Table Of ContentAgriculturalandForestMeteorology182–183 (2013) 156–167
ContentslistsavailableatScienceDirect
Agricultural and Forest Meteorology
journal homepage: www.elsevier.com/locate/agrformet
Carbon dynamics in the Amazonian Basin: Integration of eddy
covariance and ecophysiological data with a land surface model
BassilEl-Masri∗,RahulBarman,PrasanthMeiyappan,YangSong,MiaolingLiang1,
AtulK.Jain
DepartmentofAtmosphericSciences,UniversityofIllinoisatUrbana-Champaign,Urbana,IL,UnitedStates
a r t i c l e i n f o a b s t r a c t
Articlehistory: Informationcontainedineddycovariancefluxtowerdatahasmultipleusesforthedevelopmentand
Receiv ed13August2012 applicationo fgloballa nd surfac emodels,s uch aseval uatio n/va lidation, calibr atio nan dprocesspar am-
AReccceepivteedd i2n7 r Mevaisrechd 2fo0r1m3 27 February 2013 eterization for carbo n sto cks and fluxes. I n this st udy, we combine Larg e Scale Bio sphe re–Atmo sphere
ExperimentinAmazonia(LBA)projectdatacollectedfromanetworkofeddycovariancefluxtowers
deployedacrosstheAmazonianbasintoimprovethecarbonstocksandfluxestimatedbyalandsur-
Keywords: facemode l,theI nteg ratedScienc eAsse ssm entMod el( ISAM).W eeval uate the keymodel par am eter sfor
ISAM
carbonallocationfactors,maintenancerespirationaswellastheautotrophicrespirationusingtheLBA
GPP siteme asurement s.Ourm odelresultss howthatth et otalb io mas srangesfrom 0.7kgCm −2yr− 1for the
LNSBoPiAPl carbon tphaest uunrcee srittaei ntoty 2r0a.8n gk eg oCf m th−e2syirt−e 1mfoera tshu er efmoree nstt sdiat eta. .TAh les oIS,At hMe reesstuimlt astreesv feoar l eGdPtPh aantda lNl,P bPu taoren ewfeolrl ewstit shiitne
havelowernetprimaryproductivity(NPP)togrossprimaryproductivity(GPP)ratio(NPP/GPP=0.4com-
paredtothesavannaandthepasturesites(NPP/GPP=0.5).Thisisbecausesavannaandthepasturesites
experiencedthelongestdryseasonandplantsgrowinginsuchenvironmentalconditionshavestronger
efficiencytostorecarboncomparedtoforests.Theforestevergreensite(Km67)hasahighermeasured
NPP/GPPratio(0.5),becauseofhighercarbonaccumulation.Soilcarbonislowestforthepasturesite
(Km77)( 7.2kg Cm−2 yr−1)an dh ighest forthef orestsite(Km3 4)( 12.8kgC m −2yr−1 ).Th em odelres ults
suggest that all the forest sites are a net sink for atmospheric CO2, while the savanna (PDG) and pasture
(FNS) sites are neutral and another pasture site (Km77) is net source for atmospheric CO2. Meanwhile,
themodelresultshighlighttheimportanceoftheLBAsitedatatoimprovethemodelperformanceforthe
tropicalAmazonregion.Thestudyalsosuggesttheneedforanetworkoflong-termmonitoringplansto
measurechangesinthevegetationandsoilcarbonbiomassatthelocalandregionallevels.Suchprograms
willbenecessarytomakereliableglobalcarbonemissionsestimates.
© 2013 Elsevier B.V. All rights reserved.
1. Introduction have led to an increase in carbon uptake by about 3PgCyr−1 in
undisturbedforests(Houghtonetal.,2000;Saleskaetal.,2003).In
TheAmazonianforestsplayakeyroleintheglobalcarboncycle, contrast,ecosystemmodelingstudiessuggestthatduetoclimate
contributingtoabout30%oftheglobalbiomassandterrestrialpro- variability,theAmazonbasinisanetsourceduringthedrierand
ductivity(Beeretal.,2010;Houghtonetal.,2001;MalhiandGrace, warmerElNin˜oandnetsinkduringthewetterandcoolerLaNin˜a
2000; Melillo et al., 1996). The increase in air temperature, and cycle(Asneretal.,2000;Bakeretal.,2008;Foleyetal.,2002;Potter
shiftinprecipitationintensityanddurationduetoclimatechange etal.,2001,2009;Tianetal.,1998).
has altered carbon fluxes from the Amazonian rainforests (Botta ThelandscapeofAmazonisalsochangingdramaticallydueto
etal.,2002).SomestudiessuggestedthatCO2 fertilizationmight humanactivities.Overthepastfewdecades,increaseddemandfor
agriculturalproductsduetorisingpopulationledtomassivedefor-
estationrates(Asneretal.,2005;Moran,1993;Mortonetal.,2005;
SkoleandTucker,1993).Deforestationcanalsoleadtodegrada-
∗Correspondingauthor.Tel.:+12173001842. tionin the ecosyst emserv icessuchaslo wer wat erqu alit y,spread
E-mailaddresse s:belma sri@ illi nois.edu(B.El-Masri),[email protected] ofin fec tiou sdiseases ,andinc rease si ntree morta litythro ughan
(R.Barman ),meiyapp [email protected](P.Mei yap pan),song [email protected](Y.Song),
increaseinfirefrequency(Foleyetal.,2007).Itisexpectedthatthe
[email protected](M.Liang),[email protected](A.K.Jain).
