Carbon dioxide in the atmosphere over the ... - Wiley Online Library

264 downloads 0 Views 870KB Size Report
Feb 20, 1988 - exchange of metabolic CO• with vegetation and soils. selection of the data ...... exchange processes, ince O• and CO2 are negatively correlated.
JOURNAL

OF GEOPHYSICAL

RESEARCH,

VOL.

93, NO. D2, PAGES

1377-1387, FEBRUARY

20, 1988

CarbonDioxide in the AtmosphereOver the AmazonBasin STEVEN C. WOFSY

Divisionof AppliedSciences andDepartmentof Earth andPlanetarySciences Harvard University,Cambridge,Massachusetts ROBERT C. HARRISS

NASALangleyResearchCenter,Hampton,Virginia WARREN

A. KAPLAN

Divisionof AppliedSciences andDepartment ofEarthandPlanetarySciences Harvard University,Cambridge,Massachusetts

Thecycleof atmospheric CO2 in theAmazonBasinwasstudied usingmeasured verticalprofilesof CO2 concentrations in thecanopyandaloftanddirectmeasurements of CO2 emissions fromsoils.Theresultsprovide detailedinformationon daily exchanges of air betweenthe tropicalforest(0-30 m) andthe atmospheric boundary layer(30-2000m). Theforestwasa netsource of CO2 at nightanda sinkduringtheday. Highconcentrations of CO2 (380-400ppm)accumulated in theforestat nightandweretransported intothenascent mixedlayerin themorning.Between1000and1200LT, CO2in themixedlayerdeclined to concentrations 4-6 ppmlowerthanfreetropospheric valuesandremainedlow for the restof the afternoon.The afternoonmixed layerbecame isolated at night,andthelowvalues of CO2werepreserved untila newmixedlayerdeveloped the nextmorning.Entrainment of air intothemixedlayerappeared to be a one-wayprocess duringtheearlystages of mixedlayergrowth.Theratefor uptakeof atmospheric CO2 by forests(soilandcanopy) nearmiddaywas

estimated tobe9(+_4) kgC ha-i hr-i, andmean uptake overthesunlit period wasabout 2.8(+_1.2) kgC ha-1

hr-1.Forest soils emitted CO2atamean rate of1.8(_+0.2) kgCha-1hr-1. Theatmospheric CO2cycle over

wetlands wasweaker thanoverforests, withdaytime uptake rates ofapproximately 1.6kgC ha-• hr-'. Rivers remained netsources of CO2 throughout theday.

INTRODUCTION

estimationof rates for exchangeof CO2 and other tracegases betweenthe forestandthe atmosphere.

Thedaily cycle ofatmospheric CO2 inforested regions reflectsWefirstdescribe experimental methods andthen present a exchange of metabolic CO• withvegetation andsoils.selection ofthedata obtained. Observed variations ofCO•intime Photosynthesis provides anetsink foratmospheric CO•during andspace arecompared withdatafromthelidarinstrument

daylight hours; respiration and decay represent netsources ofCO•[Browell etal.,1987] and with data forOa[Gregory etal.,1987]. atnight. Consequently, atmospheric CO2 over forests usually Atmospheric CO•concentrations over forests, weftands and rivers declines during the day and increases atnight. areexamined, revealing characteristic differences. Finally, Ecologists have exploited atmospheric variations ofCO•to preliminary estimates are derived formean daily exchange rates of study metabolic rates inavariety ofenvironments. Pioneering CO•between theatmosphere and theforests and wetlands ofthe workbyOdum etal. [1970] andLemon etal. [1970] provided Amazon Basin.

basic informationon the CO• cycle in tropicalforests. These studieswere hampered,however,by the absenceof information

about CO2variations above thetreetops. Desjardins etal. [1982]

EXPERIMENTAL METHODS

Desjardins, [1985] andAlvo etal.[1984] measured CO•fluxes Gasfluxes across thesoil-air interface were determined by over Canadianforestsby eddy correlationusing an airborne samplingair in the headspace of weldedaluminumchambers(10

sensor. They observed that therate ofCO•uptake byvegetation cmhigh x20cmwide x40cmlong, placed 1-2cmdeep into the increased linearlywiththeincidentsolarflux. soilforshort periods (20-40 rain).Methods andanalysis Thepresent paper examines thecycle ofCO•inandover a procedures were thesame asthose described byKeller etal. tropical forest, aspartof theABLE2Amission toBrazil's [1986]. Pressure equilibrium between theenvironment andthe Amazon Basin [Harriss etal.,1987]. Anextensive setofdatainterior ofthechambers wasmaintained through a 5-cm-long was acquired from anairborne CO•sensor, from vertical profiles hypodermic needle with0.3mmID. Theinternal volume ofa obtained onamast extending through thecanopy inthetropical chamber was7.2L when eraplaced 1 cmdeep or6.4L when forest, and from direct fluxmeasurements atthesoil-atmosphere inserted 2 cmdeep. Thedeeper eraplacement was used inmost interface. We show that there is a predictablecycle of cases,except where soils were heavy clay. The static open

atmospheric CO2 intheforest environment and that CO•isaverychamber method waschosen asmost appropriate toobtain usefultracerfor transportprocesses in the lowestlayersof the representativeresultsin the forest environmentwhere air motions

atmosphere.Carefulanalysisof the atmospheric CO2cycleallows are usually very weak at the soil surface [Odum et al., 1970; Copyright1988by theAmericanGeophysical Union. Papernumber7D0485, 0148-0227/87/007D-D485 $05.00

Hutchinsonet al., 1981]. Samples of 40-mL volume were slowly removed from the

chamberat approximately5-rain intervalsusing50-mE (Propper Trophy)syringeswith matchedgroundglassbarrelsandplungers. 1377

1378

WOFSYET AL:CARBONDIOXIDEOVERTHEAMAZON

GTE/ABLE

2A

Chart recorder and

data logger

WAI LOG RATIO AMP INLET

e xhoust

2-st selection

solenoid valve

valve

ice

6ram gloss

both

beads

liquid water trap

HARVARDAIRBORNECO•. SENSOR Fig. 1. Schematic diagramof airborneCO•analyzer.

