubject]. Many attempts also have been made to har-ness the loss of dissolved oxygen from the water col-umn as a measure of the rates of oceanic mixing proc-esses (e.g., Riley, 1951; Wyrtki, 1962). These latterpursuits, however, have been of only marginal successbecause of our lack of knowledge of the consumptionrate of 02 within the sea.

The big breakthrough in geochemical tracing cameafter World War II with the discovery of the cosmic-ray-produced isotopes 14C and 3H. A further impetus tothis field came with the realization in the mid 1950sthat the ocean was receiving significant amounts of90Sr, 13 7Cs, 3H, 14C, etc., from nuclear testing. Becausethe distributions of radioisotopes offered informationnot so highly dependent on assumptions regarding therates at which biological processes proceed in theocean, the emphasis in chemical oceanography movedquickly away from the traditional chemical tracers tothe radiotracers. Only quite recently has interest in thechemically used compounds in the sea been renewed.Three reasons can be given for this renaissance:

(1) Radiocarbon is transported in particulate matter aswell as in solution; hence the contributions of the twoprocesses must be separated if the distribution of 14Cis to be used for water-transport modeling. This sepa-ration is based on the distribution of ECO2 (concentra-tion of total dissolved inorganic carbon), alkalinity, anddissolved 02 in the ocean.(2) The concentrations of nitrate and phosphate can becombined with that of dissolved oxygen to yield thequasi-conservative properties "PO" and "NO". As re-viewed below, such properties are needed in modelingto unscramble the "mixtures" found in the deep sea.(3) With the advent of (a) sediment trapping and othermeans for the direct measurement of the fluxes of par-ticulate matter into the deep sea, (b) devices designedto measure the fluxes of materials from the sea floor,and (c) better means for the measurement of plant pro-ductivity, interest has been renewed in generatingmodels capable of simultaneously explaining the dis-tribution of the chemical species, the distribution ofthe radiospecies, and the flux measurements.

In this chapter I shall emphasize the development ofradioisotope tracing, as I feel that it constitutes themajor contribution of geochemistry to our understand-ing of ocean circulation over the past four decades (i.e.,since the writing of The Oceans). I will mention the

434W. S. Broecker

_I 1 ___·I_ __ _ _ _

use of the classical chemical tracers only where theybear on the interpretation of the radioisotope data.

Over the last decade, the Geochemical Ocean Sec-tions Study (GEOSECS) has determined the distribu-tion of the radioisotope tracers on a global scale. At-tempts to model the previously existing 14C results(Bolin and Stommel, 1961; Arons and Stonmmel, 1967)made clear the inadequacy of this data set. HenryStommel therefore brought together a number of geo-chemists interested in this problem, and encouragedthem to think big, to work together, and to produce aglobal set of very accurate 14C data.

Because of its massive scope and of the measurementaccuracy achieved, the GEOSECS data set has becomedominant in the field of marine geochemistry. Whilepreviously existing radioisotope data (for review seeBurton, 1975) were of great importance in the devel-opment of thinking with regard to the interpretationof tracer results and in the separation of the naturaland the bomb-test contributions to 14C and 3H, the newdata set eclipses what we had in 1969 when this pro-gram began. Thus I shall refer frequently to these newresults in the sections that follow.

At the time this chapter was written the GEOSECSfield program had been completed. Maps showing theship tracks and station positions are given in figure15.1. The laboratory analyses for the Atlantic and Pa-cific phases of the program are complete. Those for theIndian Ocean are still in progress. The mammoth jobof making scientific use of this data set has just begun.Many years will pass before the meat of this effort willappear in print.

15.2 Water-Transport Tracers

The efforts in the field of radioisotope tracing can bedivided into two categories: those that are aimed at abetter understanding of the dynamics of the ventilationof, and mixing within, the ocean's interior, and thosethat are aimed at a better understanding of the origin,movement, and fate of particulate matter within thesea. While many of the tracers we use are influencedby both processes, a division can be made into a groupprimarily distributed by water transport and into agroup primarily distributed by particulate transport(table 15.1). I will simplify my task by discussing hereonly the water-transport tracers.

The water-transport tracers can be subdivided ac-cording to their mode of origin. 9Sr, 13 7Cs, 85Kr, and thefreons are entirely anthropogenic in origin and henceare "transient tracers." 39Ar, 2 2 2 Rn, 2 2 8Ra, and 32Si are

entirely natural in origin and hence are steady-statetracers. 14C and 3H are in part natural and in part an-thropogenic in origin. In the case of 3H the man-madecomponent dominates. For 14C the man-made compo-

nent constitutes about 20% of the total in surfacewaters and is negligible in deep water.

3He, the daughter product of 3H, is also a tracer. It isproduced within the sea by the decay of its parent; italso leaks into the deep sea from the mantle. Thesecomponents produce an excess over atmospheric sol-ubility within the sea. As will be shown below, thecontributions of these sources can usually be separated.

The applicability of any given isotope depends on itshalf-life (in the case of steady-state tracers), or its tem-poral-input function in the case of the transienttracers). It also depends on the geographic distributionof the input function. The differences from tracer totracer are sufficiently large that the information ob-tained from the distribution of one isotope is not re-dundant with that obtained from another tracer. In thesections that follow the geochemistry of each of thesetracers is reviewed.

Natural radiocarbon: 14C is produced in the atmos-phere by the interaction of 14 N atoms and the neutronsproduced by cosmic rays. The current production rateis estimated to be about 2.2 atoms cm- 2s -' (Lingenfel-ter and Ramaty, 1970). This production is balanced bythe beta decay of radiocarbon atoms. In its 8200-yearmean lifetime, the average 14C atom can penetrate theactive carbon reservoirs (atmosphere, terrestrial bio-sphere, ocean, ice, soils). Some 14C atoms become

Table 15.1 Ocean Tracers Currently in Usea

Half- Originlife Cosmic U+Th Weapons Other

Isotope (years) rays series testing anthro.

14C 5730 V /

226Ra 1600 /32 Si 250 /

39Ar 270 /137 Cs 30.2 /

9°Sr 28.6 /3H 12.4 / /3 He - V/ v/85Kr 10.7 / /2 2 8Ra 5.8 /7Be 0.15 /

22Rn 0.01 /

Freons - /2 39 Pu 24,400 / V240u 6540 / v

21Pb 22.3 /2 28Th 1.9 /

210po 0.38 /2 34 Th 0.07 /

a. In the column headed Isotopes, freons and all entriesabove it are water tracers; all entries below freons areparticulate tracers.

435Geochemical Tracers and Ocean Circulation

(15.1A)Figure I5.I Maps showing the locations of the stations oc-cupied during the GEOSECS program in the Atlantic (A),Pacific (B), and Indian (C) Oceans. Those properties measurableon 30-liter samples were determined at every station. Radi-ocarbon, which requires -250 liters of water, was sampledonly at the large-volume stations. The 4000-m contour isshown on each map.

436W. S. Broecker

_ L_ _ � _ _ _II��_

(15.1B)


(15.1C)

438W. S. Broecker

_ 1��11 _ I_ I__

bound into inactive reservoirs (e.g., marine sediments,peats) before their radiodemise. Table 15.2 summarizesthe distribution of radiocarbon among these reservoirs.

Because of variability in the flux of cosmic rays, theproduction rate of '4C has not remained constant withtime. Also it is possible that the rates of the processesthat distribute 14C among the various reservoirs havechanged with climate. Thus, as shown by measure-ments on tree rings (figure 15.2), the 14C/C ratio inatmospheric CO2 has not remained constant with time.As pointed out by Stuiver (1976), these variations mustbe considered when attempting to derive ventilationtimes from the distribution of 14C within the sea.

During the present century the release of '4C-freeCO2 to the atmosphere through the combustion of fos-sil fuels has measurably reduced the 4C/C ratio. Thedecline between 1850 and 1950 was about 2.4%. Againthis must be considered in models. The 4C producedby nuclear .testing became measurable in 1954(Broecker and Walton, 1959).