Pr1inCceutrornen,Nt Ja,dUdnrietsesd: SCt iavtiel s a.nd En vironmental Engi neeri ng, Princeton University, triaotnese oxfp danefdosretosttahtieocno wreilol fctohnetifnouree stto( Linacurreaanscee aest malo.,r2e0 u0r1b)a.nSiuzcah-
0168-1923/$–seefrontmatter© 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.agrformet.2013.03.011
B.El-Masrietal./AgriculturalandForestMeteorology182–183 (2013) 156–167 157
hadledtotheestablishmentofaLargeScaleBiosphere-Atmosphere
Notation ExperimentinAmazonia(LBA)(AvissarandNobre,2002)Themain
scienceobjectiveoftheLBAprojectistounderstandtheinterac-
Vegfrct vegetationcoverfraction
tionsbetweentheatmosphereandclimatechangevariability,in
Cn carbon:nitrogenratiofortheleaves
leaf theAmazonianterrestrialecosystems(Avissaretal.,2002).Such
Cnstem carbon:nitrogen ratio for the stem obs ervational d ata from t he LBA sites can sub sta ntia lly imp rove
CVncmroaoxts cmaarbxoimn:unmit rcoagrebno xraytlaioti foonr rtahtee raoto 2ts5◦C ethceo suynsdteemrstaannd dtinhge aotfm thoes pchareb roena ne dxcthhaen dgeet eb cettiwoneeonf dthefie cti eernrceisetsriianl
kn light extinction coefficient thelandsu rfac emo dels(LSMs) .For inst ance,seve ra lstudiesusin g
k baserespirationrateforleaves
leaf LSMshavepredictedadeclineintopicalforestwaterandcarbon
Kstem base respiration rate for stem fluxes duri ngthedry s eason(B ot taetal. ,2002; Tiane tal. ,1998),
Kroot base respiration rate for roots while siteme asur eme ntssug gestth eo ppo site(V onR an dow etal.,
ω allocationparameter
2004;Saleskaetal.,2003).Withoutconstrainingthemodelsusing
ε parametercontrollingallocationtoleaves
leaf siteobservations,LSMswouldperhapsunderestimatetheglobal
εstem parameter controlling allocation to stem carb onfluxesand failto detec tthecarb ondynamicof the tropi-
εroot parameter controlling allocation to roots calfore sts(Sa leska et al., 2003). The refore, theseobs erv atio nsare
k allocationparameter
crucial for the improvement of LSMs. Since climate change and
Y leaflifespan
leaf variabilityaffecttheglobalterrestrialbiogeophysicalandbiogeo-
rwmax maximum drought leaf loss rate chemical c ycles (Dav idson et al., 201 2; Malhi et al., 2008 ), LSMs
rtmax maximum cold leaf loss rate withthei rcomp lexanddet ail edb iogeoc hemic al and biogeo phys-
bw parameter for leaf loss (drought) icalp roces sesareim por tanttool stogainbetter unde rstandingof
bt parameter for leaf loss (cold) the interaction sb etweenthe Ama zo nbas interr estrialecosyste m
T temperaturethresholdforleaflossbecauseofcold
cold andenvironmentalchange.
stess
Inthispaper,wetakeadvantageofobservationaldatafromflux
Ystem stem turnover rate tower site stopr ese nta methodolo gy tocalibratean dim prov ean
Y finerootturnoverrate
froot LSMbyidentifyingandmodifyingthekeymodelcarbonschemesin
Ycroot coarse root turnover rate orde rto enhanceth eve getationa ndt hes oilcar bonest imates.W e
R maintenancerespirationforleaves
leaf usetheIntegratedScienceAssessmentModel(ISAM)toshowthe
Rstem maintenance respiration for stem feas ibili ty of our a pproach and its rele vance to othe r L SMs. T his
Rroot maintenance respiration for roots paper ext end s up on previo us m od eling appl ica tion o f the I SAM
R totalmaintenancerespiration
mtotal (Jainetal.,2009;Yangetal.,2009)bydetailingthecarbondynamics
R growthrespiration
G oftheLBAsitesrepresentingforests,savannas,andpasturesecosys-
Rauto autotrophic respiration tem so fth eAm azonianbasin .Notab ly,thever sion ofISAM usedin
C leafcarbon
leaf thisstudycontainsdetailedbiogeophysicalprocesses(Songetal.,
Cstem stem carbon 201 3)cou pledinto thebiog eochemicalcom ponento fISAM .T his
Croot root carbon extend edversi onof the modelisusedto examineth ei nteran nual
Ø leafphonologicalstatusdefinedasthecurrentfrac-
variabilityofcarbonfluxes(Grossprimaryproduction(GPP),net
tionofmaximumLAI
primaryproduction(NPP),autotrophicandheterotrophicrespira-
gt temperaturedependentfunction
TTv eg vsoeigletteamtiponer taet murpee(r◦aCt)ure (◦C) tinio nthse) aAnmd aabzoonve b aansdin b e(Tloawbl eg r1o)u. nTdh bei oombjaescst iovfe e ihgahst LbBeAen s taucdhyi esviteeds
soil bycomparingthemodelestimatedcarbonfluxeswitheightLBA
GPP grossprimaryproductivity
project flux towers measurements. The study is designed to: (1)
NPP netprimaryproductivity
provideabriefdescriptionofthecarbondynamicsofISAM,(2)cal-
Allofact allocationfactorforleaves
leaf ibratethemodelparametersusingsitedata,(3)evaluatethemodel
Allo factstem allocation factor for stem perfor man ce by comparing the m ode l cal cul ated carb on fluxes
Allo factroot allocation factor for roots with eddy co var iance flux t owe rs meas urements, and (4) evalu-
L lightavailabilityfactor
atethemodelestimatedsoilandvegetationcarbonwithpublished
ws waterstressfactor
studies.
LAI leafareaindex
LAImax biomespecificmaximumLAI
Tr root temperature 2. Methodology
Biotr biomespecificroottemperature
Biotrmin biomespecificminimumroottemperature
2.1. ISAMdescription
daylen biomespecificdaylength
TheISAMisoneofthe23modelsparticipatingintheLBAproject
tostudytheresponsesofAmazonbasinterrestrialecosystemsto
rapid changes in land use will lead to a massive carbon release environmental factors. The ISAM is a land surface model that is
fromthesoils,andvegetationspeedinguptheclimatechange.The appliedtoexaminetheimpactsofchangingCO intheatmosphere,
2
forestsandwoodlands(e.g.Savannas)alsoexchangelargeamounts fire,andlandusechangeonterrestrialecosystemsfunctions(Jain
ofwaterandenergywiththeatmosphere,andthelossoflargeareas etal.,1996;JainandYang,2005;Jainetal.,2006).ISAMhasalso
of Amazonian forest due to land-use and land-use change could beenusedtodescribethedynamicoftheterrestrialbiospherecar-
impactlocalandregionalclimates(Ramankuttyetal.,2007). bon(Jainetal.,1996;JainandYang,2005)andnitrogen(Jainetal.,
