Jul 16, 2000 - However, Bates et al. [ 1996] derived an ...... (submit- ted manuscript, 1999). ... Bates, T. S., K. C. Kelly, J. E. Johnson, and R. H. Gammon, A.
JOURN• OF GEOPHYSICALRESEARCH,VOL. 105,NO. D13,PAGES17,243-17,255, JULY 16,2000
Simulation of preindustrial atmospheric methane to constrain the global source strength of natural wetlands SanderHouweling, Frank Dentener,and JosLelieveld Institutefor Marine andAtmosphericResearchUtrecht(IMAU), UtrechtUniversity,Utrecht,Netherlands
Abstract. Previousattemptsto quantifythe global sourcestrengthof CH4 from natural
wetlands haveresulted in a rangeof 90-260Tg(CH4)yr-•. Thisrelatively uncertain estimatesignificantlylimits our understandingof atmosphericmethane. In this study we reducethis uncertaintyby simulatingpreindustrialCH4 with a three-dimensional
chemistry-transport model.Methane mixingratiosand•13C-CH4,asdeduced fromice cores,and estimatesof otherpreindustrialsourcesand sinks,are usedas constraints.This
yieldsanaverage preindustrial natural wetland source strength of 163Tg(CH4)yr-1, with anestimated 4-20uncertainty rangeof 130-194Tg(CH4)yr -• . Thepresent natural wetland sourcemay be • 10% smaller,owingto drainageand cultivationof wetlandareasince1800 A.D. The simulatedpole-to-poleconcentrationdifferenceis foundto be ratherinsensitive to the assumedrelative contributionsof importantpreindustrialsourcesand sinks, and thereforeimposesonly a limited constrainton the estimateof naturalwetlandemissions.In
contrast, •13C-CH4 couldprovide robust constraints, but,unfortunately, atpresent reliable measurements are absent. Estimatesof the historicdevelopmentof anthropogenicCH4 sources,in combinationwith our model calculations,can largely explain the increaseof methanemixing ratiosduringthe nineteenthcentury. Resultsfor the twentiethcentury indicatethat thesehistoricalemissioninventoriesunderestimateanthropogenic emissions by at least 10%. Simulationsof preindustrialand present-dayisotopicratiosshowthat the
growth of anthropogenic sources since1800A.D. mayhaveincreased ;513C-CH4 by 3%•. 1. Introduction
tion. The upscalingapproachaims to quantify this on the basis of statisticsof the geographicaldistributionof inunMethane productionby methanotrophicbacteria under dated area in combinationwith a typical emissionper unit anoxic conditions in natural wetlands constitutes the most area [Matthews and Fung, 1987; Seiler and Conrad, 1987; importantprocessby which methaneis emitted to the atAselmann and Crutzen, 1989; Barlett and Harris, 1993; mosphere.The quantificationof its global sourcestrength remainsa scientificchallengeaftermorethana decadeof re- Chappellaz et al., 1993]. Since measured emission facsearch.Sinceearly estimatesby Matthewsand Fung [1987] tors representa limited range of conditionsonly, this apand Seiler and Conrad [1987], several studiesbased on the proachis not well suited for the quantificationof wetland modelinghas the advantagethat extrapolationof process-related information(upscaling)and emissions.Process-based emissions can be estimated as a function of the process inversemodelingreportedglobal-scalesourcestrengthsin controlling chemical and physical conditions[Potter, 1997; therangeof 90-260Tg(CH4)yr-1, or •15-45% of itstoCao et al., 1996; Walter, 1998]. Many wetlandecosystems, tal source[Aselmann and Crutzen, 1989; Barlett and Harris, however, have not yet been studied in sufficientdetail to 1993; Chappellazet al., 1993; Cao et al., 1996; Hein et al., derive reliable parameterizations. In addition, the number 1997; Lelieveld et al., 1998; Walter, 1998]. A more accurate of field experiments to validate these models are limited, estimateof wetlandemissions is of importancesincethisrelwhich particularly pertains to tropical wetlands. Therefore, ativelylarge anduncertaintermin the atmospheric methane budgetsignificantlylimits our understanding of this impor- to derive global-scaleestimatesof wetland emissionsby thismethod,largegeneralizations areneeded,which signifitantchemicallyreactivegreenhouse gas. cantly increase the uncertainty of the results.Measurements Methane releasesfrom wetlandsresult from a complex of concentrations andisotopicratiosof atmosphericmethane interplayof many highly variablefactors,suchas organic can be utilized to constrainwetland emissions,as explored substratesupply,temperature,hydrologicalconditions,and by inversemodeling [Brown, 1993; Kandlikar, 1997; Hein the competitionbetweenmicrobial productionand oxidaet al., 1997; Houweling et al., 1999]. The added value of inversemodelingis ratherlimited by the availabilityof Copyright2000 by the AmericanGeophysicalUnion. measurements, asprovidedby the low-densityobservational networks such as the National Oceanic and Atmospheric Papernumber2000JD900193. 0148-0227/00/2000JD900193509.00 Administration(NOAA) cooperativeair samplingnetwork 17,243