Our knowledge of the 14C/C distribution within thesea is based on measurements on water samples takenat various regions and water depths and on measure-ments of CaCO3 formed by marine organisms at knowntimes in the past [historically dated shell collections(Broecker, 1963) and corals dated by ring counting(Druffel and Linick, 1978; Nozaki, Rye, Turekian, andDodge, 1978)]. As only a limited number of water sam-ples were collected prior to the onset of large-scalenuclear testing, the results of measurements on thesesamples are extremely important in distinguishing nat-ural from bomb 14C. For deep waters, measurementson post-1954 samples are adequate as significantamounts of bomb-produced 14C have not as yet pene-trated to great depth in the ocean (as demonstrated bythe absence of measurable tritium in these waters).

As the counters used to measure radiocarbon cannoteasily be absolutely calibrated, 14C measurements are

Table 15.2 Distribution of Natural Radiocarbon amongReservoirs

IC Z'44C/C

(1016 moles) (14C/C),at (104 moles)

Atmospherea 5 1.00 6

Ocean 280 0.87 290Ocean sediments - - - 0 b

Terrestrial biosphere -7 0.95 -8Soils -20 0.90 -2Continental sediments - - <1

a. The absolute 14C/C ratio for preindustrial atmospheric CO2was about 1.2 x 10-12, corresponding to a specific activity of8.0 dpm (gm C)-1 .b. Assumes calcite-rich sediments cover 20% of sea floor,that the mean calcite accumulation rate in these areas is1.5 gm cm-2 per 1000 years, and that the initial 14C/C for thiscalcite is equal to that for atmospheric carbon.

!0

I)UU IbUU I UU IUU IYUUYEARS (A.D.)

Figure 15.2 A14 C for atmospheric CO2 from 1450 to 1950 as

reconstructed from tree ring measurements (Damon, Lerman,and Long, 1978). Prior to 1900 the variations must reflectchanges in the production rate of natural 14C. The declinesubsequent to 1900 is caused by the release of 14C-free CO2 asthe result of burning fossil fuels.

made by comparison with a standard (National Bureauof Standards oxalic acid). The results are given as permil difference between the specific activity A of thesample carbon and 0.95 times the specific activity ofa standard carbon sample (the 0.95 multiplier is chosento bring the- activity of the standard close to that ofage-corrected 1850 wood; Broecker and Olsen, 1959,1961).

In order to separate the 14C differences produced byisotope fractionation, the 14C/C ratios in the samplesare normalized to a common 13C/C ratio (Broecker andOlsen, 1959, 1961). The formulas used in these calcu-lations are as follows:

814C = Aamp - 0.95 ANBS stan X 1000,ANBS stan

A14C = 814C - (281 3C + 50)(1 + '-) '

813 C = 1000 13C/12 C)samp - 13C/1 2 Cstan13C/12Cstan

Table 15.3 gives the L'4C values for natural radio-carbon in a variety of water types. The range is about200 per mil (from -4 0%bo for temperate surface waterto -250%o for 2500-m-depth water in the North Pa-cific). This distribution reflects not only the nite mix-ing rate of waters within the sea but also the finite rateof carbon-isotope equilibration between atmosphericCO2 and surface ocean XCO2. The difference betweenthe YCO2 content of average deep water and averagesurface water bears witness to the removal of carbon


Table 15.3 A 4C Values for VariousWater Types in the Ocean

A14CWater type (%O)

Surface WatersTemperate -40Equatorial -60

Deep AtlanticNew LSW - 70New GFZW - 70New DSW - 70NADW (mean) -100N. Western BasinNADW (mean) -115Eastern BasinAABW -160

Deep AntarcticCircumpolar -160

Deep PacificSamoan Passage -190OverflowN. Pac. Bottom -225N. Pac. 2500 m -250

YEAR

Figure 15.3 Excess atmospheric "C (resulting from bomb test-ing) as a function of time based on direct measurements ofthe 14C/C ratio in tropospheric CO2 samples. These measure-ments were made mainly by Nydal, Lvseth, and Gulliksen(1979). (See figure 16.1.)

from the photic zone by organisms and its return todissolved form at depth in the sea. This "short crcuit-ing" by raining particulate debris must be taken intoaccount in models used to derive ventilation rates fromthe 14C/C distribution. This is done by joint consider-ation of the stable carbon and radiocarbon distributions(Lal, 1962, 1969; Craig, 1969; Broecker and Li, 1970).

Bomb-produced radiocarbon: Beginning with thefirst fusion-bomb tests in 1954 significant amounts ofanthropogenic 14C appeared in the atmosphere. The14C/C ratio in atmospheric CO2, which had been fallingsince the turn of the century, began to rise sharply.This increase continued until 1963 when the U.S.-So-viet ban on atmospheric testing was implemented.Since then the atmospheric 14C/C ratio has been stead-ily falling (see figure 15.3). This decrease reflects thedilution of the bomb-produced "4C through mixingwith ocean and terrestrial biosphere carbon.

Through gas exchange with the atmosphere the sur-face ocean has taken up bomb ' 4C. Concurrent verticalmixing has carried this 14C to the interior of the ocean.Profiles of total 14C (natural plus bomb) as measuredduring the GEOSECS program are shown in figure 15.4.From the data we have concerning the prebomb distri-bution it is possible to estimate the depth distributionof the excess due to bomb testing. An ocean-wide meanpenetration depth of about 300 m is obtained as of 1973(the mid-point of the GEOSECS Atlantic and Pacificfield programs).

As of 1973 the distribution of bomb 14C was as fol-lows. About 35% remained in the atmosphere, about45% had entered the ocean, and about 20% had takenresidence in the terrestrial biosphere (table 15.4).Within the oceans the bomb-produced 14C is symmet-rically distributed about the equator. The temperategyres (15-45°) have considerably higher water-columninventories (i.e., bomb-produced 14C atoms cm-2) thandoes the equatorial belt (Broecker, Peng, and Stuiver,1978). shall discuss the oceanographic implications of

5 this equatorial anomaly later in the paper.Natural tritium: Like radiocarbon, tritium is pro-

duced by cosmic rays (von Buttlar and Libby, 1955).The production rate is estimated to be about 0.25 at-omscm-"2 s- (Craig and Lal, 1961). It enters the oceanvia vapor exchange and rainfall shortly after its birthin the atmosphere. As relatively few measurements ofoceanic tritium were made before the commencementof nuclear testing, our knowledge of its distribution isrudimentary (Giletti, Bozan, and Kulp, 1958). Further-more, because the blank corrections associated withthese measurements were quite large and also quiteuncertain, the validity of these data remains in ques-tion. Consequently, little use can be made of naturaltritium in ocean-mixing studies. Fortunately theamount of this tritium is surely too small to be signif-

440W. S. Broecker

U75

75

50(

25

- ------.

, ark

A 14C %o-120

1'4C %o-80 -40 0 40 80 120 ;,00

0 0100 PREBOMB /

200 -- 0 A ST 113 1 NO ST III 2°NA ST' 107 12°S

300 EASTERN EQUA. ATLANTIC

400 .

A 14C %.-120 -80 -40 0 40 80 12(

0 0 .

100 - PREBOMB ,

_20 A/ ao n _200 A A ST. 37 12°N

/ l ST 40 4°No lOST 48 4S

300 / o 0 ST 49 8S300 _ 0 i

WESTERN EQUA. ATLANTIC

400

100

200

300

400I-r

0 W 500

c3

600

700

800

Qnn

Figure 15.4 A14C vs. depth in the north temperate and equa-torial Atlantic Ocean as measured by Stuiver at the University

icant in modeling the distribution of tritium producedduring nuclear testing.

Bomb-produced tritium: Nuclear tests during the1950s and early 1960s produced large amounts of tri-tium (Begemann and Libby, 1957). As most of thisproduction was during tests in the northern hemi-sphere, this tritium was added mainly to the northern-hemisphere stratosphere. From here it leaked to thetroposphere (on a time scale of a year or so). Once inthe troposphere it is removed to the land and oceansurface by vapor exchange and rainfall on the timescale of weeks). Fortunately there are enough tritiumdata on rains and tropospheric water vapor collectedduring the peak fallout years to permit a reasonablysound reconstruction of the temporal and geographicpattern of its input to the ocean.

Furthermore, a picture of the time history for bombtritium in at least the northern-hemisphere surfaceocean can be reconstructed from measurements madebetween 1965 and the present.

Maps showing the tritium content of surface waterare given in figure 15.5. As would be expected, thetritium content of northern-hemisphere surface watersis on the average much greater than that for southern-hemisphere surface waters. The boundary betweenthese high- and low-tritium-content waters, however,lies near 15°N rather than at the equator. As discussedbelow, this unexpected result has important implica-tions with regard to the dynamics of the equatorialthermocline.