GiventheimportanceoftheAmazonbasintowardtheregional 2009;Yangetal.,2009).ItwasapartoftheIPCCassessmentsof
andglobalbudgetsofcarbon,energy,andwaterfluxes(Andreae futureclimatechangescenarios(Schimeletal.,1996;Prenticeetal.,
etal.,2002;CostaandFoley,2000;Foleyetal.,2007;Werthand 2001).ThelatestversionoftheISAMhasalsoparticipatedinthe
Avissar,2002),aninternationalscientificendeavorheadedbyBrazil ModelingandSynthesisThematicDataCenter(MAST-DC)studyas
158 B.El-Masrietal./AgriculturalandForestMeteorology182–183 (2013) 156–167
partoftheNorthAmericanCarbonProgram(NACP)toquantifyand
Reference Araújoetal.(2002) Hutyraetal.(2007) DaRochaetal.(2004), (2004)etal.Miller VonRandowetal. (2004)Bormaetal.(2009) DaRochaetal.(2002) Andreaeetal.(2002), VonRandowetal. (2004)Sakaietal.(2004) erucsseiiatntn sreadkobTsle sl.oh ,utr, n eo2sat ,t i0nr loaae1ndnnng2i ditida;nro noRvsncpgdeioaceanlmah/ntttci ao,pao rmrloedni ennaustsineoeln tnbrndiegpyt ne yt losetet, y afma atnltle nps.ht,mchd oe2ae r p0lIwaseSo1ilszA ar2 dai(tM;nHeil sSgrr utc u erNnhflissbtAauoezuedlxCiuftne Peiti gso nird oen eaantrsttth se a oait0 sral faa .. n5,scln .2td◦a,gu 02 ×rmi db1n0 y0o21og .n)d25sf. i;er◦sm oKlossmeuu,p eflra rahncotteaaiemaslnsfl,
hourlytoyearly.TheISAMhastwomaincomponents:(1)abiogeo-
05 04 03 02 06 03 01 chemic al module wh ereso ilc arbo n,litt erproduction ,re sp iration,
000 0 000
ars 02–202–201–2 00–2 04–202–299–2 01 pesrtoidmuactteivdit(yFi, gc.a1r)b(oJnai, nanetda slt.,o1c9k9s 6f;oJra diniffaenrdenYta vnegg,e2t0a0t5io;nY apnogolest aarle.,
Ye 202020 20 202019 20 2009)and (2)a bi ogeop h ysic almod ule repr esent ingsun lit-sh ad ed
photosynthesisschemes(Daietal.,2004),energy,andsoil/snow
hydrology(Songetal.,2013;Olesonetal.,2008;Lawrenceetal.,
mate yseason:July–September yseason:July–November yseason:July–November yseason:June–August yseason:May–September yseason:April–September yseason:May–September yseason:July–December 2maaoctiynnf0v acdd1tiTrlth 1eyyvhSe) oaOme.( tFnl IMIwanoiSdgtd Ae i.aspl unMe1idozltc)deayo oi( its lttnYeis hioc daoia(gonnanFthm rg ,ia etgy Isp esS. mf totwoA1i saamr)Meee.nl l.das, Ela b tt 2gaaosei0ceso,fs Ch 0d gsnn3 e9 eigoiv t;oatrr rercJni oiadnuhfidgi nenvc ecCmma enee4tgl tia illges o inaritmtnsrlaaa. y,,gsit o td is2eahocee0dtcnasni0 us,o li p9Ca ptnaorn)3is .ieodd fi adoly cnsc nn abdoa tapyn vCimo led4aan itn tecc, y itr llgp neoephaeiparttssclor ta hlo(d inpoitguntdnerecgisne--r,,
Cli DrDrDr Dr DrDrDr Dr C3andC4pastures,shrubland,tundra,urban),andbaregroundas
describedinMeiyappanandJain(2012).
The ISAM utilizes Farquhar et al. (1980) C3 photosynthesis
model as implemented by Collatz et al. (1991), and the Collatz
m) et al. (1992) C4 photosynthesis model. The stomatal conductance
y( modelisavariantoftheBall-Berrystomatalconductancemodel
Canopheight 35 55 35 35 16 10 0.5 0.6 c(Boanldl uetc taal n., c 1e9a8n7d; Cpo hlola ttozs ye tn athl.e, 1si9s9f1o)r . sTuhne aISnAdMs h caodmedpuleteavs esst owmhaetrael
leafareaindex(LAI)isusedtoscalethosevariablesfromtheleafto
thecanopylevel.GPPismodifiedthroughthefeedbackofnitrogen
W) availability on carbon assimilation, and is obtained by dynami-
◦on( 0.21 4.96 4.97 1.93 0.16 2.36 2.36 4.89 1ca9l9ly2 )c.omparing plant nitrogen demand and supply (McGuire et al.,
L 655 6 566 5
GPPisallocatedtoleavesandtheleafNPP(GPPminusleafmain-
tenanceandgrowthrespirations)isallocatedtostem,androots.
Maintenancerespirationiscalculatedseparatelyforleaves,stem,
◦Lat(S) 2.61 2.863.02 10.08 9.82 21.62 10.76 3.02 aatnudr ero doet p(ceonadresnet a Qnd10 fifnaec)t obra s(eAdro orna ,S 2it0c0h3 e)t. aTlh. (e2 g0r0o3w) athn dr eas tpeimraptieorn-
foreachpoolisassumedtobeafixedpercentage(25%)ofthediffer-
encebetweenGPPandmaintenancerespiration(Sitchetal.,2003).
Equations describing carbon allocation, maintenance respiration,
est estest est and phenology key processes that determine NPP, litter produc-
dy. Biometype Evergreenbroadleaffor forEvergreenbroadleaf Evergreenbroadleaffor Evergreenbroadleaffor Semideciduousforest Woodysavanna Pasture Pasture thtcBinhieoaoceovenlTsueur,eh s adn( e flu2ebo upt 0itoao xh0y tme5eewrsno)te apoibsant hlene(o eitWrdgchn ysreW hdt e iirIosmtShepecAsip uitserMelm aet f t maeaaeictolrn e.ten,ato n e1,lgr t.ad 9 ie(nvt,9d1 haed7 9anni )an9td bi a7 tnthonh)r ,eviAde gnew pgIt cSapheietnAereh tsdnM h u dsbse soeiie esxemnl qoboAeeuwafs. ascmm egetidrnoiono cdnouieimsnfin d fcduAoea mrrbsto ciih oorrLaenmiAb rsaIbia nnttasdoogs-
stu indicate the start of the leaf fall period. The phenology for decid-
nthis uthoeusst faorrteosft tehceogsyrostweimngs hseaass foonu,rn sotramgeasl:g lreoawf othn,sleeta wffhailcl,ha ninddaicdaoters-
usedi BAN) Tmhaen te,v werh girc ehe inn dfoicraetsetss na ore leaalfw par yesseonncen (oTrambalel Ag 2ro owf tthh.e Ta hpepeh ne drbixa)-.
Table1 oftheeddyfluxtowersitesList Sitename Manaus(Km34) (Km67)Santarém Santarém(Km83) Jarú(RJA)Reserva JavaesRiver-BananalIsland( ReservaPe-de-Gigante(PDG) FazendaNossaSenhora(FNS) (Km77)Santarém tnucawedhennyooheodvnutAi ui epacsprls hnmholfh oo tbenayiecrincmrsaoreeoi tvmstoe ilthteoonlhseoegng.tes geayd U stihl to c atanocaoratmlo vmili aoa kneecnanne do tano dihndtsttt hrdi e ht sfoeaihcretnteotiaerid somg p iifi un nepsh ro setaosttu rruhntanims rgtredo henfg ilto toreemh.pergi r eTegereieysohch dtpta de tm sh (wl uA, te vsarchhrnteiaeoeanegonsrgrntelsabaeol ti ciaisatgatneic tty ndor eeil dotnuesroat na ruBfiafiv nrasnogopel geeelemsosr nc atoa, o o hvn sol2ssfie dn0lr y tiooe0rhpsSnn t5ergOs gemo) trdC m.aadoee.gswusnc Tet cittd hhdeohaoees-fl,
B.El-Masrietal./AgriculturalandForestMeteorology182–183 (2013) 156–167 159
Fig.1. SchematicdiagramofallreservoirsandflowsofcarbonandnitrogenintheISAM.