The fu-stsamplewas taken less than 1 rain after cmplaccmcnt-introduced. The air then flowed through a microprocessor-

Sampleswere analyzedin the field, usinga gaschromatograph controlled solenoid valveintoa glasstrapfilledwith6-ramglass with a thermalconductivitydetector. Concentrations in dry air beads,maintainedat 0øCin an ice bath,andfinally intoa modified were determinedby comparisonwith commerciallyprepared BeckmanIndustrialmodel 865 nondispersed infrar• analyzer. standards of CO2 in air and/orin nitrogen(ScottEnvironmentalPressure in the instrument wassensedby an MKS 220B absolute

Corporation,Plumsteadville,PA). We assignedabsolutepressure gaugeandmaintained constant at 400 tort, independent concentrations to each commercialstandardby comparingto of altitude,by the microprocessor (MKS 250 B valvecontroller). standard referencematerials(SRM) from theNationalBureauof The 90% response timefor the instrument andthe delaytimein Standards. thesamplingline werebothabout3 s. Fluxes acrossthe air-soil interfacewere calculatedusing a Humidityin thesamplecell wasmaintainedat 46 tortby theice

linearleastsquares fit to thefouror morepointsin thetimeseries bath. The instrumentwas equipped with a narrow-pass of concentrations.Observed concentrationsusually increased interferencefilter to reducethe sensitivityto H20 vapor. Total linearlywith time exceptin the caseof very largefluxes,where responseto HzO undertheseconditionswas equivalentto a CCh

concentrations approached a steadyvalueat longtimes.In these level of -1 ppm, maintainedconstantby the experimental casesthe flux was calculatedfrom the slopeof a line drawn arrangement to betterthan :L-0.1ppm (equivalentCOz) for both between the first two data points. Occasionaloutliers were ambientair andstandard mixtures.Pfima• standards at 336.2 and

rejectedby comparing residualvariances for linesfittedto three 365.2 ppmCO• in air wereobtainedfrom ScottEnvironmental time points. Results of a few emplacements were rejected Corporation.We measured theserelativeto a set of four SRM altogether because concentrations at the first timepointdiffered CChstandards. Resultsagreedwith themanufacturer's valuesto

markedly fromambient levels,indicating disturbance of thesite betterthanx'-0.2 ppm. duringemplacement or incomplete flushing of theboxpriorto the Instrument precision wasaboutx'-0.5ppmfor a 10-saverage m startof theexperiment.About10%of thefielddatawererejected level flight (:L-0.7 ppm in the strongest fair-weather turbulence).

using theabove criteria. Minimum detectable fluxes were3 x 10• During spirals thetiltof theaircraft induced a shiftof 1-2ppm,a molecules cm-z s-• (2g C ha-•h-•). generic problem forthistypeof instrument. Theinfluence of tilt Vertical profiles of CCh in the forest environmentwere was effectivelyremovedfrom the data by frequentcalibration, obtainedby filling previouslyevacuated 300-mi.,glassflasksto a exceptin instances wheretheaircraftexecutedunscheduled turns. pressureof 2 atm, usinga Teflon diaphrampump. The samples Theseinfrequentmaneuvers werecarefullynotedin our log and wereobtainedfrom 0 to 28 m by raisinga 30-m-longfluorinated the associated data were rejected.Becauseof the rapidrate of ethylenepropylene(FEP) Tefloninlet tube,1/4-.inch OD, on the descent (or ascent) duringspiralsandtherequirement for frequent

aluminum mastdescribed by Kaplanet al. [thisissue].Vertical calibration, only one standard was routinelyused.Postflight profiles wereobtained duringtwo24-hour cycles onJuly26-27 analysis of thetwo-point calibrations revealed a slowgaindrift andAugust2-3, 1985. Samples werereturned to Harvardand accounting for errorstypically d:0.5ppmor,in a fewcases, d:l.0 analyzed for COz,usinga gaschromatograph with ultrasonic ppm.Thetimeconstant forthisdriftwasabout1 hour,whereas phase-shift detector.The samples werealsoanalyzed for the thetimerequired for anindividual profilewasusually about15 •3C/zCratio;in COzby P. Quay(University of Washington, min.Hencetherewasnegligible effectonthemeasured variations Seattle)andfor CH4concentration, usinggaschromatography of CO2concentration withaltitude, although variations between with flameionizationdetector.

flightsor duringlongflightsmayhavebeenaffected.A particular

Theairborne COzanalyzer is shown schematically in Figure1. anomaly wasrecorded in flights 6 and7, whichshowmeanCOz Air wasdrawn through theinstrument at3-61rain -• bya dual-levelsabout 2 ppmhigher thanotherflights; a problem withthe headAir Cadetpump.Air enteredan inletof 1/2-inchOD zerogasissuspected inthiscase. aluminumtubing, mounted 15 cm above the fuselageof the

Data were recordedat 8 Hz, usinga Hewlett-Packard349'/A

NASA Electra, which was connectedvia 1/2-inch OD data acquisitionunit. coupledto an HP-85 computerand polypropylene tubingto a valve where standards could be associated disk drive. Data were averagedover 10 s during

GTE/ABLE2A

WOFSY ETAL:CARBON DIOXIDE OVER TimAMAZON

1379

al. [1980] which indicatethat respirationand decomposition of motsaredominantfactorsin softCO: production (seealso,Keller

! I I ! t I ! ! I ! .i !

et al., [1986]). 3

-

'E 3.0--

-2--':

_

o

A summaryof mean COz fluxes measuredfrom terra firme forestsoftsnearManaus,Brazil,is givenin Table 1. The standard deviationfor eachdata set is relativelylarge,usually30-50% of the mean value, and 90% confidenceintervalsare typically 1020% of mean values, reflecting intrinsic patchinessin the distributions[Keller et al., 1986]. There is an indicationfrom

Table 1 thatsoftCOz fluxesin the wet season (November-June) maybe lowerthanfluxesearlyin the dry season (July-August). Thisdifferen9e couldreflectseveralenvironmental factors.The

x

o• -I

mean incident solar flux to the forest is lower during the wet season,which couldlead to reducedroot metabolismby reducing

overall community photosynthesis. Production ofCOzfromdecay

o

Normol

of organicmattercouldbeinhibitedby lowerratesfor diffusionof Oz into the soB. Decayof soblitter maypeakin the'earlydry

distribution

season because of the enhanced fall of leaves at that time.

S - 0.94

0 .Ol

i .05

I

I .20

I

I I i I .40 .60

I .80

I

I .95

I

CO2WrIItlN THEFOREST CANOPY

.99

FR ACT I LE

The vertical distributionof CO2 from the soil surfaceto 27 m Fig. 2. Probabilitydistributionof soil CO2fluxes. The straightline height,(3 m abovethe mainforestcanopy)wasstudied during represents a normaldistribution with the samemean and standard deviationasthe datasample.

several24- hourperiods.The primarypurposeof thesestudies was to investigatethe use of CO2 as a tracerfor exchange processes betweenair within the canopyand the atmosphere aboveand to define the lower boundaryconditionfor the CO2

postflight processing, except forcalibrations where thesteady profiles obtained bytheaircraft. segment wasselected byhand from therawdata. A high- Figures 3and 4 show vertical profiles ofCO2 obtained onJuly performance chart recorder (Soltee) wasused asbackup data26and August 2-3.Weusually observed accumulation ofCO:at

system. A fewprofiles were notrecorded bythecomputer andnight intheforest layer, followed bydecline toanafternoon were retrieved fromthechart record. Detailed analysis showed minimum (Figures 3 and4). Highconcentrations ofCO2were thechartrecord to bereliable to better than0.1ppmat all observed between 0 and6 m dayandnight,reflecting the frequencies thatcould beexamined, i.e.,upto0.3Hz. influence ofsoilemissions. Thepattern issimilar toobservations byLemonet al. [1970]in a tropicalrainforestin CostaRica.