-50 0 50 100 50

I PREBOMB

IV laI

A +*1III

.

o+x

13

200I I I I I - -

I NORTH TEMP + A

I ATLANTIC + x +x 0A

0

o ST29 36N -o ST 31 27 N& ST 33 21'N* ST. 120 34°Nx ST. 117 31°N+ ST.115 29°N

I I I

I +I

,

I

of Washington and Ostlund at the University of Miami onsamples collected in 1972 as part of the GEOSECS program.

Table 15.4 Distribution of Bomb '4C as of 1974

EffectiveIC % Bomb(1016 moles) 14C

Atmosphere 5 -35

Surface ocean -20a -45Terrestrial biosphere -3 b -20

a. Assumes a characteristic penetration depth of 300 m anda bomb '4C/C ratio for surface ocean water averaging 30% ofthat in the air (in 1973).b. Assumes a characteristic replacement time of 15 yearsand the terrestrial biomass given in table 15.2.

44IGeochemical Tracers and Ocean Circulation

oA

CI

x

II-aw0

E

ILii0

.1_ . @ e- a

-%--e*

U -

- I

i

-

-

-

I II~vv

x. 90 70 50 30 10 10 30 A

A,' 90 70 50 30 10 10 30 '0"0

(1 5.5A) Figure 5.5 Maps showing the geographical pattern of tritiumconcentrations in surface ocean water as determined by theGEOSECS program in the Atlantic (A) and Pacific (B) Oceans.The measurements were made by Ostlund of the Universityof Miami. The units are T.U. (i.e., 10- ' T atoms per H atom).

442W. S. Broecker

* ~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -_ _

(15.5B)


Primordial 3He: Clarke, Beg, and Craig (1969) dis-covered that primordial 3 He stored in the mantle isgradually leaking into the deep sea. Subsequent workhas shown that this leakage is mainly from centers ofsea-floor spreading in the Pacific Ocean. This 3He hasbeen shown to diffuse laterally away from the ridgecrest, forming a Pacific-wide "plume" centered atabout 2500 m depth. This excess 3 He can be distin-guished from the atmospheric component either bysubtracting the air-saturation value of 3He contentfrom the observed 3He content or by comparing the3He/4He ratio in the water sample with that in atmos-pheric helium (corrected by 1.4% for higher solubilityof 4 He relative to 3 He). The ratio method is preferredbecause the abslute-concentration approach is madesomewhat uncertain by the fact that deep-sea waterappears to be several percentage points supersaturatedwith conservative atmospheric gases such as N2, Ar,and Ne, and hence is probably supersaturated with Heas well. This supersaturation is presumably induced byevaporative cooling of a surface film or by bubble en-trainment generated by breaking waves. As measure-ments (Lupton and Craig, 1975) of sea-floor hydro-thermal samples and of ridge-crest volcanic glassesreveal that the 3He/4He ratio in the exhaled mantle gasis rather uniform (eight to ten times larger than that inatmospheric helium), if the excess 3He is known to besolely of mantle origin (rather than in part from thedecay of tritium), then the excess can be calculatedfrom the 3He/4He ratio alone.

One of the interesting results of the measurementsof excess 3He in deep water is the contrast between theAtlantic and Pacific (Lupton, 1976). The Pacific showsa large (up to 30%) mid-depth anomaly (figure 15.6).This anomaly has its origin in the East Pacific Rise. Bycontrast the excess He in the Atlantic is carriedmainly by water returning from the Pacific via theAntarctic Bottom Water (Jenkins and Clarke, 1976).The anomalies generated by 3He release in the Atlanticare very small, the largest lying along the continentalmargin in the western North Atlantic rather than nearthe ridge crest (figure 15.7).

Radiogenic 3He: The decay product of 3H is 3He. Insubsurface waters the 3He produced by the decay of 3Hremains in the water (mass 3 is conserved). Whenwaters containing radiogenic 3He reach the surface,however, the excess 3He escapes to the atmosphere.Hence the ratio 3He/3H in a subsurface water samplecontains information with regard to its time of isola-tion from the surface (Jenkins and Clarke, 1976). For-tunately the 3He produced by the decay of tritium isspatially separated from that of the 3He leaking fromthe mantle. Where a significant overlap is suspected,the two excess components could be separated by pre-cise measurements of absolute 4He concentration (cou-pled with Ne, Ar, and N2 measurements aimed at es-

o

2

aI.0CUia

3

4

0 +10 +20 +3083He (%)

Figure I5.6 Distribution of excess 3He with depth in the Pa-cific. The profiles are by Clarke, Beg, and Craig (1970).

tablishing the degree of He supersaturation). Theexcess 4He content could then be multiplied by the3He/4 He ratio for average mantle-derived helium toyield the mantle component of the 3He excess. Theresidual 3He excess would then be that due to the decayof 3He. In practice this proves difficult because the ratio3He/4He in mantle helium is an order of magnitudehigher than that in atmospheric helium. A 1% error inthe measurement of 4He concentration would lead toa 83He error of about 10%.

2"6Ra: 226Ra is added to the ocean by diffusion fromdeep-sea sediments. With a half-life of 1600 years ithas been a prime candidate for a mixing tracer (Petter-sson, 1955; Koczy, 1958). Several factors have dimmedthese hopes. First it was found that the deficiency of226Ra in surface water relative to deep water is far toolarge to be explained by radiodecay. Removal of 226Raon settling particulate matter was invoked to explainthis difference. This hypothesis was confirmed bymeasurements of barium (Wolgemuth and Broecker,1970). As barium and radium are nearly identical intheir chemical behavior, it was hoped that through theuse of 226Ra/Ba ratios (analogous to the use of 4C/Cratios) the effects of particulate transport could be sep-arated from those of mixing [see Chan et al. (1976) fora discussion of the GEOSECS results].

This strategy suffers, however, from a severe problemof dynamic range (i.e., ratio of signal to observationalerror). If the deep sea is ventilated on a time scale of1000 years, then for a tracer added mainly to the bot-

444W. S. Broecker

STATION NUMBER

E

I_a.wr~

S

(15.7A)

LATITUDE N

STATION NUMBER

70 60 50 40 30

(15.7B) LONGITUDE

Figure I 5.7 Sections along the western Atlantic basins GEO-SECS track (A) and the GEOSECS track across the NorthAtlantic (B), showing the 3He anomalies generated within theAtlantic Ocean (given in percentage of 3 He excess over at-mospheric). The contribution of excess 3He carried by deepwaters returning to the Atlantic from the Pacific (via theAntarctic) has been eliminated through use of the linear"NO"-SiO2 relation for deep Atlantic waters (see Broecker, inpress). The data on which these plots are based are fromJenkins and Clarke (1976) and Lupton (1976).

tom of the sea the percentage difference A between theratios of radioisotope to "carrier" ratio in deep waterand in surface water should be approximately

A = 100 { 1- exp [ 000(VS/VD)]Tisotope I I

For a ratio of VS (volume of surface water) to VD (vol-ume of deep water) of about 0.1, the surface-to-deep-water difference predicted for 226Ra (half-life = 2300years) is only 6%. The observational error on the GEO-SECS 226Ra measurements averages about 3% and thaton a GEOSECS barium measurement about 1%. Thusany age message is buried in observational noise. De-spite the fact that 14C has a half-life 3.5 times longerthan that of 226Ra, because it enters the ocean from thetop rather than from the bottom, it has a more thancompensating advantage. As mentioned above, a 21%range is seen in the natural 14C/C ratio within the sea.Thus the range of natural 14C/C ratios in the sea shouldbe about three times the range of 226 Ra/Ba ratios. Fur-thermore, the 14C/C ratio can be measured with tentimes higher precision (±0.4%) than the 226 Ra/Ba ratio(+4%). The overall dynamic-range advantage of 14Cover 226Ra is thus thirty!

Although it may be possible to improve the meas-urement error for 226Ra (to perhaps 1%), uncertaintiesin the geographical pattern of 226Ra input from marinesediments will always plague the use of 226Ra as a timetracer. For these reasons 226Ra should perhaps be de-leted from the list of mixing tracers. Nevertheless, asshown by results from the deep northeastern Pacific,where a great excess of 226Ra relative to Ba has beenfound (Chung, 1976; Chung and Craig, in press), the226Ra distribution will contribute to our understandingof the release patterns of metals to deep water fromsediments. Also, the 226Ra distribution in the oceanmust be known if we are to use the distributions of228Ra and of 222Rn as time tracers for mixing, and if weare to use 210Pb and 216po as time tracers for particulateprocesses.