allocationofcarbontothesevenvegetationpoolsisdependenton are available until that month. The Javaes River–Bananal Island
light,waterandphenologystagesofthecanopyandfollowsthefor- (BAN)siteisasemi-deciduoussiteanditresemblesthefeatures
mulationsofFriedlingsteinetal.(1999)andAroraandBoer(2005) ofevergreenforests.Hence,wetreatthissiteastropicalevergreen
(seeAppendixA),whereduringleafonsetcarbonisallocatedonly sitesimilartoothermodelingstudies(e.g.Poulteretal.,2010).The
totheleaves.Duringnormalleafgrowth,carbonisallocatedtothe readersarereferredtodeGonc¸alvesetal.(2012)fordetailsabout
sevenvegetationcarbonpoolsanditceasestotheleavesduring thesiteusedinthisstudy.
leaf fall. The allocation of carbon to the vegetation pools is con- Toimprovethemodelestimatedcarbonpoolsizesandfluxes
strainedduetoadverseeffectsoflimitedavailabilityoflight,water, fortropicalforestsites,thecarbonallocationfactorsandthemain-
andnitrogen.Accordingly,theallocationinrootscanhelpaddress tenancerespirationrates(valuesareprovidedintheappendix)are
thelimitedavailabilityofwaterandnitrogen,whileinvestmentsin calibrated based on LBA site data. Following Malhi et al. (2011),
stemstendtoincreaseaccesstolight. theallocationfactorsareadjustedtohave35%ofthecarbonallo-
Thecarbonallocationprocessisconstrainedbythreeconditions. catedeachtothestemandleafpoolsand30%totherootspools.
First, for the cold deciduous trees all carbon is allocated only to Thecalibratedallocationfactorsaredynamicandtheirvaluescan
theleavesduringleafonset.Asecondconditionisthatsufficient changewithtimedependingonwaterandlightstressfactors(Arora
woodybiomassmustbeavailabletosupportthemassofleaves.A andBoer,2005).Studiessuggestthatmaintenancerespirationrates
thirdconditionistoassurethattheplantstructureispreserved, aredifferentfordifferentcarbonpools(Amthor,1984,2000;Ryan
meaningsufficientrootandstembiomassmustbebuiltpriorto etal.,1995).Therefore,differentmaintenancerespirationratesfor
supporttheleafbiomass. leaf, stem, and, root pools for each biome are introduced in the
The model estimates litter fall from leaves, stem and roots, maintenancerespirationequations(TableS2).
whichisafunctionoftemperature,waterstress,andturnoverrates. TheISAMparametersforcarbonallocation,maintenancerespi-
Theturnovertimescaleforleavesisbiomedependentandvaries ration,andturnoverratesareparameterizedusingthemeasured
from0.75yearforsavannato2yearsforevergreenforest(Foley data from tropical biome site Km34. We optimize the model by
etal.,1996).Theturnoverratesforthestemvaryfrom2.5yearsfor usingtrialanderrormethodtofindparametersthatcanachieve
savannato50yearsforevergreenforest(Malhietal.,2004)and, theminimumabsolutedifferencebetweentheobservedandthe
between1and8yearsfortheroots. estimatedvariables.Theoptimizationisdoneforeachbiometype
becausetheparametersarebiomespecific.Inaddition,measured
2.2. Modelcalibrationandspin-up parametersthatareavailablefromtheliteratureareuseddirectly
withoutopt imiz atio n.Weeva luate theo ptimized mo delus ingfour
Inthisstudy,weusedatafromeightLBAfluxtowersites:cov- other tropical sites (Km67, Km83, RJA, and BAN). Similarly, we
ering four tropic al e ver green fore st site s, on e tr opical decid uous calibr atethea bove mention edmod elpa rame tersfo rtheFNSp as-
forest ,one savanna ,onepastu re,and onep astu re/agric ulturalsite turesite and tested themodel usingt heotherpa stur esi teKm 77.
(Table 1).T heSanta rém Km77w aso rigin allyapasturesitef rom Also , we run and e valu ate the mod el p erform ance fo r a woody
Septem be r200 0untilNo vembe r200 1,butwa sc onverte dto arice savan nas iteP DG. Wecould not validat eitwithother LBA s avanna
siteinFebr uary2 002( Sakaietal. ,2004 ).W esim ulatethis sit e asa sites, du e to the lack of da ta. H owever , w e co mpar e ou r model
past ur e(January 2001 –Nove m ber 2001), disc ontinuin g.th emo de l result swi tho the rpub lish edm easuredst udie s(Gracee tal., 2006).
calculat ionsafter November2001 becaus ethefluxmea sure ments Finally, wea lsoev aluateand compare ourmod elresu lts wi thsite
160 B.El-Masrietal./AgriculturalandForestMeteorology182–183 (2013) 156–167
Table2
Summ aryofISAMestimatedabovegroundandbelowgroundcarbon(KgCm−2yr−1)reportedasthemeanforthestudysitesfluxmeasurementperiod.
Km34 Km67 Km83 RJA BAN PDG FNS Km77
Cleaf MISAeaMsu rement –0.7 –0.6 –0.7 –0.6 –0.6 00..42 –0.4f –0.2 –0.1
Cstem MISAeaMsu rement 1–7.2 1–2.6 1–6.9 1–6.2 1–4.7 00..57 –1.2f –– ––
Croot ISAM 2.9 2.2 2.8 2.7 2.4 1.0 1.2 0.6
Measurement – – – – – – – –
AGB ISAM 17.9 13.2 17.6 16.8 15.3 0.9 0.2 0.1
Measu rement 16.5 ±5a 14.8 ±3a 15.6 b18.5–20.4c – – – – –
BGB ISAM 2.9 2.2 2.8 2.7 2.4 1.0 1.2 0.6
Measu rement 3.8±0.6a 1.8±0.7b – – – 0.7 –1.1d – –
TotalBiomass ISAM 20.8 16.4 20.4 19.5 17.7 1.9 1.4 0.7
Measu rement 20.3 ±5.6a 16.6 ±3.6a 19.9 –21.8±0.2a – – 1.7 –2.3e – –
a MeasurementsdataarefromMalhietal.(2009),unlessindicatedbyaletter.
b Measurementf rom Sale skae tal.(20 03 ).
c Measurement from Millere ta l.(2 004).
d Measurement from DeCas tro an dKaufman(1998).
e Measurement from Da Rocha etal .(2002).
f Measurement sfrom Gr aceet al. (2 006).
ecophysiologicalmeasurementsforNPP,autotrophicrespiration, Table3
heterotrophicres piration,bioma ss,s oilca rbon,andlea flitter. ISAM estimated soil carbon (Kg C m−2) for the LBA study sites reported as the mean
forthestudysitesfluxmeasurementperiod.