CO2DISTRmLrHONS: RESULTS ANDDiSCUSSiON

Nearlyuniform CO2concentrations wereobserved between 6 and 30 m (Figures3 and 4), indicatingsignificantturbulent

Wewillfirst discuss thecharacteristics ofsoilCO2 emissions in mixing. Vertical mixing isexpected inthedaytime, because of theDucke forest near Manaus, Brazil [Harriss etal.,thisissue], buoyancy generated bysolar heating ortoimpinging gusts from followed byananalysis offactors influencing theCO2 distribution above [Denmead andBradley, 1985; Finnigan, 1985; Grant etal., in the surfacelayer of the atmosphere from the ground to 1986]. Nighttimemixingis somewhat unexpected. It probably

immediately above theforest canopy. Then wewilldiscuss the results fromradiative andevapoi'ative cooling of theupper large-scale distribution of CO2in theplanetary boundary layerand canopy,whichcouldcreatenegativebuoyancyat the top of the

middle troposphere over theAmazon Basin.

forest layer [Shuuteworth etal.,1984].

SOIL EMISSIONS

Measurements ofCO2 fluxfrom forest soils aresummarized in TABLE 1.Mean CO: Fluxes From Soils inTropical Forests

Figure 2.The mean net CO2 flux from these softs was 2.45 x10 t4

Near Manam, Brazil.

molecules cm -2s-• (1.76kgC ha-• t•'-•),consistent withotherSite

data from these environs [Keller et al., 1986; Goreau and

CO 2Flux n

DeMello,.1985].Thestandard deviation onthemeasurements was Ducke

25:L9

36

Date July1985

References ThisStudy

9.4 x 10•a cm-2 s-• and the 90% confidence interval for the mean

was +2.6 x10•acm -2s-t(+10%). Figure 2compares the observed Ducke 29+13 12 July-August 1984

1

distribution o.ffluxes toanormal distribution withthesame meanEsteio

21+11

4

July-August 1984

1

20

July-August 1984

1

andvariance.The experimental distribution appearsto be close,

indicating thatthemeanandtheconfidence interval should be Fazenda" 27+13 robust estimators.

The meanannualinputof canopylitterto soilsin Duckeforest

Fazenda"

is approximately 6 t dry organicmatter,ha-•yr -• [Klingeand Fazenda" Rodrigues, 1968]. If we assume 0.5 g C per gramof organic

16

7

December 1983

2

19

9

March 1984

2

matter [Whittaker andLikens, 1972], thecarbon source tosoils Fluxes are given as10 • molecules cm -2s-i;Reference 1,Goreau and DeMello [1986]; Reference2, Keller et al. [1986]. Rangesrefer to one from canopy litter isonly 300gCm-2yr-• (0.34 kgCha -• hr-•), standard deviation, nindicates number ofsamples. Soil type isyellow

approximately 19% of thesoilCO2flux. Thisresultagreeswith oxisol inallcases. previousstudies (for example,seeSchlesinger [ 1977];Medinaet "Fazenda PortoAlegro, 75ionNEofManaus.

1380

WOFSYET AL: CARBONDIOXIDEOVERTI-IEAMAZON Mean

-

26July

[t'l • • 24• •o- ••,•,,,•" ,'

--

CO2DISTRIBUTIONS IN TIlE PLANETARY BOUNDARY LAYER

Daytime conditions.The mixed layer developsrapidly after sunriseas solar heating producesbouyancyat the top of the canopy (see Garstang et al., [this issue], for a complete discussion).Since the surfacelayer containshigh concentrations of COz, a sharp COz gradientdevelopswhich unambiguously delineatesthe growingmixedlayer. Figures6a and 6b showthe first appearance of a COz-richmixedlayeroverforestsat 150 m altitude

oF ,

2A

interruptedat 2000 LT (Figure 5). Hence,there clearly were importantexchangesof air at night betweenin the forestcanopy and higher altitudes during the August experiment. Such exchangesappearto havebeenlesssignificantduringJuly, when muchhigherconcentrations of COz andCH4 wereobservedin the early morning.

value

I-oo 0700 28•--x% J1400 • •

GTE/ABLE

between

0800

and 0900

LT.

The

remnants

of

the

detachedrelic layer may be seenjust abovethe new mixedlayer. The top of the developingmixedlayer was observedto moveup

,

rapidly,at 5-10 cm s-•, erodingthe low-COzrelic layerabove (Figure6c). Growth of the mixed layer was somewhatretarded

CO2 (ppm)

Fig.3. Distribution ofCO•withheight onJuly26,1985, intheforest at overwetlands, asillustrated in Figure6d. Herethereliclayer DuckeReserve.The maincanopytop is at 24-m height.The mean from the previousday was undisturbed at 0840 LT in contrastto

concentration inthefreetroposhere was344.6ppm,denoted byanarrow profiles in Figures 6a-6c. at the top of the figure. The insert comparesprofilesnear sunriseon August2 and August3, 1985.

The compositionof the relic layer appearedto remain unchangedjust above the top of the new mixed layer. The gradient was very sharp and there was little observable

Sharp decrease inCO2concentrations wasobserved attheupperentrainment of high-CO2 airfromthemixed layerintotherelic canopylevelduringthe nightof August2-3 (Figure5). layer.Duringtheperiod of rapidrisein mixedlayerheight, Concentratiom ofCOzatthecanopy topdeclined tovalues typicaltransport appears tobeone-way, entraining airfromthereliclayer oftheafternoon, withaconcentration lessthan340ppmobserved aloftintothedeveloping mixed layer. at 28 m at 0100LT. As shown later,formation of thenocturnal Themorning profileof CO2overtheocean wasverydifferent

inversion in lateafternoon separates theatmosphere above 35m fromprofiles obtained overland(Figure 6e),showing, atmost, a fromtheforest canopy layer.Airfromtheafternoon mixed layer,2-ppm enhancement atlowaltitudes. Themixed layeroverthe with3-6ppmle• COzthan themean values observed aloft(344.6ocean does notdecay atnight, andthere isnoreliclayer depleted ppm,Table2),is stranded above thecanopy. Subsidence of air inCO2between 1 and2 kin.Theoceanic source ismuch weaker fromthis"relic" mixed layerintotheforest canopy appears to thanthenocturnal terrestrial source, andthesmallobserved have occurred atoursiteonthenightofAugust 2, staffing aboutenhancement mayhave been duetoadvection ofterrestrial CO2by 2000LT. Thesame phenomenon isalsoevident indataforCH4 thelandbreeze. fromthesample flasks, wherenocturnal accumulation was A typical progression ofCO2profiles overforests isshown in Figure7. Thesedata(Flight 6), representa traverseof the forests in the easternAmazon Basin betweenManaus and Rio Xingu

i

•"Q\

?'

1

!

i

,

!

!

i



i

I

i

i

i

i

+

|

20- •

3Aug 0600 -

1.68 +

CH4

_

_

(ppm)I.64 /

-

/

t I t I

'"T, ,

_ _

I

• I • I I

I

390

II

ONSET

COz

OF VENTILATION

370 --

(ppm)

L I

....... "'...... N•

'•

• •930

-

/%+.