228Ra: 228Ra has a half life of 5.8 years. Like 226Ra itenters the ocean primarily by diffusion out of marinesediments (Moore, 1969). Because of its far shorter half-life, however, it has a much different distributionwithin the sea than does 226Ra. Profiles with depth (seefigure 15.8 for an example) show maxima at the surfaceand at the bottom with low values in the mid-watercolumn. The high surface values reflect input of 228Rafrom the continental shelves coupled with rapid hori-zontal dispersion, and the high bottom values reflectthe release of 228Ra from deep-sea sediments. The lowmid-depth values are the result of lower ratios of sed-iment area to water volume, and of the finite rate ofvertical mixing. (The 228Ra released from shelf and bot-tom sediments cannot reach this depth range beforeundergoing radiodecay.)


E

0Jmgr

V

2

3

4

2ZERo Concentrotion (dpm / 100 kg)

0.5 1 0 1.5 20

25 26 27

a (%)

28

Figure 5.8 228Ra concentration as a function of depth at theGEOSECS Atlantic calibration station (36°N, 68°W) (Trier,Broecker, and Feely, 1972).

That 228Ra that enters the deep sea probably can betreated as if it were chemically conservative. In surfacewater, however, removal by particulate matter maycompete with radiodecay and mechanical mixing. For226Ra the probability of removal from surface water viaparticulate transport appears to be three times higherthan by vertical mixing. If the residence time of waterin the upper 400 m of the ocean is taken to be 50 years,then the residence time of 226 Ra with respect to partic-ulate removal is about 18 years. As the mean life of228Ra is 8 years, correction for particulate transporteffects may prove significant (i.e., -20%).

2 22Rn: Radon (half-life 3.85 days), is produced withinthe sea by the decay of 226Ra dissolved in the sea. Wellaway from sedimentary margins of the ocean, in situ226Ra decay is the only source of 222Rn. This is true forthe surface mixed layer of the open ocean (the atmos-pheric 222Rn content is too low to be a significantsource). Loss to the atmosphere of the radon producedwithin the surface-ocean mixed layer provides a meansof determining the rate of gas exchange between theocean and atmosphere (Broecker, 1965). Close to thesediment-water interface, radon released from the up-per few centimeters of the sediments (and from theupper few millimeters of Mn nodules and Mn-coatedrock pavements) is found. This excess radon providesa means of determining the rate of mixing in the near-

bottom ocean (Broecker, Cromwell, and Li, 1968). Bothof these applications require a knowledge of the 226Racontent of the water. To determine gas-exchange ratesthe difference between the equilibrium 222Rn content(calculated from the 226Ra content) and the observedmixed-layer 222Rn must be known. For near-bottommixing the sediment-derived 222Rn is estimated by sub-tracting the 2 26Ra-supported 222 Rn from the measured222Rn content of the abyssal water.

9Sr and 137Cs: The usefulness of fission products inoceanography was first demonstrated by Bowen andSugihara (1957, 1958, 1960). Prior to nuclear testingthe environment was free of the fission products 137Csand 90Sr (both have half-lives close to 30 years). Theamounts of these isotopes currently present in theocean are entirely the result of these tests. Like 3H theywere added mainly to the northern-hemisphere strat-osphere. Like 3H they were deposited mainly in thenorthern hemisphere. Like 3H their ocean distributionshows much higher water-column inventories north of15°N than south of this boundary (Kupferman, Liv-ingston, and Bowen, 1979). In fact they are so muchlike 3H that no one has yet demonstrated that theiroceanic distributions contain any information with re-gard to mixing not contained by the 3H distribution(Dreisigacker and Roether, 1978). As 1

37Cs and 90Sr re-quire far larger water samples than 3H, as they cannotcurrently be measured with the accuracy or reliabilityof tritium, and as there are indications that '37Cs maynot be conservative within the sea, emphasis has beenplaced by most geochemists studying mixing processeson the measurement and interpretation of tritium data.

90Sr should, however, not be written off as a mixingtracer. Its input distribution at the earth's surface isbetter known than that of 3H (because it accumulatesin soils). It has a longer half-life (30 years) than does3H (12 years), and therefore will be around for a longertime. Its concentration in tropical surface water hasbeen recorded by corals (there is no similar paleotri-tium recorder).

X5Kr: 85Kr is a product of fission. Thus it is producedduring bomb testing, plutonium production, andpower-reactor operation. Much of that produced hasleaked to the air (Pannetier, 1968; Schroeder andRoether, 1975; Telegadas and Ferber, 1975). Measure-ments reveal a strong buildup in this reservoir (seefigure 15.9). By gas exchange 85Kr reaches the surfaceocean (its current specific activity is about 3 dpm m-3 ).Because of this very low activity only a few measure-ments of 85Kr have been made in the sea to date(Schroeder, 1975). As expected, its vertical distributionshows a first-order similarity to that of the other an-thropogenic tracers (3 H, 14C, 9 0Sr, 137Cs). If this covari-

ance were perfect, then there would be no interest in

446W. S. Broecker

_ 1--

228, Ra

, :

,- .III"",I

__ _ _1___1_1___ 1~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

I . I I . . I . . .

0EE

EE

a.IL~y

0

YEAR

Figure I5.9 Kr as a function of time in the atmosphere(Schroeder and Roether, 1975).

going to the trouble to collect and process the largewater samples necessary for measuring this isotope.

There is, however, at least one very important ap-plication for this isotope. As is discussed at greaterlength later in the paper, a difficulty arises in estimat-ing the concentration of excess CO2 (i.e., that stem-ming from the production of fossil fuels) in currentlyforming deep water. Because the chemical equilibra-tion time of CO2 between mixed-layer water and theatmosphere is of the order 1 year while the coolingtime is on the order of 1 month, newly formed deepwaters are likely to carry less than their capacity ofexcess CO2. The most promising approach to findinghow much less appears to be through the use of twotracer gases: one a gas that equilibrates on a time scalelonger than that for CO2, and the other a gas thatequilibrates on a time scale shorter than that for CO2."4CO2 is the only choice for the former. As discussedby Broecker, Peng, and Takahashi (in press) the isotopicequilibration time for carbon exceeds the chemicalequilibration time by a value numerically equal to thebuffer factor (-12 for cold surface water). 85Kr and thefreons are the prime candidates for the short-exchange-time gas. Like all gases of normal solubility their ex-change time is about 1 month (Peng et al., 1979). Al-though the freons have the advantage that inexpensivemeasurements can be made on small samples, theyhave yet to be proven conservative in sea water. Proofof their long-term stability can only be achievedthrough comparison with a second tracer of the sametype. 85Kr is the only candidate for this task.

39Ar: Cosmic-ray-produced 39Ar is the ideal deep-ocean tracer. As it exchanges rapidly with the air, theproblem of establishing its concentration in newlyformed deep water is far smaller than for 14C. Its half-life of 270 years is far more appropriate to deep-watermovement times than that of 14C. The problem in its

use is technical (Loosli and Oeschger, 1968, 1979;Oeschger et al., 1974). Very large samples (several tons)and ultralow background counting are needed. OnlyLoosli and his colleagues at the Bern laboratory cur-rently are capable of measuring this isotope in the sea.Because its potential is so large, however, the nextdecade is bound to bring forth many 39Ar measure-ments.

32Si: Since Schink's (1962) early attempts, two de-cades of effort have gone into determining the distri-bution of 2Si (a cosmic-ray-produced isotope with ahalf-life of -250 years) in the sea. Like 39Ar its collec-tion and measurement are fraught with technical dif-ficulties. Using the in situ extraction technique ofKrishnaswami et al. (1972) about 200 samples werecollected during the GEOSECS program. Three profileshave been published (Somayajulu, Lal, and Craig, 1973).The remainder of the analyses are still in progress atthe Physics Research Laboratory at Amedabad in India.Even when these data appear, three serious problemswill remain for their interpretation. First, the perform-ance of the bag samplers used to collect the silica wasfar from perfect. Second, since the particulate cycledominates the movement of silica down the water col-umn, isolation of the water dynamics from the particledynamics constitutes a difficult problem. Finally, themeasurement errors are sizeable and will surely lead toconsiderable ambiguity in the interpretation. If the 39Armethod succeeds, 32 Si may well be relegated to the areaof particulate research. Were the circulation dynamicsindependently known, the 32Si distribution could tellus much about the pattern of dissolution of siliceousparticles.