Themodelisinitializedusingsitelevelsoilpropertiesandsite
meteorological data and with the absence of CO and nitrogen Soilcarbon(0–1m)
2
disturbance.Thespin-upisachievedbyrepeatingthesitemete-
orologicalda tafo rtheran ge ofavailab ley earsuntil the carb onand ISAM Measurement
Km34 12.8 12.7a
fnoirtrtohgeenso pilocoalsr broenac.hN ethxet, sttheeadmyo sdtaetleis, wruhnicfho rtaekaecsh asbioteutfo 7r0t0h yeesaitres KKmm6873 1122..62 172..80a±1.1–11.2±1.3b
measurementperiod. RJA 10.0 –
BAN 10.2 –
PDG 8.3 8.3–8.9c
3. Results FNS 10.3 10.0–12.0c 10.0-10.8d
Km77 7.2 –
Thisstudyusesanetworkofmeteorologicalandclimatedata
a MeasurementsarefromMalhietal.(2009).
tfrhoemc aArbmoanzoflnu xseistews ittoh drerlievvea tnhteo bISsAeMrv attoio cnaslcoufleactoe saynstde mcoflmupxaerse. debptMh.easurements are from De ca rva hlo Conceic¸ ão Telles et al. (2003) for 1 m soil
Specifically,thisstudytakesadvantageoftheLBAprojectupdated c MeasurementsarefromFishersetal.(1994)forSavannaandpasturesitesin
database fro m e ight fl ux tow ers across a gra dien t of eco systems Colombia.
d MeasurementsarefromTrumboreetal.(1995).
withinAmazon.TheISAMsystematicallycalculatescarbonfluxes
andstorageintheAmazonvegetationandsoils.
3.1. Biomass
The ISAM calculated aboveground biomass (AGB) ranges
betwee n 0.2k gCm−2yr−1 for pasture a nd 17.9kg Cm−2 yr−1 for
tropical evergreen forest (Table 2). The AGB for the Km34, the
Km67,andtheKm83sitesarewithintherangeofthesitemea-
sureddata.TherearenomeasureddataavailableforAGBforthe
RJAandBANtropicalevergreensites,buttheISAMestimatedAGB
for this site falls within the range of Km34 and the Km67 sites
(Table3).Thepasture(FNSandKm77)andthesavanna(PDG)sites
storemorecarbonintheroots(Croot)thanintheabovegroundand
range from 0.7to1 .2k gC m−2y r−1(Ta ble2 ).B elo wgrou ndbiom ass
(BGB) for the Km34 and Km67 sites is estimated to be between
2.2an d2. 9kg Cm−2y r−1 withBG Bfor th eKm34sit eb ein gslightly
lowerthanthemeasuredvalue(Table3).Asexpected,carbonin
forest ecosystems is mostly stored in the stem (Cstem), whereas
for grassland in the roots (Fig. 2). On the other hand, ISAM esti-
ma ted total b io mas s for t he K m8 3 si te ( 20.4kg Cm−2 yr−1) was
within the range of measured total biomass (Table 2). The ISAM
modeled total biomass for the other tropical sites ranges from
16.4 to 2 0.8kg Cm−2yr −1, whic h is w ithin th e me asured range
of values of total biomass at the Km34, Km67 and Km83 sites
(Table 3). The modeled total biomass for the savanna (PDG) site
is1.9k gC m−2 yr−1,whi chfal lswithin the mea suredva lueoft otal
bi oma ss ,w hiletotal bioma ssfo rthepa stu resites(FN Sand Km 77) Fig.2. ComponentoftotalbiomassforeachsiteinkgCm−2yr−1 groupedwith
are1.4an d0.9 kgCm −2yr−1 (Ta ble 2),respe ctive ly.Th eva riation resp ec ttosoiltypes .
B.El-Masrietal./AgriculturalandForestMeteorology182–183 (2013) 156–167 161
inthetotalbiomassforthetropicalevergreensitesisduetothe
factthatthesoilcharacteristicsvarywithsite.Thesiteswithclay
soil have higher biomass than the sites with silt or sandy soils.
However,thetopicalevergreensiteKm67isexceptional,whichhas
beendisturbedoverthetimeandthereforehaslowertotalbiomass
comparedtootherLBAforestsites(Fig.2).
3.2. Soilcarbon
Themodelestimatedsoilcarbonforthetop1msoildepthranges
betwee n 7.2k gCm−2 (K m7 7 site) and 12 .8kgC m−2 (Km3 4 site)
(Table3).TheISAMestimatedsoilcarbonfortheKm67siteisonly
4.5%higherthanthesitemeasurement.Thesoilcarbonestimates
for the RJA and BAN forest sites are within the range of the site
measurementforKm83.Theestimatedsoilcarbonforthesavanna
site(PDG)isatthelowerendofthesitemeasurement.Theesti-
mat edsoil ca rbo nfo rthep astur es ite( FNS) (10.3kgCm−2) ish igher
thanthatofthesavannasite,butthemodelestimatedsoilcarbon
forbothofthesesitesarewithintherangeofmeasuredsoilcar-
bon (Tab le 3).The past ure site(Km 77) soilca rb on(7.2KgC m− 2)is pFiegr. y3e. aSrcfaotrteera cphlosti tfeorin thgeC rmel−a2tiyorn−s1h.iTph beestowliedelnin IeSAreMp reessteimntasttehde G1P:1P lainnde. flux GPP
lowe rthan the oth erpastu resi te(FNS) wh ichcan bere lat ed todi f-
ferenceinsoiltypesandtextures(Andreaeetal.,2002;Kelleretal.,
2005)th at det ermin esth eamoun tofcarbo n sto redin theso il. Similarly, the ISAM estimated autotrophic respiration (Rauto)
(thesumofleaf,stem,roots,andgrowthrespiration)forthetropical
ever gree ns ites varyb etwee n1 .5and2 .0kgCm−2y r−1 ,wh ichfall
3.3. GPP
withintherangeofmeasuredRautofortheKm34andtheKm67sites
(Table4).Thesavanna(PDG)siteandthepasturesite(Km77)have
Overall, the ISAM GPP results are in good agreement with mea- thelow es tRau to (0.5k gCm− 2yr− 1),w hil epastur e(F NS)site Rauto
sGuPrPemvaernytbse (tTwaebelen 42)..6 Faonrd tr3o.0pikcgalC emve−r2gyrre−e1nc soitmesp,a mreoddteol fleustximtoawteedr (1.0 kg C m− 2yr−1) is alm o st half of t he fore st sites, which cou ld be
mea sured valueso f2. 7–3. 1kg C m −2yr−1(Table4).Th eG PPf orthe attributed to lower GPP compared to the evergreen sites (Table 4).
savannaa ndpast ur esitesa re lo werthanthefo res tsit es,b eca use Rleaffor the tropical evergreen sites is about third of the total Rauto
(Fig.4).TheRrootforthepasturesites(FNS)isthehighest,whichis
herbaceousecosystemsarelessproductivethanforestecosystems.