17,244
HOUWELING
ET AL.: PREINDUSTRIAL
METHANE
[Dlugokencky et al., 1994] andthe CooperativeAtmospheric 2. Model Description Data IntegrationProject [GLOBALVIEW-CH4, 1999]. ParModel simulationspresentedin this studyhavebeencarticularlyin the tropicsthe numberof measurements is insufficientandgenerallytoo far from the sourcesto providesig- ried out using the global three dimensional(3-D) Tracer nificant constraintson wetland emissionestimates[Houwel- Model 3 (TM3) [Houweling et al., 1998, 2000; Dentener ing et al., 1999]. et al., 1999; Lelieveldand Dentener, 2000]. The geographIn this study,an alternativemethodto constrainthe natu- ical resolutionappliedis 10ø in the longitudinaland7.5ø in ral wetlandsourceis proposedandtested,thatis, by examin- the latitudinal direction with 19 vertical levels. The vertiing the preindustrialmethanebudget.A studyof the prein- cal levelshavebeendefinedasterrainfollowing coordinates dustrialperiodhas the advantagethat, sinceanthropogenic near the surface,pressurelevels in the stratosphere,and a sourceswere relativelyunimportant,naturalwetlandsdom- hybrid of the two in between. The horizontaland vertical inated the global methane source. On the other hand, ice transportof tracersis basedon six hourly mean meteorocore analysesprovide the only observationalevidenceof logicalfields,includingwind, surfacepressure,temperature, the compositionof the preindustrialatmosphere, and there- and humidity,derivedfrom EuropeanCentre for Mediumfore methanemixing ratios are known over Greenlandand RangeWeatherForecasts (ECMWF) re-analyses for theyear Antarcticaonly. Still, thesemeasurements give a clear in- 1993. For a descriptionof the parameterizations of transport dicationof the preindustrialglobal mean methaneconcen- and chemistrythat areusedin TM3 we refer to [Houweling tration,being ,-•40% of the presentlevel [Craig and Chou, et al., 2000]. Emissionsof photochemicaltracersotherthan 1982; Staufferet al., 1985; Battle, 1993; Chappellazet al., CH4 are basedon the Global EmissionsInventoryActiv1997; Etheridge et al., 1998]. In addition, the difference ity (GEIA) andEmissionDatabasefor GlobalAtmospheric between methane
as measured
in Arctic
and Antarctic
ice
Research (EDGARV2.0) inventories [Olivier et al., 1999;
coresquantifiesits preindustrialnorth-southconcentration Guentheret al., 1995; Yiengerand Levy, 1995; Benkovitz gradient[Nakazawaet al., 1993; Chappellazet al., 1997; et al., 1996]. To representthe time evolution of anthroEtheridge et al., 1998].Further, the13C/12C isotopic ratio pogenicsourcesfrom the start of industrialization,the hisof methaneasmeasuredin Greenlandice cores[Craig et al., toric emissioninventoryby J. A. Van Aardenneet al. (A 1988a] provides a constrainton processesthat fractionate 1øxløresolution dataset of historicalanthropogenic tracegas thesecarbonisotopesdifferently, such as biomassburning emissionsfor the period 1890-1990, submittedto Global (-25%0)andnaturalwetlands(-60%0) [Levin, 1994]. BiogeochemicalCycles,1999, hereinafterreferredto asAarTo utilize thesemeasurements, we simulatepreindustrial denneet al., submittedmanuscript,1999) hasbeenused. methaneby meansof a three-dimensional(3-D) chemistry As a test of boundarylayer mixing and regional-scale transportmodel (CTM). From ice core measurements it can transport, 222Rnsimulations at variousmodelresolutions be inferredthatthe anthropogenically inducedupwardtrend havebeencomparedwith observationsat continentalandreof atmosphericmethanestartedabout 1800 A.D. To mini- mote locations[Denteneret al., 1999]. From this studyit mize the contributions of anthropogenic sourcesto the over- follows that measured and simulated radon concentrations all emissions,we focuson the precedingperiod(1500-1800 agreequitewell; generally,deviationsare