Freons: Most of the freon produced as a propellentand as a refrigerant ultimately finds its way to theatmosphere. Reliable and accurate means of measuringit in air and sea water have been worked out (Lovelock,Maggs, and Wade, 1973). Measurements on sea watershow its distribution to be similar to that of tritiumand other "fallout" nuclides (Hahne et al., 1978). Asmentioned above, once it can be established that freonsare conservative in the sea, then this substance willbecome one of the more useful of the transient tracers.

7Be: With a half life of 53 days, cosmic-ray-produced7Be has great potential as a tracer of upper thermoclineprocesses (Silker, 1972a,b; Silker et al., 1968). An ex-ample of its distribution in the upper water column isshown in figure 15.10 (36°N, 68°W). Unfortunately,because of the large volume of water needed (thousandsof liters) and of the complexity of the counting systemneeded for its detection, little work has been done totake advantage of the information carried by this iso-tope. Were it to be used, attention would have to begiven to its removal by particulate matter and to theseasonal variation in its distribution.


TEMP (C)

?Be (dpm/m 3 )

Figure iS.io 7Be as a function of depth at the GEOSECS At-lantic calibration station (36°N, 68°W) occupied during Au-gust 1970 (Silker, 1972a). Solid curve is temperature profile.

15.3 Water-Mass Tracers

All deep waters in the ocean are mixtures from two ormore sources. Since the concentrations of the timetracers in each source are potentially different, thesemixtures must be unscrambled if we are to use theisotope data successfully to yield ventilation rates. Forthis job we need properties that are conserved once thewater ends contact with the atmosphere. Candidatesare as follows:

(1) Conductivity. The conductivity of a water parcelisolated in the deep sea changes only because smallamounts of SiO2, C0 2 , P04--, Ca++ and NO0 are addedvia respiration and mineral solution. As discussed byBrewer and Bradshaw (1975), these changes are justmeasurable. For any of the applications involving ra-dioisotope "dating" of sea water, the salinities calcu-lated from conductivity measurements may be treatedas perfectly conserved.(2) Temperature. As heat transfer by molecular con-duction is of negligible importance to the internal en-ergy content of a sea water parcel, potential tempera-ture can be treated as conservative. The only source ofheat within the deep sea of significance is geothermalheat. Except in rare circumstances this contribution issmall enough not to create difficulties. Because heat-flow rates are well known (except perhaps along ridgecrests), the effects of geothermal heating can often betaken into account in models of radioisotope data (i.e.,the temperatures can be corrected by iteration).(3) 2H and 180. The isotopic composition of the watersventilating the deep sea varies from source region tosource region (Epstein and Mayeda, 1953). As discussedby Craig and Gordon (1965), these variations do notstrongly correlate with either salinity or temperature.

The isotopic messages from 2H and 180, however, arehighly correlated with each other. These signatures areconservative. The problem lies with the measure-ments. An acceptable dynamic range can be achievedonly by painstaking care using the best available massspectrometers and preparation lines. Because of thisonly a limited number of such measurements is cur-rently available. Even for measurements of the highestattainable precision the dynamic range may prove tobe too small to be useful.(4) "NO" and "PO". Although dissolved-oxygen gas isutilized by animals and bacteria living beneath thephotic zone in the sea, a correction can be made forthis utilization by taking into account the fact that foreach 135 molecules of 02 consumed about 15 mole-cules of NO0 and about one molecule of P04-- arereleased to the water as dissolved ions. Thus the sums[02] + 9[NO3] and [02] + 135[PO4] are potentially con-servative properties (Broecker, 1974). The near con-stancy of the NO3/PO4 ratio in most parts of the deepsea supports this concept. Since the coefficients 9 and135 are not truly "stoichiometric" quantities, however,care must be taken in using "NO" and "PO". Theactual coefficient applying to any given mass of watermight deviate from these values. "NO" has an addi-tional drawback. In areas of the ocean where the dis-solved 02 content reaches 1% or less of its saturationvalue, NO3 is used by bacteria as an oxidant. Thus, inregions adjacent to zones of severe 02 depletion (likethe thermocline off Central and South America), "PO"rather than "NO" should be used in "unmixing" cal-culations.(5) SiO2 and Ba. Silica and barium are concentrated indeep waters relative to surface waters, and show largedifferences from water type to water type (Edmond,1974; Chan, Drummond, Edmond, and Grant, 1977;Chung and Craig, in press). As these differences aregenerated by the removal of these substances from theupper ocean followed by regeneration in the deepocean, Ba and SiO2 are by no means conservative.Nevertheless, observation has shown that in certainregions of the ocean they are nearly conservative andcan be successfully used for "unmixing" calculations.An example is given below.

15.4 Modeling Tracer Data

The ultimate use of the radioisotope data is to obtaininformation regarding transport of water and of speciesdissolved in the water within the sea. Deconvolutionof the tracer field into the various modes and directionsof transport remains an unsolved problem. As mightbe expected, modeling of radioisotope data has under-gone an evolution that paralleled the size and qualityof the data set. The first models were of the box variety.

448W. S. Broecker

The ocean-atmosphere system was approximated by afew well-mixed reservoirs. The '4C data were used todefine transfer coefficients between these reservoirs.Initially, three-box models that separated the systeminto a well-mixed atmosphere, a well-mixed surfaceocean, and a well-mixed deep ocean were used (Craig,1957). The fluxes derived were the air-sea gas exchangerate and the water transfer rate across the main ther-mocline. When 14C data for the Pacific Ocean (Bien,Rakestraw, and Suess, 1963) as well as the AtlanticOcean (Broecker, Gerard, Ewing, and Heezen, 1960)became available, a large difference in the apparent agefor the deep water in the two oceans became evident.This generated an incentive for models with moreboxes. It was also realized that the regions of deep-water formation (North Atlantic and Antarctic) shouldbe treated separately. Various attempts were made togenerate multibox models (Broecker, 1963; Keeling andBolin, 1967, 1968; Broecker and Li, 1970). In all theseattempts, owever, it was realized that because thenumber of fluxes needing definition increased as thesquare of the number of boxes while the number ofnew pieces of information increased linearly with thenumber of boxes, this approach was destined to yielda considerable degree of ambiguity. The idea of usingsalinity and temperature to aid in the definition offluxes, while promising in a mathematical sense, alsoproved impractical because of the uncertainties in theatmospheric boundary conditions for heat and salt.Geochemists fell back to the use of the simpler three-box models as a guide to geochemical thinking, andsuspended attempts to use box models to answer phys-ical oceanographic questions.

Munk (1966) initiated an alternate approach to theuse of isotope data. He focused his attention on thetrend of 14C/C ratio with depth within the deep Pacificand showed that, if the type of advection-diffusionmodel used by Stommel (1958), Robinson and Stommel(1959), Stommel and Arons (1960b) and Wyrtki (1961b,1962) were employed, the 14C trend would yield abso-lute values for the upwelling velocity and the verticaldiffusivity. Craig (1969) elaborated on Munk's idea,pointing out that 14C and C should be treated separatelyin the calculation. By fitting both the 14C concentrationprofile and the XCO2 concentration profile, the contri-bution of 4C through the rain of particulates could beaccounted for. A reader interested in a review of thissubject is referred to the excellent paper by Veronis(1977). As the GEOSECS data came in, it became ap-parent that the horizontal gradients within the deepPacific could not be neglected. For this reason interestin the simple one-dimensional advection-diffusion haswaned.

Kuo and Veronis (1970, 1973) initiated the first two-dimensional modeling of the deep-ocean tracer field

including advection and diffusion. They assumed thatdeep water formed at two small high-latitude regionsand upwelled uniformly over the entire ocean. Theyused a modified form of the horizontal advection pat-tern envisioned by Stommel (1958), and assumed uni-form lateral and vertical diffusion rates. Their endeav-ors focused on matching the deep-O2 distribution.Fiadeiro and Craig (1978) adopted a variation on thisapproach and did more elaborate three-dimensionalcalculations for the deep Pacific. Interest in this ap-proach has also waned, however, because of seriousreservations with regard to the suitability of the ad-vection pattern adopted for the models.