TheISAMes timatedGPP for the individuals itesis withi n15%offlux about 50% of the total Rauto, while Rleafleaf respiration accounts for
about10%ofthetotalRauto(Fig.4).Themodelisabletocapturethe
measurementvalues.Whenconsideringthelargeuncertaintiesin
themeasured fluxGP P(±5– 20%)thatar ere lated toerrorsint he variability in the maintenance respiration for each of the vegeta-
tionpoolofforestandherbaceousbiomes.Modelresultsshowthat
partitioningofnetecosystemexchange(NEE)intoecosystemres-
theherbaceoussites(especiallyforthepasturesites)havehigher
pirationandGPPanderrorsinenergybalancecloser(Hollingerand
Richards on,2 005 ;Wi lsonet al .,2002) ,themo delGPP estimates are Rroot, whereas for the forest biomes Rleafis the highest (Table 4).
within the u ncerta inty ran ge of the m easu red flu x GP P values. eveTrhgree seinmfuolraetsetds s(0o.i9l –re1s.1pikrgatCiomn− (2Ryhert−ro1)) easrteimalasotews fiothr itnhteh terorpanicgael
TheGPPfortheFNSpasturesiteisabout3timeshigherthanthe
ofmeasuredvalues(Table4),exceptfortheKm67site.TheISAM
otherpasturesite(km77).Oneexplanationforthelargedifference
inpro ductivit ybe tweenth eset wopastures ite sist hatth eproduc- estimated Rhetrofor Km67 is lower than the measured fluxes, while
leaf, stem, root respiration and the Rauto are within the range of
tivityoftheKm77siteislowerduetowaterandnutrientstresses
(Sakaietal.,2004).Therefore,Km77siteLAIisalmosthalfofthat
oftheFNSsiteLAI.
To evaluate the model performance for the yearly changes in
GPP,wecomparedtheestimatedGPPwithsitemeasurements.The
modelperformancesforall,butpasturesite(FNS)foryear2001are
comparedwellwiththemeasurementsforallthemeasuredyears
(Fig.3).WhileexactcauseofunderestimationofGPPfortheFNS
pasturesiteGPPforyear2001isunknown;wespeculatethiscould
beduetoerrorsinsomemeasuredinputvariable.
3.4. Respiration
TheISAMestimatedmaintenancerespirationfortheleaf(R )
(0.4–0. 7kgC m−2yr−1), stem (Rstem) (0.3–0.4kg Cm −2 yr−1 ), alenafd
roots(Rr oot )( 0.3–0.4kg Cm−2 yr−1)fo rtropical ev e rgreensite sare
withintherangeofmeasuredvalues(Table4).Leafmaintenance
respirationratesarehigherthanstemandrootsbecauseofthehigh
leafturnoverrates(0.75–2yearsfortheleavescomparedto1–50
yearsforrootsandstem;TableA1).ThemodelestimatedR is
leaf
lowerforthesavanna(PDG)andthepasture(FNS)sitesthanthe
f(oTraebslte s4it)e.s because of lower LAI (data not shown) and leaf carbon wFiigt.h 4r. eCspoemcptotonesonitl otyf pauesto.trophic respiration for each site in kg C m−2yr−1grouped
162 B.El-Masrietal./AgriculturalandForestMeteorology182–183 (2013) 156–167
Table4
Summ aryofISAMestimatedcarbonfluxes(KgCm−2yr−1)reportedasthemeanforthestudysitesfluxmeasurementperiod.MeasurementsforGPParefromeddyfluxtower
measurements;whiletheremainingvariablesareecophysiologicalmeasurementsfromMalhietal.(2009).
Km34 Km67 Km83 RJA BAN PDG FNS Km77
GPP ISAM 3.0 2.9 2.9 2.9 2.6 1.2 2.1 0.8
Measurement 2.9 3.1 2.7 3.0 2.7 1.3 2.2 0.7
NPP ISAM 1.1 1.4 1.1 1.1 1.0 0.6 1.1 0.4
Measu rement 1.0 ±0.1 1.4 ±0.1 – – – – – –
Rauto MISAeaMsu rement 22..00 ±0.5 11..55 ±0.4 1–.8 1–.7 1–.5 –0.6 1–.0 0.4
Rleaf MISAeaMsu rement 10..07±0.4 00..74 ±0.4 –0.6 –0.6 –0.5 –0.2 –0.2 –0.1
Rstem MISAeaMsu rement 00..44 ±0.1 00..34 ±0.1 –0.4 –0.4 –0.4 –0.05 –– ––
Rroot MISAeaMsu rement 00..45 ±0.2 00..34 ±0.8 –0.4 –0.4 –0.3 –0.1 –0.5 –0.3
Rhetro MISAeaMsu rement 11..10±0.1 11..50±0.1 –1.0 –1.0 –0.9 –0.6 1–.0 –0.6
Rtotal MISAeaMsu rement 32..19 ±0.5 23..50 ±0.4 2.8 2–.7 2–.4 1–.2 2–.0 1–.0
NEE ISAM −0.1 0.4 0.1 0 .2 0 .2 0 .0 0 .1 -0.2
Measu rement 0.5±0.2a1.4±3.8b 0.5 ±0.1a−1.1b – – – – –
NPP/GPP ISAM 0.4 0.5 0.4 0.4 0.4 0.5 0.5 0.5
Measurement 0.34 0.5 – – – – – –
a Ecophysiologicalmeasurements.
b Gasexchangeme asurements.
sitemeasurements.ItishypothesizedthattheKm67sitehasexpe- thepasture(FNSandKm77)siteshavehigherNPP/GPPratioof0.5
rienced mortality disturbance (Malhi et al., 2009; Saleska et al., (Table 4) mainly because of lower GPP and Rauto. Moreover, the
2003), but this disturbance has not been quantified in the liter- savannaandthepasturesitesexperiencethelongestdryseason
ature(MarcosCosta,pers.comm.).Theinabilitytodeterminethe andplantsgrowinginsuchenvironmentalconditionshavestronger
rightdisturbancehistoryoverthemeasuredtimehindersthemodel efficiencytostorecarbonastheyhavelessplantmaterialtosustain
ability to estimate R , because R is expected to increase comparedtoforests(Zhangetal.,2009).Also,thetransitionfrom
hetro hetro
followingdisturbanceduetoincreasedmicrobialrespirationwith drytowetclimateleadstoanincreaseinplantrespirationwhich
vegetationgrowth(Concilioetal.,2006;VargasandAllen,2008). maycauseadecreaseinNPP.