350 --

/+

330 --

320

•0

3•

380

4•

420



4•

a•

576

Aug 85 I

310 0

4



3 Aug 85 I 8



I

,

12

I 16

,

I 20

,

I 24

,

I

,

4

TIME

Fig. 4. Dis•budon of c•

wi• heiCt on Au•st 2-3, 1985, at •cke

Reserve.Notetheverylow valuesobserved at nightjustabovethecanopy Fig. 5. Temporalvariationsof CO• and CI-hat 28 m, just abovethe top. canopy, onAugust2-3, 1985.

TABLE2. CO?VerticalProfiles Flight

[COz] bl

Zbl,

Local

intCOz

m

Time

O-Zlkm

(lc•a)

[COz]

Terrain

(1.8-2.1

laxi)

July 19, 1985

338.6

442 859 1372 1244

0950 1110 1230 1350

3122.0 • 3115.5 3149.9 3096.6

345.6 342.3 343.7 342.8

forest/wetland forest forest forest

359.0 354.3 347.2 342.0

244 573 1036 1360

825 1000 1100 1240

3113.0 3136.9 3129.7 3115.3

342.4

forest

342.9

forest

342.9 344.6

river/wetland forest

6.1

358.3

6.2 6.3 6.4 6.5 6.6 6.7 6.8 6.9

360.6 351.8 351.9 348.8 346.8 345.2 344.0 --

244 408 1006 975 945 1250 1676

840 910 943 1005 1050 1120 1205

3141.2 3156.7 3162.1 3161.6 3149.1 3142.8 3138.5

345.2 346.1

palms/fiver low vegetation

344.9

forest

345.8 345.5 345.7

forest forest forest

347.4

forest/RioXingu

1890 ND

1230 1330

3124.6 3143.8

347.0

forest

346.2

ocean

348.2 343.2

ocean

forest/RioXingu forest(6.6) forest(6.5) forest(6.4) forest(6.3) forest(6.2)

night

4.1

357.1

4.2 4.3

345.6 348.6

4.4

July 21, 1985 5.1 5.2 5.3 5.4

July 23, 1985

July 24, 1985 7.2 7.3 7.4

-347.0 347.4 346.9

ND 1097 732 914

840 950 1015 1105

3143.3 3133.7 3140.6 3134.5

7.5 7.6 7.7 7.8

351.1 347.6 347.4 344.1

1067 1219 1067 1280

1125 1215 1230 1330

3161.7 3143.7 3143.4 3130.1

345.5 345.1 345.8 345.7 347.0 346.6

00

3108.2

343.3

1274 1402 1682 1676 838

1250 1415 1625 1645 1730

3126.5 3131.4 3128.9 3117.3 3109.3

ND 152

401 810

152 290

7.1

Amazon delta

July 25 and 26, 1985

8/9t'

(340.5)

ND

July 29, 1985 10.1

345.1

10.2 10.3 10.4 10.5

345.4 345.1 343.0 340.2

July 31, 1985 11.1 11.2

(340.0) 356.7

343.5

Rio Solimoes/wetlands

344.4

Rio Negro

344.2

Rio Solimoes/wet!ands

345.8

Rio Negro/islands

343.1

forest/Ducke

3108.3 3134.5

344.5

Rio Solimoes/wefiands

343.5

Rio Negro

845 905

3118.2 3127.6

344.7

wetlands

344.6

forest/wetlands

August5, 1985 14.1 14.2 14.3 14.4 14.5 14.6 14.7 14.8 14.9

353.7 349.1 352.9 356.0 346.6 347.0 342.7 340.7 343.6

180 290 391 367

925 948 1011 1033

3114.5 3130.7 3109.0 3110.7

343.5

wetlands/forest/fiver

345.2 343.4 344.5

wetlands forest Rio Solimoes

692 872 1001

1102 1125 1157

3112.8 3106.8 3119.9

344.9

forest

343.4 343.9

forest forest

15.1 15.2 15.3 15.4 15.5 15.6 15.7

372.9 364.7 368.6 350.8 347.8 348.4 349.5

149 241 274 362 579 594 671

825 845 920 940 1020 1040 1110

3134.4 31 40.5 3129.6 3124.0 3121.1 3127.3 3136.8

345.2 343.9 343.3 344.7 343.3 344.3 344.8

forest forest forest wetlands/lake forest forest/wetland forest

15.8 15.9 15.10

347.3 -•

732 ND ND

1135 1200 1220

3128.0 3120.2 3132.5

345.8 343.5 345.9

forest wetlands wetlands

August6, 1985

Zbl = mixed layerheight;ND, couldnot be boundarylayer,for nocturnalprofiles. a Corrected for calibration error.

eComposite offournocturnal profiles.

determined

from the data,parenthesis indicatea meanin remnant

1382

WOFSYET AL:CARBONDIOXIDEOVERTIlE AMAZON

Flight5

816 localt•me

( o)

GTE/ABLE

2100

2A

(b)

Ftt6

5000

Spirol I

840/t

1800 F

TWI

2•0• TR

:

1'

-J

R'CCL ,•

F

9oo

i

•00

RMLI • (noctu rnol

=

340 9O0

)45

I 350

I 355

I , 360

365

[CO2] (ppm)

600

RIV L2 • 0

I

I

I

I

I

550

555

•00

5

10 15 20 250

[CO2] ppm

I

I

I

I

WINDSPEED

I

10 •

!

•0

3300--

AT(øC)

(KNOTS)

Fit

14

840

om

3000 --

I

•F (d)

3300 --

FI! -• OCEAN' ( e 0840

am

local

time

,

300O

I i

---

2l'00 --

2100

24O0--

2400 --

2100 --

2100-

1800--

1800-I



1500--

1200 --

1200 --

90• --

900-

ß

ßß

600 -

500 t new b.L • o/,540

, I 550 ,,-'! 540 , 545

, 545

[COz](ppm)

550

, 555

, •0

, 5t15

, 5•0

, 0

340

345

CO2 (ppm)

3•5

340

345

GO 2 (ppm)

Fig. 6. EarlymorningC02profiles.(a) Verticaldistributions of C02,windspeed, anddew-pointdepression overtheforestat 0816 LT on July21, 1985. The nascentmixedlayerwith highCO•is clearlyseenbelow250 m. The relicmixedlayerfromtheprevious dayis seenbetween250 and 1500m (denotedRML). The relicconvective cloudlayer(RCCL) is seenbetween1500and2400 m,

with top definedby a wind shear,a sharpchargein dew-point depression, anda temperature inversion.A transition sublaye r appearsabovethat, cappedby a stronginversionand very dry air above2700 m. A pointwith errorbarsdenotesmeanand standard deviationof datafrom 1 to 4 minuteslevelflight. (b) Profileof CO•overforestat 0840 LT on July23, 1985,showingthe nascentmixedlayerandrelic layersabove,asin Figure6a. (c) Profilesof CO•overforeston August6, 1985,showingthe mixed layerbetween0825 and0940 LT, andthe 1ow-CC•reliclayersabove. (d) Profileof CC•overwetlandsat 0836 LT. Thereis no signof thenascentmixedlayerdownto 150m altitude.(e) Profileof CO•overthecoastalocean,150km NE of Belem,Paraguay.