During the last 2 years Sarmiento has been applyingfluxes of bomb-produced isotopes to Bryan's quasi-di-agnostic model (Sarmiento and Bryan, in preparation),the idea being to see whether the model reproduces theobserved distribution of these tracers. This work is stillin progress.-'Despite the failure to produce a satisfactory modelwith which to match the tracer data for purposes ofelucidating the dynamics of deep-ocean ventilation,one very important application has emerged. Throughtracer-tracer analog modeling the distributions of theradioisotopes can be used to predict how much fossilfuel CO2 has been and will be) taken up by the sea. Anumber of investigators have used three-box modelscalibrated using the distribution of natural radiocarbon(Bolin and Eriksson, 1959; Broecker, Li, and Peng, 1971;Machta, 1972; Keeling, 1973). Oeschger, Siegenthaler,Schotterer, and Gugelmann (1975) have used a vertical-diffusion model calibrated with natural radiocarbon.Broecker, Takahashi, Simpson, and Peng (1979) havesubsequently shown that the diffusivity chosen byOeschger et al. (1975) is consistent with the verticaldistribution of the bomb-produced tracers. Severalgroups of investigators are currently developing moreelaborate models of this type. While not giving us anybetter idea of the physics of ocean circulation, thesemodels do provide us with an interim approach to aproblem of great import to society.

15.5 Current Applications

We shall consider four applications of the tracer results.The first is the ventilation time of the deep sea. Thesingle great triumph of the natural radiocarbon meas-urements in the ocean has been the establishment ofa 1000-year time scale for the residence of water in thedeep sea. Now that we have the GEOSECS data, whatmore can be said on this subject? The second is theventilation time for the main oceanic thermocline. Thetransient tracers 14C and 3H prove ideal for this task.The third is the dynamics of deep-water formation. Inthis process the interaction between surface water andthe atmosphere constitutes a rate-limiting step for the


tracers used. Again, the transient tracers prove idealfor the task. Finally, the distributions of the short-livednatural tracers and of the anthropogenic tracers allowupper limits to be placed on the rates of vertical (dia-pycnal) mixing.

15.6 Ventilation of the Deep Sea

Although a number of estimates of the mean ventila-tion time for the deep sea have been proposed, that ofabout 1400 years, based on radiocarbon, is the mostwidely quoted. In the three-box model the ventilationtime T of the deep sea is given by the following equa-tion:

R (Vl(4 c c s ] I

where VD is the volume of the deep sea, R the venti-lation rate of the deep sea, ('4C/C)s and (14C/CD thecarbon isotope ratios for the mean surface and meandeep ocean (prior to the industrial revolution), and Tl4cthe mean life of 4C (i.e., 8200 years). Taking(14C/C)s/(14C/C)D to be 1.17 + 0.03 i.e., A14CS = -50 -10%bo and A' 4CD = -190 - 20%°o), a ventilation time of1400 ± 250 years is obtained. Among other things thismodel assumes that the ventilation times for waterand carbon are equal. Because the 4C/C ratio in newlyformed deep water has been shown not to reach thatobserved for the warm surface ocean, the ventilationtime for water is surely less than that given by the boxmodel. If the box model calculations used to obtainthis result were repeated using average 14C/C ratiosobtained from the GEOSECS data set, the result wouldnot change significantly. Evolution in geochemical andoceanographic thinking since the heyday of box mod-eling makes clearer the problems associated with using14C data to estimate water residence times in the deepsea. Until means of overcoming these problems areachieved it will not be possible to evaluate properly

the ventilation times derived from '4C data, nor will itbe possible to improve on them.

One problem is related to the fact that while themixed layer of the ocean can be cooled (through contactwith winter air) in a few weeks, equilibration betweenthe 4C/C ratio in the carbon dissolved in the mixedlayer and that in the atmospheric CO2 requires a decadeor more. Harmon Craig recognized this problem longago but until recently it has received only casual at-tention. The upshot is that the 14C residence time inthe deep sea must exceed the water residence time.Results from the Weddell Sea (Weiss, Ostlund, andCraig, 1979) dramatically illustrate this point, forthe new deep water formed there contains nearly thesame '4C/C ratio as ambient circumpolar deep water.

The other problem has to do with the recirculationof old water within the deep sea. Rarely is it possibleto isolate the component of 4C/C change due to in situaging. Rather, the changes in '4C/C ratio within a givendeep-sea basin can be accounted for largely by the mix-ing of end members with different 14 C/C ratios. Thisgreatly complicates the use of 14C data to obtain theflux of deep water into a given basin.

The trend of A14C along the 4000-m horizon is shownin figure 15.11. The pattern shows a gradual decreasein 4C/C "down" the Atlantic and "up" the Pacific(with similar values in the Antarctic segments of thetwo oceans). The down-Atlantic decrease is about 80%oand the up-Pacific decrease is about 70%o. While thismight be interpreted as evidence for gradual aging asthe water moves by advection slowly around the globe,a close look at the situation reveals that at least in theAtlantic the trend is largely the result of mixing be-tween low 14C/C waters of circumpolar origin and high14C/C waters of northern Atlantic origin. When thismixing is taken into account the residence time ofwater in the Atlantic is considerably decreased and theflux of new North Atlantic deep water necessary tomaintain it is correspondingly increased (table 15.5).

Table 15.5 Evolution of Calculations of the Apparent 14C Age of a NADW Sample with a A14C of -105%o

A14C values

Warm New A due A duesurface no. comp. New to mixing to in situ Apparentawater water AABW with AABW aging age

Broecker et al. -28 (-28) -155 0 -77 6751960

Broecker and Li -40 (-40) -155 -12 -53 4701970

Stuiver (GEOSECS) - -70 -160 -10 -25 2251976

Broecker (GEOSECS) - -70 -160 -26 -9 801979

a. t = 8200 in 1 - (A1 4Ciniti a) X 10-31 - (A14Cinitiai + Aagin) X 10- 3 '

450W. S. Broecker

--

(15.11A) Figure I5.II Maps showing the distribution of 14C along the

4000-m horizon in the Atlantic A) and Pacific (B) Oceans.The analyses were made by Ostlund at the University ofMiami and Stuiver at the University of Washington as part ofthe GEOSECS program.

45 IGeochemical Tracers and Ocean Circulation

.....- ·· · · _~~~~~~~~~~~~~~~~~~~~ 1 ~......--.. ...

(15.11B)

452W. S. Broecker

_ ___ �XI_ I

-...r

_I

�____I__II ��

Before going into this matter further, it is appropriateto review the evidence leading to the conclusion thatthe back mixing of Antarctic deep water is the majorcause for the 14 C/C gradient observed in the Atlantic.Confusion on this point has arisen because of the am-biguity in estimates based on conventional tempera-ture-salinity (-S) analysis of the fractions of variouscontributing water types. There are three such north-ern water types: deep water formed by winter coolingin the Labrador Sea (LSW); water spilling over the sillconnecting Iceland to Greenland (DSW); and waterwhich ultimately comes from overflow across the sillconnecting Iceland with the British Isles but enters thewestern basin through the Gibbs Fracture Zone. Asshown in table 15.6, the traditional O-S approach is notparticularly sensitive, because the 0 and S ranges forthe three contributing northern water types are largecompared to the differences from the dominant south-ern type, i.e., the Antarctic Bottom Water (AABW).Temperature and salinity could be used only if therelative proportions of the northern-source waters oneto another were known. They are not (see chapters 1and 2).

By contrast the SiO2 and the "NO" contents of thethree northern water types are almost identical and arevery different from that for AABW (see figure 15.12).Thus, were either of these properties conservative, wewould have a means of determining the fractions ofhigh-1 4C northern and low-'4C southern waters in anygiven sample from within the mixing zone. One testof conservation is a plot of SiO2 versus "NO". Linearitywould require either conservation of both properties ora fortuitous correlation between the true "NO" coef-ficient and the amount of SiO2 added to any givensample through the dissolution particles. As shown infigure 15.12 the relationship is remarkably linear ex-cept for waters in the southern eastern basin.