The R for the tropical evergreen site, Km34
(n3o.1m kega Cs umre−tdo2taydlra−t1a)a ivs a vilearbyl es ifmoriltahre too tthh eer mfoereasstusrietde sR,tboutatl.t Th heemreo adreel 3.7. Leaf litter
aseinsttedism th(aTetaebdpl aRes tto4ut)ar.l eiTs(h FcelN oISsSeAa tnMod tehKsemt imm7e7aa)tsesudirt eeRsdto avtaralelufoleorsw tahet reK tmsha3avn4a natnhnaed K(fPomDre6Gs7t) (tThaeTb Khleme5 6m)7.oT sdhiteeell eaedna dflel riataftn elgirteftoe frrr otihsm ec 0lpo.a2sse kt ugto rC et mhsie−t e2m(yFerNa−s1Su)tro(e0 m0.2.3ek nkgtgsC C em mx−c−e22pyytr r−f−o11r)
sites ma inlybec ause oflo werRau to for the setwo sites (Ta ble4). isatth eh igh ere ndof the rep ortedm eas urem ents ,w h ileforthe
teEhvxeea rmIgS rAineMienn ges stti hitmees actoaenmdd Rp aoaunbto eonuistt a(hbRoaa luuftto ftowarnodt th hRe ihredstar oov)fa tnohnfe at Ro t(toPatDla lGrfe o)srp atinrrodaptii tochnael, lotoht wheeem rr opedna esdtl uoerfs ett hisme it areet ep( Kdomrlt ee7ad7f )m litthetaeesr uvfaroelrum eteh (ne0t.st1.w kTogh eCp admsif−tfu2erryee rn−sc1ite)e bsis e ctaawtn etehbnee
pasture(F NSan dKm 77)si tes(F ig.5 ),w hichare consist entw ith
themeasurementdata.
3.5. NPP
The ISAM NPP estimates for tropical evergreen forest varies
betwee n1.0a nd1 .4kgCm−2 yr− 1 which fallswithin thera ngeof
reportedNPPvaluesforthesamebiome(Table4).Themodelesti-
matedNPPfordeciduousforest(BAN),savanna(PDG)andthetwo
pastur esite s(F NSandKm 77)are 1.0,0. 6,1.1,and 0.4kg Cm −2y r−1,
respectively(Table4).
3.6. NPP:GPPratio
The model estimated forest carbon use efficiency (NPP/GPP
ratio)variesfrom0.4to0.5forthetropicalevergreensites,which
issimilartothemeasuredNPP/GPPratioof0.34(fortheKm34)and
0.5(fortheKm67).TheKm67sitehasahighermeasuredNPP/GPP
ratio(0.5)comparedtoothertropicalevergreensites,whichsug-
gestthatthistropicalevergreensiteisaccumulatingmorecarbon
thantheotherLBAtropicalevergreensites.Malhietal.(2009)sug-
gestthattheenhancedNPPfortheKm67siteisduetoashiftin
cthaerbootnh earllotrcoaptiiocanl teov ewrgoroede nprfoodreuscttisointe sb.eTchauessea vGaPnPn ais (sPiDmGil)ara ntdo rFeigsp. e5c. tCtoomsopiolntyepnets o.f total respiration for each site in kg C m−2yr−1grouped with
B.El-Masrietal./AgriculturalandForestMeteorology182–183 (2013) 156–167 163
Table5 fortheKm67sitecouldberelatedtothemortalityhypothesisfor
ISAM estimated leaf litter (Kg C m−2yr−1) for the LBA sites reported as the mean for the Km 67site ass ociate dw ithdrou gh tst ressandt hatcouldh ave
thestudysitesfluxmeasurementperiod.
causedanincreaseintheobservedleaflitter.
Leaflitter The measurement for the Km67 site suggests that this site is
netsourceforcarbon(Malhietal.,2009),whileourmodelresults
ISAM measurements sho wtheo pp ositebe causet he mo delun derest ima tedsoi lrespi-
Km34 0.3 0.25±0.35a
KKmm6873 00..33 00..34bc sraintikofno.r Tchaerb mono.dOeul rremsuodlte sluregsguelsttssa tghraete twheit hKtmh3e4s isteiteec aocpthedys aiosl onge-t
RJA 0.3 – icalmeasurements,butdisagreewiththesitefluxmeasurements
BAN 0.3 – (Ma lhietal.,2009) (Tab le2).Bec ause ofth ela rgeu ncertaintiesin
PDG 0.2 –
FNS 0.2 0.1–0.2d the site measurements, particularly for Rleafand Rhetro(Malhi et al.,
2009; Ryan and Law, 2005; Suseela et al., 2012), and the differ-
Km77 0.1 –
encein them eas urem entsre sultsbas ed on thetwo mea sure ments
a MeasurementistheaverageofthethreeestimatesfromLuizãoetal.(2004).
bc MMeeaassuurreemmeenntt ttaa kkeenn ffrroomm RHiic resc eht eatl . a(l2.0(20 040).4). tmoectohrordesc t(leycoepvahlyusaiotelotghiecaml aondde ledNdEyE fleusxti mcoavtaersiaanncde)t,h ite rise fdoirfefictuhlet
d Measurement taken from Trumb ore et al.(1995). ro leofthee cosystem asa carbon sink orsource .
Thetotalecosystemrespiration,R ,forthetropicalevergreen
total
sitesdecreaseswithdecreasingprecipitationduringthedryseason
relatedtothedifferenceintheLAI(maximumLAIfortheFNSis
5.7m2m − 2co mparedto 3.2 m2 m−2 fortheKm 77s ite) (da tapr o- because soil respiration is constrained by desiccation (Saleska et al.,
2003).BothKm34andRJAsiteshavethesamedryseasonlength(3
videdbytheLBAcoreteam),whichisrelatedtothenutrientand
months),butthedryseasonprecipitationfortheKm34site(35cm)
soilmoistureavailability(Sakaietal.,2004).Forthesiteswithno
repo rtedmea surements (RJA,B AN ,an dPDG site s), weco uldn ot is 40% larger than the RJA site (25 cm). Rtotal for the Km34 is 15%
drawany conclusionsas towh ether them ode lisove rest imatin gor higher than the RJA site, which indicates that Rtotalis driven by dry
seasonprecipitationamountandnotbydryseasonlength.There-
underestimatingleaflitter.
fore, the respond of Rtotalto climate change will be dependent on
theseverityofthedryseasonandthewetseasonlength.