(Figure 7a). The heightof the mixed layer rose more or less uptakeby the forest.The regionof CO2drawdownis particularly uniformlyduringthe morning,while the CO2 contentdeclined. clearwhenonecompares adjacentverticalprofilesobtained over By midday the CO2 profile was inverted, with lowest land (Ilha de Marajo) and over the sea at the end of flight 6 concentrations at the lowestflight level, reflectingphotosynthetic(Figure 7b). Depletionof CO2 often extendedup to 1800 m

GTE/ABLE 2A

WOFSY ETAt.:CARBON DIOX•E OVERTim AMAZON

1383

(b)

Fllghl #6Belem

Monaus--

-

23 July 1985

3000

_

(o)

F•ight #G

,

Monaus--Belem

'

2:5 July ]985 2400

2400 --

'

_

_

•E•eO0-

ø

inland

o

i

I

335

/ •2zø! 345

'•% , •ooI

0

I

1330

I

I ace on

I

355 365 335 345 355 365 C02(pom) C0z(ppm) Fig. ?. Midmorningandafternoon CO:profiles. (a Progression of CO:profilesduringthe morningovertheforestsof the Eastern AmazonBasin,from flight 6. (b) Land-oceancontrastsfor early afternoonfrom flight 6 (local times as shown). Note the drawdownobservedoverland(Ilha dc Marajo) but not overocean.

altitude. This behaviorof CO: concentrations in the lowest2000 versa. Elevated CO: and depressedozonelevels were notedover

m of the atmosphere is similarto thatreportedby Schnellet. al. forestedareasandlow CO:/highozonelevelswerelocalizedover [19811overtropicalEastAfrica. rivers. Large variationsof CO: and O3 were observedin Nocturnalconditions.Nocturnalprofilesareshownin Figure8, association with forestedzonesor riversonly 1-2 km wide. which gives the averageof severalprofiles and the associated The small spatial scalesof CO: and Oa variationsprovide variancesfor data takenover 2 nights. The upperportionsof the evidence that the concentrations of these gases at 150 m are variousprofiles are essentiallycoincident.There is a marked controlledmainlyby localexchangebetweenthe atmosphere and increase, however,in the varianceamongtheprofilesbelow1500 the surface. Nocturnalfluxes of CO: are highestover forests,

m. Theincreased variance coincides withthemixedlayerof the intermediate overwetlands, andlowestoverriven,reflecting mc previous day. The decayof themixedlayerleavesbehindair relative metabolic ratesof thesesystems. Deposition ratesforOa parcels witha rangeof CO:contents, reflecting different timesof would be expectedto follow the same relative order. contact withtheforestcanopyduringtheafternoon.

Concentrations of the two gases are thereforenegatively

It is interestingto notethattherelic boundarylayersobservedat correlated. The observedvariationscould also reflect vertical

night(Figure8) andin theearlymorning(Figure6) werenearly exchange processes, sinceO• andCO2arenegatively correlated indistinguishable from the middayand late afternoonprofiles with altitudeduringtheearlymorning.However,in thatcasewe (Figures7a-7c). The closeagreementbetweennighttimeand wouldexpectto find lower dew point temperatures coincident early morningprofilesshowsthat at night there was little withhighOdlowCO2values.Thisrelationship wasnotobserved. exchangeof CO: betweenthe forestlayer and the detachedrelic Land/river differences in CO: were also present in the layer. Nocturnalcirculations of the kind indicatedin Figure5 afternoon, asshownin Figure9b. However,themagnitude of the must have been too shallow to influence CO: concentrationsat variations was smaller and the boundariesbetween the different

300 m, theminimumaircraftaltitudein nighttimeflights. INFLUENCE OF VEOETAT•ONTYPES AND RPe•s

vegetationdomainswere not as sharplydefined. The forest efficientlyremovesCO: duringthe daytimehours,while the fiver remainsa net source[Richeyet al., 1987]. Thus a drawdownof COz was observedover the forestedislandsof the Rio Negro, but

Therewereimportant 0ffferences between COzdistributions nodrawdown wasfoundovertheadjacent water(Figure 9b). The overrivers,wetlandsandforestswhichreflectdifferences in rates differences shownin Figure9b extendedverticallythroughout the for CO: metabolismand for atmospheric mixing. Figure 9a mixedlayer,asshownin Figure9c. Theseprofilesweretakenat presents datafrom a level flight segmentat 150 m showingthat eitherendof thelevelflightsegmentshownin Figure9b. Profiles concentrations of COz werehighestover forests,intermediate over wetlands,and lowestover rivers. Fluctuationsin COz alongthe flight trackcloselymirrorfluctuations of Oa [Gregoryet al., this issue],with high O• levels coincidentwith low COz and vice

of COz over wetlandswere intermediatebetweenprofiles over land andover rivers: drawdownof COz was observedduringthe afternoon,but it was less than half as large as observedover forests.

1384

WOFSYL:rrAL:CARBONDIOXIDEOVERTHEAMAZON

3300. • ---T .......1-' I L•q-•

2A

mean COzconcentration inthemixed layer, mixed layer height,

••T

3•-

GTE/ABLE

mean COz between 1800 and 2100 vegetation type ofthe underlying terrain, local time, and anm, estimate of the integrated

-

•(• -

COz contentfrom the groundto 2100 m. The boundarylayer heightwasdeterminedfrom CO• andO• data,in caseswherethese gasesclearlydelineatedthe mixed layer. The integralof the COz

•400-

content assumes that the concentration is uniform from the lowest

aircraft altitude to the ground. This approach deliberately

... •)00•-

*

excludes the shallow nocturnal andearlymorning boundary



layers.

"' •800-

The rateof declineof CO•.columncontent duringtheday

• • 1500-

* -



providesa meansfor estimatingthe rate for CO•. uptakeby vegetation.This estimateis subjectto severalsourcesof error.