If this linearity is (as I believe it can convincingly beshown to be) a demonstration of near conservancy of"NO" and SiO2, then the fraction of southern compo-nent in each sample of deep Atlantic water collectedby the GEOSECS program can be determined eitherfrom its SiO2 or "NO" content (these two independentestimates generally agree to 0.03). The plot of A14Cversus fraction of southern component so obtained isgiven in figure 15.13. The dominance of mixing isclearly demonstrated. The deviation attributable to ra-diodecay averages only 9%0o (the measurement error ona given sample averages 4%bo).

The geographic and depth patterns of the residualanomalies (i.e., those produced by radiodecay withinthe Atlantic) are shown in figure 15.14. The only pro-nounced feature is the larger anomaly for the easternthan for the western basin. As shown in table 15.5,estimates of the residence time of waters in the deepwestern Atlantic have dramatically decreased withtime. Part of this decrease is accounted for by the re-alization that water descending in the deep northernAtlantic has a lower 14 C/C ratio than ambient Atlanticsurface water. The rest is due to the increase in theestimates for the role of AABW back mixing.

As I have shown in a separate publication (Broecker,1979), if an advective model is used to reproduce boththe distribution of northern and southern componentsin the western Atlantic and the distribution of theresidual (i.e., in situ decay) anomalies, then a flux of20 x 106 m3 s - 1 of northern component (some mixtureof LSW, GFZW and DSW) and 10 x 106 m3 s- l of south-ern component (AABW) are needed. As the deep por-tions of the eastern basin are known to be ventilatedfrom the western basin (via fracture zones), no extraflux of northern component water is needed for itsventilation (part of the water exiting the western basindoes so through the Romanche and possibly other frac-

Table 15.6 Characteristics of the Contributors of the NADW Complex

0 S "NO" SiO2 &'4CWater type (C) (%bo) (/,M kg- ) (M kg-') (%o)

Labrador Sea 3.3 34.94 430 12 -69Deep WaterDenmark Straits 1.5 34.91 427 10 -69Overflow WaterGibbs Fracture 2.8 35.00 429 13 -63Zone WaterMediterranean Sea 12 36.4 300 8 -- 60Overflow WaterCircumpolar 0.0 34.70 511 125 -163Bottom WaterCircumpolar 2.2 34.63 472 85 -147Intermediate Water


'a

o0Z

470

I450

430

410A..

(15.12A) Si 2 (m/kg) 0 20 40 60(15.12A~~ si o, (pm/tg Si 02 (m/kg)

(15.12C)

A=,.qrU

2 450

Z430

410

MOW ilU2 ktLm/ yJ(15.12B)

a

Figure I5.I2 Plots of "NO" versus SiO2 for all the samplestaken by the GEOSECS program below 1500 m in the AtlanticOcean. In the western basin (A) the entire range of composi-tions between northern (DSW, LSW, and GFZW) and southern(AABW) component water is seen consistent with the two-end-member hypothesis. No change in slope is seen at theTwo Degree Discontinuity (Broecker, Takahashi, and Li,1976). In the northeastern basin (B) those stations nearest tothe Straits of Gibraltar show a deviation toward low "NO"value for any given SiO2 content caused by the admixing oflow "NO" Mediterranean water (see table 15.6). Deep watersin the southeastern basin (C) show a deviation toward highsilica at any given "NO" value. This presumably reflects insitu production of silica through the dissolution of particulatematter.

FRACTION SOUTHERN COMPONENT

Figure I 5. 3 A'4C for western basin deep waters as a functionof the fraction of southern component (F.S.C., as estimatedfrom the "NO"-SiO 2 relationship). Were mixing alone re-sponsible for the trend, then the points should lie along theline joining A14C = -160%o, F.S.C. = 1.0, and A14C = -70%o,F.S.C. = 0.0. Deviations below this line are presumably ameasure of the loss of radiocarbon within the mixing zone byradioactive decay. The radiocarbon measurements were madeby Stuiver of the University of Washington and Ostlund ofthe University of Miami.

454W. S. Broecker

T-rIa_LdC

31 117 115

TI-a-LIi0

I r-TI nr--

(15.14A) LONGITUDE (W) (15.14B)

48 III

E

I-F-

C

EII0LLUC

(15.14C) LATITUDE (15.14D) LuI'I I UUt - VV)

Figure I5.I4 Four sections showing the geographic and dedistribution of the 14C anomalies attributable to radioacldecay within the deep Atlantic. The contours show the ftion of southern component (i.e., AABW) within the dAtlantic.


. .-- I.-I

ture zones which cut the Mid-Atlantic Ridge). Thequestion how much more water is needed to ventilatethe eastern basin above the ridge crest is not so easilyanswered. I guessed that 10 x 106 m3 s-' more of north-ern component is required. If so, then the total flux ofnorthern component is 30 x 106 m 3 s-.

The volume of the deep sea (below 1500 m) dividedby 30 x 106 m3 s-' gives 900 years. Thus, were NorthAtlantic Deep Water the only source of carbon isotopeventilation, the mean age of deep-sea carbon with re-spect to that entering with new deep water formed inthe northern Atlantic would be 900 years. Starting withA14C values of - 7 0%o, this would give a mean A14C fordeep-sea carbon of -167%o (a value close to the ob-served mean).

This does not imply that ventilation of the deep seaby waters descending around the continent of Antarc-tica is negligible. Rather, as shown by Weiss et al.(1979), these waters exhaust heat without substantiallychanging their 14C value. Thus if, as commonly esti-mated, the flux of new deep waters formed around theperimeter of Antarctic is in the range 10 to 40 x106 m3 s- (Gordon, 1975b; Gill, 1973; Killworth, 1974,1977; Carmack, 1977), then the water ventilation timeof the deep sea is probably considerably less than 900years.

While these calculations are preliminary and subjectto many criticisms, they do serve to indicate that de-rivation of water fluxes from '4C data is far from astraightforward exercise. The conventional 14 C-resi-dence time (based on box models) provides only anupper limit on the water-residence time. As tempera-ture is probably the property most rapidly equilibratedwith the atmosphere and carbon isotopes the mostslowly equilibrated, the renewal time for other sub-stances (CO2, NO3 , SiO2, . . .) will lie in between thewater and carbon-isotope renewal times.

15.7 Ventilation of the Main Oceanic Thermocline

The GEOSECS program has provided for the first timea reasonably detailed coverage of the distribution inthe ocean of the 14C and 3H produced by nuclear testing.These distributions emphasize the importance of twophenomena evident but perhaps not fully appreciatedthrough more conventional oceanographic observa-tions. First they suggest an upwelling flux in the equa-torial ocean comparable on a global scale to the flux ofnewly formed deep water (Broecker, Peng, and Stuiver,1978). Second, they illustrate the importance of ther-mocline fronts located near lats. 15°N and 15°S(Broecker and Ostlund, 1979).

Strong upwelling in the equatorial ocean (or in theeastern boundary regions feeding into the equatorialzone) is suggested by the distribution of bomb-pro-duced "4C. The equatorial zone is characterized by low

surface-water bomb 14C/C ratios and shallow penetra-tion of bomb 14C (relative to the adjacent temperategyres). If the gas-invasion rate (from the atmosphere)into equatorial waters is comparable to that into tem-perate waters (as I believe it must be), then the lowinventory of bomb 4C in the equatorial zone can onlybe maintained by the input to the equatorial zone ofwater deficient in bomb 14C. The only source of suchwater is upwelling from depths of at least 500 m. Asthe bomb 14C distribution within the equatorial zoneis quite homogeneous along isopycnal surfaces (figure15.15), this isotope is not a sensitive indicator of theplace at which upwelling occurs. However, by combin-ing 14C data with data for tracers like Pco,,, PN2 , NO3,NO,, which have shorter surface-water response timesthan does 4C, it may be possible to get at the entrypattern of upwelled water. Furthermore, time series of14C (as measured directly on water samples and as re-constructed through measurements of coral-growthrings) may reveal fluctuations in the rate of upwelling.

The distributions of 9Sr and 3H in the Atlantic dem-onstrate a front near 15°N, because the water-columninventory of these isotopes drops by an order of mag-nitude near that latitude. As shown in figure 15.16, theconcentration of tritium along any given isopycnal alsodrops by an order of magnitude from north to southacross this front (Broecker and Ostlund, 1979). Onereason for this difference is that the bulk of the 90Srand 3H "fallout" occurred to the north of the equator.This asymmetry between hemispheres does not ex-plain, however, why the boundary is so abrupt and whyit lies near 15°N.