First,smallinconsistencies in calibrations maycontribute to •200-

-

s00-

-

s00-

_

300 -

_

variationsin the integrals.Second,exchangesof CO•. with air above 2.1 km have been neglected. Third, regional-scale advectioncan influenceCO•.levelsbetween1.5 and2.1 kin; as shown clearly during episodesof pollution from agricultural burning [Andreaeet al., this issue]. Finally, CO2 gradients betweenthegroundandthelowestaircraftaltitude(-150 m) could

alsoinfluence theanalysis. In order to minimi7g the influence of theseerrors,trendsin the

0

I

I

3•0 3•t 3•

I

I

34• •440

CO•l•rn)

! ! I I



I I

4

s

I

integralcontentare computedusingprofilesfor the sameflight,



reducing theeffect ofinslrumental driftandofanyreallong-term

CO•W•c• ½•me) changes in regional CO•.concentrations or environmental

Fig.8. Nocturnal CCh profiles, averaged fromfourspirals andanumber of conditions (e.g.,insolation). Ourattention is focussed onthetime

level flight legs onflights 8and 9.in(Left) Observed mean concentrations. period between 0800and1300LT. During thistimeperiod the (Righ0 Varianceof measurements left panel. The relic mixedlayeris clearly defined bylarge variances. Notethetopofthereliccloudmixed layerandthecloud convection sublayer werestrongly convective layer, marked bya CO•anomaly, asinFigure 6a.

coupledto the groundby small-scale convection, but deeper

convectionusuallywasnot stronglydeveloped. Figure 11a showsthe CO•. column contentobservedduring

Several factors influencing mixedlayergrowth aresuggested in flights4 and5 andalsoshowsthe verticalgradient of CO•. Figure 10. Herethetopof themixedlayer,asdefined bytheCO• observed in thelowermixedlayer.Bothflights covered thesame gradient, is shownas a function of timefor various typesof areaof undisturbed forest SEofManaus infairweather. TheCO•. vegetation. Therewasa tendency forthemixedlayerovereasterncontent isseentorisebetween 0800and1000LT, thentodecline Amazon forests to be higherthanoverwestern forests at any between 1000and1400LT. Theriseandfallof CO•.content is specific timebetween 1000-1300 LT, reflecting twofactors. A mirrored bytheCO•. gradient, which ispositive between 0800and

rainfall gradient existed during ourexperiment, withconsiderably 1000andbecomes negative thereafter. Therateof change of

more convective activity inthewestern Basin [Harriss etal.,thiscolumn CO•. during themidday period was-4.5kgCha-1h-1and issue]. Higher insolation andlower surface moisture would be -7.4kgC ha-1h-I forflights 4 and5,respectively. A similar

expected toproduce morevigorous dryconvection in theeasternpattern wasobserved inflight6, which crossed theforests ofthe partoftheBasin, andhence more rapidgrowth ofthemixed layer.eastern Amazon Basin(Figure11b).TheCO2uptake rateat Also,theheating of theforest canopy begins about 1 hourearliermidday wassomewhat larger inthiscase, corresponding toaflux

ontheeastern coast thanon thewestern sideof theBasin,of-12.7 kgCha-• h-l. Essentially thesame uptake ratewasfound

following thesun. forflight7,which retraced theflightpathofflight6. Thetrend in Growth ofthemixed layerwasslower overwetlands (3.5-5cm ratesforCO•.uptake correlates withtherainfall gradient, and s-•) thanoverforests (8-12cms-•),asshown inFigure 10.Mixedpresumably theinsolation gradient, across theAmazon Basin layerheights werehigher overtheblack water environments ofthe during ABLE2A. RioNegro watershed thanoverthesediment-rich whitewaters of Thepattern of rising,thendeclining, CO•.content in the theRioSolimoes. Observations witha downlooking Precision morning appears toreflect theinfluence of highlevels of CO2 Radiation Thermometer (PRT5) fromtheaircraft indicated thatthe stored intheforest canopy during theprevious night.Thecanopy blackwatersurface wastypically 2ø-4øC warmer thanthewhitelayerexchanges thisstored CO•.withthemixed layerrelatively waters. Thusthedifferences in mixed layerheights could reflectslowly, asmaybeseen inFigure 5. Since wecould notmeasure inputs ofsensible heatfromthesurface waters. theCO2profile between 0 and150m,thisexchange appears asan

•ARY

ESTIMATES OF CO•.

EXCHANOERATESOVER FORESTSAND WETLANDS

increasein the CO•.columncontent. Uptake rates over wetlandsappearto be smaller than over

forests, asexpected from Figure 9e.Unfortunately, asshown in

Table2 andFigure1lc, we haveto rely on a composite of profiles Table 2 summarizesmore than 50 verticalprofilesof COg from Flights10 and 11 in thiscase. The columncontentof COa obtainedin the ABLE'2Amission.Here we haveexcludedresults declinedvery little over the floodplainduringthe day,reflecting from disturbedweatherand from flights16 and 17. The latter lowernet photosynthesis rates.This resultis consistent with the were heavilyinfuencedby biomassburningand are discussed in interpretation of CO•.andO2cyclesin floodplainwaterspresented detailby•4ndreaeet al, [thisissue].We list for eachprofilethe by Richeyet al. [1987]. Theseauthorsfoundfloodplain?atersto

GTE/ABLE

2A

WOFSYET AL:CARBONDIOXIDEOVERTI-IEAMAZON DISTANCE

O

ALONG FLIGHT PATH (km)

25

14- •

I

1385

• ' I

5O

I

i

i

I /,•

I

i"

12

•0

8

6•WETLAND-..ffT•RIVER •-'#ETLAND-•..f..•-FOREST --.----WETLAND '

FORESTED

•o•



L[v[[



FORESTED

/• '

FOREST

L[V[,

-

360

340[

•- ::

I 5

q

4•7'S 61•I'W

•o

TIME ALONG FLIGHT PATH (m•n.)

3045'S

(833-843 localtime)

61'55'W

Fig. 9a. Transectof CO•concentration overrivers,forests,andwetlandsat 150 m altitudeon August6, 1985,between0833 and 0843 LT.

beanetsource ofCOzwitha mean emission rateof 1.8kgC ha-c net fixationof CO•.proceeds with aboutthe samequantum h-1. Theorganic material beingoxidized on thefloodplain is efficiency in Canadian andAmazonian forests. Thisresult should thought to beprovidedby influxfromtheforestand/orby primary perhaps havebeenexpected, sincenearlyall of theincidentlightis production duringperiods of lowwater. intercepted by leaf surfaces in bothsystems, andtheplantsare MeanratesforCO•.uptakearesummarized in Table3. Midday presumablywell adaptedto other environmental conditions CO•.fluxesfortheforests wereestimated tobe-9J:4 kgC ha-1h-1, (temperature, soiltype,moisture, etc.)ineachforest type. usingdatafromflights4, 5, 6 and7. Theobserved ratesfor COz We computed a verticalexchange coefficient for the height uptake in Amazon forests areclosetovalues (-7+3kgC ha-1h-1) range150-300m, usingtheestimated noontime fluxof COzand observed by Desjardins et al. [1985]overCanadian forestsat theobserved gradient.Thederivedvalues,1-2x10 scmzs-1,arein

similar solarirradiance (--600W m-z) andaresimilar alsotothe therangeexpected at thisaltitude for an activemixedlayer meanvalueinferredby Odumet al. [1970]for forestsof Puerto [JacobiandAndre,1963;Liuetal., 1984]. Thisresultprovides an Rico. The uptakerate for wetlandswasmuchsmallerthanfor independent checkonourapproach for estimating fluxes.

forests withanestimated fluxof-1.6kgC ha-1h-1. It appears that

Thedaytime meanvalueforuptake of atmospheric COzcanbe

DISTANCE TRAVELLED (km) o

35o

--

25

50

I

I

(b)

75 I

ABLE 2A-FLIGHT 10

,

240o I (c)

ALTITUDE - 150m

dULY 29, 1985 348 --

1620-]640

....

LOCAL TIME

346

i ....

/

x•., •

F 29JULY 1985

18•

60"28'W •

/

2o26'S

'• 61ø01'W

x'

t•me

16311•al •,(

/

344

3,• 340

338

336

I

o t

3o0's 60o28'W

TIME ALONGFLIGHTPATH (rain)

I

2':'26'S 61o01'W

•1x i I•A•S •i i !

0 /

3•

Fig. 9b. Transeet of riverandislandsystemat 150m altitudeat 1520LT on July29, 1985.

Fig. 9c. •fil•

i

i



RIO• NEGRO I i i I

i

345

COz (ppm) of C• at eider endof •e •ns•t

i

3•

in Figu• Fi•

9b.

1386

WOFSYET AL:CARBONDIOXIDEOVER•

AMAZON

GTE/ABLE

2A

l

I

1800--

/

FORESTS

--

1200

--

_

/.+

Eastern (6)'"x.. / .//•.Centrol (4,5) -

ß

o

5150

_

•-•

_

e• /

0

I

• 14(•A•zonR.)o 15 i

-

_

-•,•1Ch(:/

•:•o

I

I

i

I

I I ! 1000 II00 1200 LOCAL TIME (hr)

,I 1500

!

Fig. I lb. Columnabundance of CO• observedat varioustimesof day for flight 6, covering forests of the eastern Amazon Basin. The CO• concentration was not observedbelow 150 m, leadingto an underestimate for thefirsttwo time points(seetex0.

WETLANDS • _

1500 --

I 0900

0800

x• r•

_

I

31•0 I



I

_ •

e•/e•,••/'•' Western (14,15) _

300--

I

Forests

600--

-r

I

Flight6

+/

/

900--

I

(b)

3160

1500

I

1200 --

flux is estimated to be -2.8 kg C ha-• h-•. If we assume that daytime COz fixation shouldbe balancedapproximatelyby net evolutionof COz at night, thenthe meannocturnalCOz flux for

900--

600-

/ //io Solimoes 8 • 08•

0

I 10•

canopyandsoilsshouldbe about+2.8 kg C ha-• h-•. Emission ratesfwm soils,1.8 kg C ha-• h-1, accountfor 65% of the estimatednocturnalCOz flux, providingan indicationof themajor wle playedby soilsin oxidationof organiccarbonin theseforests.

•' wetlands I 12•

LOCAL

I 1•

TIME

I 16•

I 18•

SUMMARYAND CONCLUSIONS

(hr)

The cycle of atmospheric COz in the AmazonBasinprovidesa Fig. 10. Mixed layerheightsas definedby CO•concentrations, over detailedpicture of daily exchangesof air betweenthe tropical

various vegetation types, shown asfunctions oftimeduring thedayfor forest(0-30m) andtheatmospheric boundary layer(30-2000 m).

flight numbers asindicated.

Theforest wasanetsource ofCOzatnight andasink during the

day. Largeconcentrations of COz (380-400ppm) accumulated in estimatedby assumingthat total photosynthesis ratesareroughly the forestat nightweretransported intothe nascentmixedlayerin proportionalto incidentsolarflux [Odumet al., 1970;Desjardin.vthemorning.Between1000 and 1200 LT, COz in themixedlayer eta/., 1985].The 12-hourmeansolarirradianceat the canopytop declinedto concentrations 4-6 ppm lower than free tropospheric

wasabout190W m-1, andthecorresponding daytimemeanCOz values,thenremainedlow for the rest of the afternoon.The 5150

•0

:

aRemoonmixed layer became isolated at night and the lower valuesof COz were preserveduntil a new mixed layer developed the next morning. Entrainment of air into the mixed layer appearedto be a one-way processduring the early stagesof growth,but a significantexchangewith higheraltitudesoccurred in laterstages. We have shown,by analyzingobserveddistributions of COz, that the lower atmosphere over forestsfunctionsseparatelyfrom that over rivers or wetlandsduringthe night and to someextent duringthe day. The basicdiurnalcycleof COz over wetlandsis muchweakerthanover forests,andthe cycleis almostabsentover rivers. This result is consistentwith expectations basedon tae biogeochemistry of organiccarbonin thesesystems.Examination of the COz cycle in and over tropical forestsprovided basic

....

(o)

CO zCOLUMN [elABUNDANCE OFIt

ß FIt5 I

516O -- -4.5 kg Cha-lhr '1 3000 I

4--

I

I

I

I

!

information about rates forforest-atmosphere exchange ofthe gas.

+ - CO zGRADIENT-LOWER -- 4

IXEDLAYER --2• •

0 ......

0 u

!



I

I

,•

,•

el

I

,2m ,•

I

.....

E



,4m I•

• ]l,O•

LOCAL TIME Fig.1la. (Up•rpanel) •lculated •lu• abun•nce ofC• over foms•

• • •100• • Fit !1

d•bed 0 in •e 2150 text,m for fli•ts at 4 va•s and5. •mes •e low value at 0830LT between and al•mde ofday, •lculat• as rapresents •e relic••

layerandconv•dvecloudlayer,since•e new

•xedlayer hadnotyetgm• toaimm• •tudes. (•wer panel) Gradients ofC• •tw•n 150and3• matva•ous •m• of daydu•ng •i•ts 4 and5.

WETLANDS e Fit

•0ffi•

4•

I

6•

. I

I

I

I

I

8• 1• 12ffi 14ffi 16• 18ffi LOCAL TIME(hr)

Fig. 1lc. Same asFi•m 1lb for wet,ds in •e •o Ne•

region.

GTE/ABLE 2A

WOFSY ETAL:CARBON DIOXIDE OVERTIlEAMAZON

1387

TABLE3. CO1Fluxes FromtheCentral Amazon Forests andWetlands Grant,R. H., G. E. Bertolin,and L. P. Harrington,The intermittentvertical heatflux over a spruceforestcanopy,BoundaryLayer Meteoral.,35,

Midday Forest

-9'J:4 •

Wetland

-1.6 a

Day Mean -2.8+1.2b

317-330, 1986.

Gregory,G. L., E. V. Browell, and L. S. Gahan,Boundarylayer ozone: An airbornesurveyabovethe AmazonBasin. J. Geoph•s.Res.,,this issue.

Fluxes arein kgC ha-! h-l.

ßFrom d/tit {f[COz] dz}, Figure 11.

bFromnoonvalueandassumed proportionality tonetinsolation.

Harriss,R. C., S.C. Wofsy, M. Garstang,L. C. B. Molion, R. S. McNeal, J. M. Hoell, R. J. Bendura,S. M. Beck, R. L. Navarro,J. T. Riley, andR. C. Shell, The Amazon BoundaryLayer Experiment(ABLE 2A): Dry season1985),J. Geophys.Res.,thisissue.