An answer to this dilemma comes from the equato-rial upwelling inferred in the zone over which the ther-mocline is thin and shallow (i.e., 15°N to 150 S). As thisupwelling is driven by the poleward divergence of equa-torial surface water, any 3H or 90Sr reaching the equa-torial zone by fallout or by leakage across the 15°Nfront will be pulled back to the surface and carriedpoleward across the front. Thus upwelling accounts forboth the location of the fallout front and its sharpness.

As shown by Broecker et al. (1978), the distributionof 4C in the equatorial zone yields a ratio of the fluxof upwelled water to the invasion rate of CO ,. Sincethe latter has been independently determined, the up-welling flux can be obtained. Using this approach weobtained a flux of about 15 x 106 m 3 s- ' for the Atlanticequatorial zone. A preliminary analysis of the PacificOcean 14C data suggests that the upwelling flux thereis about 30 x 106 m3 s-'. As mentioned above, no firmstatement can yet be made as to where this upwellingoccurs, for a high rate of lateral mixing obscures itsorigin.

456w. s. Broecker

__I ____

Co

CD

o0P

(15.15A) H (TU)

15.8 Formation of Deep Waters

As pointed out, there are serious difficulties in usingthe distribution of natural radiocarbon to obtain deep-water formation rates. In the Antarctic this stems fromthe fact that the 14C clock is not reset during the for-mation process. In the deep Atlantic the anomaly dueto radiodecay is small compared to the anomaly due tomixing. Also, there is no simple way to separate thecontributions of the three (or possibly more) northernwater types contributing to North Atlantic DeepWater. It behooves us therefore to seek other geochem-ical methods to gauge these fluxes. One obvious ap-proach is to use the substances added to the ocean asa by-product of man's activities to trace recentlyformed deep water. The GEOSECS 3H section alongthe western basin in the Atlantic (figure 15.17) showsthat the pathway followed by the newly formed waterwas in 1972 clearly "stained" with fallout products. Sothe potential is there. At a minimum, surveys spacedat decade intervals would clearly demonstrate the man-ner and rate at which the front of tritiated water ispushing its way into the western basin of the Atlantic.

The problem of inverting distributions of these tran-sient tracers into fluxes is a difficult one. The geometryof the system is complex. Simple box-model represen-tations are bound to give misleading answers. I do notmean to imply by this that the situation is hopeless.Rather the job will be a long one, requiring imagina-tion, diligence, and excellent measurements.

It will be helpful, moreover, to explore what limi-tations the relationships among these tracers place onthe types of dynamic models that might be employed.An example drawn from the GEOSECS observationswill serve to illustrate this point. Despite quite differ-

0

CD

0(15.15B) Z4c(/o

Figure I5.I5 Plots of tritium (A) and of radiocarbon (B) v.potential temperature in the equatorial thermocline of theAtlantic Ocean. The tritium data give the impression thattritium-bearing surface water is being mixed down into tri-tium-free water of potential temperature about 8C. If 8°Cwater is taken to be the source for upwelling, then the 4Cdata can be interpreted as a mixture of surface water of A14C =100 °bo and 8°C water with a A14C value of about -90%o.


~n

',%O

2

;

I

C

Figure I5. 6 Tritium distribution along the isopycnal of (o =26.80 in the Atlantic Ocean. The 15°N thermocline frontconstitutes a very pronounced boundary between high-tritiumwaters to the north and low-tritium waters to the south.

458W. S. Broecker

l

ent input mechanisms, bomb-produced 14 C and 3Hshow a very high degree of covariance in the northernAtlantic (figure 15.18). The samples included in thisdiagram cover the entire range of depth, geographicallocation, and water type. Modeling shows that differentcombinations of advection (away from the source re-gion) and diffusion predict a great variety of 4C-3Htrends. Thus the combined tracer fields can perhapstell us not only the pathways followed but also some-thing of the dynamics of the water movement.

15.9 Vertical Mixing Rates

Although there is no way to demonstrate how the iso-topes we study penetrate into the interior of the ocean,we can still use their distributions to place upper limitson the rate of vertical mixing in the sea. This can bedone by matching any given depth profile to a one-dimensional model, as has been carried out for profilesof 222Rn and 228Ra in the deep sea, and for profiles of

GEOSECSSTA 74 68 67 64 60

bomb 3H and 14C in the main oceanic thermocline, asshown in figure 15.19. The apparent vertical diffusivi-ties obtained in this way show an inverse correlationwith density gradient (Sarmiento et al., 1976; Hoffertand Broecker, 1978). Quay, Broecker, Hesslein, andSchindler (1980), using tritium injections into twosmall lakes in Canada, were able to obtain verticaldiffusivities for much higher density gradients. As thetracers were allowed to spread laterally over the entirebreadth of each lake, these results are true rather thanapparent vertical diffusivities. As shown in figure15.19, the tracer results from the lakes fall close to theextension of the oceanic curve.

It is difficult, however, to assess the significanceof these results. Sarmiento and Rooth (1980) haveshown that the bottom 222Rn results could be ex-plained entirely by mixing along isopycnals. Sar-miento (1978) has shown this to be the case for 228Rain the deep sea as well. It has long been suspected byoceanographers that the main thermocline of the ocean

TRITIUM IN THE WESTERN ATLANTIC 19758 56 54 49 48 40 39 37 34333231 30 29 27 3 5 11 14 15 16 17

2-73

L. . 7> ' -.i II

0.2

L;r5i'JD

>10TU

0.2,0.4,--- 0.6. and

0.8 TU

1.0

2.0

3.0

4.0

5.0

500S 40S 30°S 20°S 10°S EO 12N 20N 30N 40°N 5dN 60°N 70°N

LATITUDE

Figure I 5.7 Tritium section along the western basin of theNorth Atlantic, prepared by Ostlund of the University of

Miami, who made all the tritium measurements for the GEO-SECS program.


0.2

1.0

KM

2.0

3.0

4.0

5.0

<0.2

- - - s -

<I

g

TRITIUM (TU.)(15.18A)

Jo1

TRITIUM (T.U.))15.18B)

Figure IS.I8 Tritium (in T.U.) vs. radiocarbon (in A14C) forwater samples (from all depths) in the Norwegian Sea (A) andnorthern Atlantic (B}. The solid curves show the results pre-dicted by a one-dimensional diffusion model (the diffusioncoefficients are in the units cm2 s-').

Density Gradient (1 lgm/cm 4 )2 4

N

NESYC,

rC1a)a

NZ(s- 2)

Figure 5.I9 Apparent eddy diffusion coefficients for verticalmixing as measured by 22Rn and 22Ra in the deep sea (Sar-miento et al., 1976), 3H in the pycnocline of the NorwegianSea (Hoffert and Broecker, 1978) and bomb 14C in the SargassoSea (Broecker et al., 1978) as a function of density gradient.Also shown are the results of Quay et al. (1980), obtainedby tritium injections into small lakes.

is ventilated mainly along isopycnal surfaces. Certainlythe distributions of bomb-produced 3H and '4C in themain oceanic thermocline could be explained by thisprocess. The 7Be profile shown above could be inter-preted as an artifact of winter convection. Perhaps anargument relating to the angle between isopycnal sur-faces and the tracer source (sea surface or floor) can bemade as an alternative to the correlation between ap-parent vertical diffusivity and density gradient.

In my estimation the only way in which the relativeimportance of vertical and horizontal mixing can beestablished is to perform tracer injections at variouspoints in the ocean interior. Such experiments (using3 He as a tracer) would be comparable in cost to that forthe large ocean science projects currently underway.The technology needed is available. By following a 3Heor tritium addition to an isopycnal horizon for severalyears, it would surely be possible to obtain a betterestimate than we now have for the relative magnitudesof mixing along and perpendicular to isopycnal hori-zons. A knowledge of the relative importance of theseprocesses is critical to the development of adequatemodels for the interpretation of radioisotope tracerdata.

460W. S. Broecker

1_� ___ _ __ ______I _ _

I -_6

Tracers have always been an important adjunct to Tracers ... · PDF fileI5 Geochemical Tracers...

Documents

Transcript of Tracers have always been an important adjunct to Tracers ... · PDF fileI5 Geochemical Tracers...