Hutchinson, (3. L., andA. R. Mosier,Imposedsoil covermethodfor field measurementof nitrous oxide fluxes, Soil Sci. Soc. Am. J., 45, 311-316,

Themidday rate forCOz uptake was about 9(+4)kgCha-•h -•, Jacobi, 1981. W., and K.

Andre, The vertical distribtuion of radon 222, radon

andthedaytime mean uptake ratewas2.8(+1.2) kgC ha-1h-1. 220, and their decay products intheatmosphere, J.Geophys. Res., 68,

The flux of COz from soilsof the tropicalforestat Duckereserve

3799-3814,1963.

averaged1.8(+0.2)kg C ha-lh-1. If we assumeapproximate Kaplan, W. A.,S.C.Wofsy, M. Keller, andJ.M. daCosta, Emissions of balancebetweendaytimeCOz uptakeand nocturnalCOz NOanddeposition of03ina tropical forest system, J. Geophys. Res.,

emissions inforests, thenocturnal production rateforCOzwould this issue.

be 2.8 kg C ha-lh-1, of whichsoilswouldprovideabouttwo- Keller,M., W. A. Kaplan,andS.C. Wofsy,Emissionsof NzO, CH4, and

thirds.

CO2 from tropicalforestssoils,J. Geophys.Res., 91, 11,791-11,8012,

1986.

Future workshould helptorefine thepresent results. Improved Klinge, H., andW. A. Rodrigues, Litterproduction in anarrea of

determinations of COz verticalfluxes (for example,by the eddy- Amazonian terrafirmeforest, Amazoniana, 1,287-310,1968. correlationtechnique),in concertwith carefulcharacterization of Lemon,E. H., L. H. Allen,andL. Miller,Carbon dioxideexchange of a

COzandtracegasdistributions, should providea powerful toolfor

tropical rainforest, II, Bioscience, 20,1054-1059, 1970.

analysis of gasandtraceelement exchanges between terrestrial Liu,S.C.,J.R.McAfee, andR.J.Cicerone, Radon 222and tropospheric

systems and theatxnosphere.

vertical transport, J.Geophys. Res., 89,7291-7309, 1984.

Lundegardh,H., Carbondioxideevolutionandcropgrowth,Soil Sci., 23,

Acknowledgements. We ar• indebted to Fanya(3im. melfarb,Irma 417-453, 1927. Tovbina, Roger Navarro, James Hoell,andthelateDr.OttoGrubnet for Medina, E., H. Klinge,C. Jordan, andR. Hartera, Soilrespiration in their assistance in this experiment.This work was supported by NASA gramNAG 1-55to HarvardUniversity.

Amazonianrain forestsin the Rio Negro Basin,Flora, 170, 240-250, 1980.

Odum, H. T., A. Lugo, G. Cintronn, and C. F. Jordan,Metabolism and evapotranspiration of some rainforestplants and soil, in A Tropical Rainforest,edited by H. T. Odum and R. F. Pigeon,pp. I103-I164, Division of TechnicalInformation,U.S. Atomic Energy Commission, Alvo, P., R. C. Desjardins,P. H. Schuepp,and J.I. MacPherson,Aircraft Washington,D.C., 1970.

measurements of COz Exchange overvariousecosystems, Boundary-Raich,J., J. Ewel,andM. Olivera,SoilCOz effluxin simpleanddiverse LayerMeteorology, 29, 167-183,1984. ecosystems ona volcanicsoilin CostaRica. Turrialba,35, 33-42, 1985. Andreae,M. O., et al., Biomass-burning emissions and associated haze Richey,J., A. H. Devol,S.C. Wofsy,R. Victoria,andM. N. G. Ribeiro, layersoverAmazonia.J. Geophys. Res.,,thisissue. Oxidationand reductionratesfor organiccarbonin the Amazon

Browell,E. V., G. L. Gregory, R. C. Harriss, andV. W. J. H. Kirchoof, Mainstem,Tributaryand Floodplain,inferredfrom distributions of Tropospheric ozoneandaerosol distributions across theAmazon Basin, dissolved gases, Limnol.Oceanogr., inpress, 1987. J. Geophys. Res.,,thisissue. Schlesinger, W. H., Carbonbalance in terrestrial detritus.Annu.Rev.Ecd. Denmead, O. T., andE. F. Bradley,Flux-gradient relationship in a forest Systematics, 8, 51-81,1977. canopy, pp.421-442,in TheForest-Atmosphere Interaction, edited byB. Schnell, R. C., S.-A.Odh,andL. N. Njau,Carbondioxidemeasurements

A. Hutchinson andB.B.Hicks, D. Reidel, Hingham, Mass.,1985. intropical EastAfrican biomes, J.Geophys. Res.,86,5364-5372, 1981. Desjardins, R.L., J. L. MacPherson, P. Alvo, and P.H. Schuepp,Shuttleworth, W.J.,et al., Dailyvariations of temperature andhumidity Measurements of turbulent heatandCO2 exchange overforestsfrom withinandaboveAmazonian forest,Weather, 40, 102-107,1985. aircraft,pp. 645-658,in TheForest-Atmospheric Interaction,editedby Whittaker,R. H., andG. E. Likens,Carbonin thebiota,in, Carbonandthe B. A. Hutchinsonand B. B. Hicks,D. Reidel,Hingham,Mass., 1985. Desjardins,R. L., E. J. Brack, P. Alvo and P. H. Schuepp,Aircraft monitoringof surfacecarbondioxideexchange,Science,216, 733-735,

Biosphere,edited(3. M. Woodwelland E. V. Pecan,pp. 281-302, U.S. Atomic EnergyCommission,Washington,D.C., 1972. (Availableas Conf.720510, Nat. Tech.Inf. Serv.,Springfield,Va.

1982.

Finnigan,J. J., Turbulenttransport in flexibleplantcanopies, pp.443-480, in, The Forest-Atmosphere Interaction,editedby B. A. Hutchinsonand

R. C. Hatriss, Mail Stop 483, NASA Langley ResearchCenter,

b. B. Hicks, D. Reidel,Hingham,Mass., 1985. Hampton,VA 23665.• W. A. Kaplan ara:;S.C. Wofsy, Division of Applied Sciencesand Garstang,M. A., et al., Tracegasexchanges andconvectivetransports over Departmentof Eartl'• and Planetary Sciences,Harvard University, the Amazonianrainforest,J. Geophys.Res.,,this issue.

Goreau, T. J. andW. Z. DeMello,Effectsof deforestation onsources and Cambridge, MA 02138, sinksof atmospheric carbondioxide,nitrousoxide,and methanefrom someAmazonianbiota and soils,in Workshopon Biogeochemistry of TropicalRainforests:Problemsfor Research,editedby D. Athaie,,T. E. Lovejoy,andP. deM Oyens,Universidad de SaoPauloandWorld

WildlifeFund,Diracicaba, S.P.,Brasil,Piricicaba, Sao Paulo,Brasil, 1985.

(Received December 18, 1986; revised May28, 1987; acceptedMay 29, 1987).