4. Discussion While land surface models suffers from the inability to accu-
rately estimate soil carbon, ISAM soil carbon estimates for the
WeusetheISAMtosimulatethecarbonfluxesandcarbonpools Amazoniansitesareincloseagreementwiththesitemeasurements
inthevegetationandthesoilsforeightLBAsitesrepresentativeof (within4.5%).Indeed,thestrengthoftheISAMisinitsabilityto
threedifferentbiometypes(forest,savanna,andpasture).TheISAM accuratelycalculatesoilcarbondynamics,whichisimportantfor
estimatesforGPParewellwithintheuncertaintyrangeofthesite predictingitsroleinclimatechangeassoilrespirationisexpectedto
measurements. increaseduetoincreaseinsoiltemperature(RaichandSchlesinger,
GPPfortheFNSpasturesiteishigherthantheKm77pasture 1992;Rustardetal.,2002).Itwillbeadvantageoustohavelong
site.OneexplanationforthisdifferenceisthatthesoilintheEast- termmeasurementsofsoilcarbontohaveabetterinsightofthe
ernAmazon,theregionwhereKm77islocated,isbelievedtohave interannual dynamics of this carbon pool since tropical soils are
lower soil nutrients than the soils in the Western Amazon, the suggestedtohaveapotentialtosequesteranadditionalamountof
regionsforFNSlocation(Asneretal.,1999).Moreover,deMoraes carbon(Lal,2002,2004).
etal.(1996)andDias-Filhoetal.(2001)studiessuggestthatpasture TheannualtotalmagnitudeoftheISAMleaflitterestimatesare
growthislimitedbynutrientavailability,suchasphosphorus,cal- closetothesiteestimates.However,whencomparedwithleaflit-
ciumandpotassium.Studiesalsosuggestthatsoilmoisturestress terestimatesfromRiceetal.(2004),theseasonalvariationsdonot
atKm77ishigherduringthedryseasonduetoshallowroots(Sakai match(notshown).TheISAMlitterproductiondonotshowsea-
etal.,2004).ItisinterestingtonotethattheISAMisabletotrack sonalvariabilityfortheKm67sitebecauseleaflitteriscontrolled
down productivity for these two diverse pasture sites (FNS and by cold temperature and water stress and the cold temperature
Km77),becausetheISAMisabletoaccountforwaterandnutrient stresswillnothaveaneffectonleaflitterfortropicalsites(Arora
limitationsunderthevariableenvironmentalconditions. andBoer,2005).BasedontheISAM,thewaterstressfactorforthe
ToinvestigatewhethertheLBAsitesactedasanetsourceorsink Km67siteisalmostuniformthroughouttheseasonandthecurrent
foratmosphericCO ,NEEiscalculatedasthedifferencebetween litterproductionschemehasfailedtocapturetheseasonalityofleaf
2
NPPandR .Themodelresultsindicatethatalltheforestsites litterfortropicalevergreensites(thisisthecaseofthemajorityof
hetro
arenetsinkofcarbon,whilethesavanna(PDG)andpasture(FNS) landsurfacemodels).Wewilladdressthisissueinthefuturestud-
sitesareneutralandonlyKm77isnetsourceofcarbon.Basedonthe iesbymodifyingtheISAMlitterproductionmodeltoaccountfor
modelresults,theKm67hasaccumulatedmorecarbonandhence theseasonalvariationinLAIwhenestimatingleaflitter,insucha
thissiteactedascarbonsink,whichisevidentfromthefactthatthe waythatleaflittershouldincreasewithadecreaseinLAIandvice
estimatedandmeasuredNPPforthissitearehighercomparedto versa.
theotherLBAforestsites.Pyleetal.(2008)suggeststhatthissitehas The improvement introduced to the carbon allocation and
possiblyexperiencedmortalityinthe1990sandre-growingthin autotrophicrespirationschemesareinterconnected,becauseRauto
youngerforesthaveabsorbedmorecarbon(Figueiraetal.,2008). is dependent on the vegetation carbon pool size. By adjusting
Basedonourmodelresultsandmeasurements,ourassumptionis the parameters for the vegetation turnover rates, we are able to
thatthissitehasbeenoccupiedbytheyoungerorsmallertreeswith improvethemodelestimatedtotalvegetationbiomass(Table2).
higherNPP/GPPratio(Deluciaetal.,2007)andlowerRauto com- Furthermore, model assumes different maintenance respiration
paredtotheotherforestsites.Weaccountforthedisturbancesfor rates for the different vegetation pools and for different biomes.
Km67sitebyreducingthemodeledvegetationcarbonby20%atthe Thesemodificationsareintroducedtobeconsistentwithstudies
beginningofthemodelsimulationyear.TheNPPforthedisturbed showingthatleafmaintenancerespirationrateishigherthanthe
case is 1.4 Kg Cm −2yr− 1 (Table 4) , whic h is com pa red well with stem and the roo ts (Ryan et al ., 1996). To our kn owled ge, th ese
themeasureddataandabout50%higherthanwithoutdisturbance implementations are unique to ISAM and help to improve the
case. In addition, the underestimation of the modeled leaf litter modelperformanceasevidentfromthemodelresults(Table4).
164 B.El-Masrietal./AgriculturalandForestMeteorology182–183 (2013) 156–167
5. Conclusion foreststobettercapturetheseasonalvariabilityintheleaflitter.
Furthermore,leaflitterseasonalitycanbeimprovedbyadjusting
WeshowthatbycalibratingtheISAMusingsitedatafortheLBA the seasonali ty of the l eaf carbon p ool th rough the mo dification
sites, t he res ults co mpare wel l wi th the site mea sure me nts and oft hecarbonal loc atio npar ameter sand accounti ngt heimpactsof
withi nth eirunce rtaintyra nge. TheI SAM isc alibratedusing data lig hts tresson thesepa rameters,sp ecifi callyforin tro picaleve r-
forthe Km34 tropicalfor estsite and tested fo rtheothe rfour topi- green fores ts, Finall y, LAI is the main drive r fo r the leaf litter
cal fore stsite sinclude dinth eLB Apr oject.A lso ,ISA Mis calib rated season ality,an dwepla nto int rodu cedyn amicLA Isc hem esto ISAM
for theon epa sturesite (F NS) and validate dfor theo th erpasture toimprovet heL AIs easo na lity.
site (Km 77) .TheISA Mis calibr ated forthesa van nas ite(PD G),but
couldnotbefurthervalidatedduetolackofdataavailability.The Acknowledgments
ISAMestimatedresultsforthemeasuredsitescomparewellwith
thesitemeasurementsthusvalidatingourcalibrationtechniques. Wethankthescientistswhocollectedtowerobservationsdur-
Despite the limitation in the site measurements and data avail- ing field campaigns under the LBA project, funded by NASA,
ability,itcanbeconcludedthatgloballandsurfacemodelscanbe the Brazilian Ministry of Science and Technology and European
improvedsubstantiallybyintegratingsiteleveldata. Agencies. We also thank the LBA-Data-Model-Intercomparison
In this application of the ISAM, we revise and modify the project funded by NASA’s Terrestrial Ecology Program (grant
autotrophicrespirationandthecarbonallocationschemes,which NNX09AL52G),whichprovidedthedatasetsthatmadethiswork
provedsuccessfulasevidentfrommodelresults.Webelievethat possible.TheISAMdevelopmentworkpresentedhereissupported
thesemodifiedschemesandcalibratedparameterscanbeimple- by the National Aeronautics and Space Administration (NASA)
mented in other land surface models depending on their model LandCoverandLandUseChangeProgram(No.NNX08AK75G)and
structure to improve their estimates. We learn from this anal- theOfficeofScience(BER),U.S.DepartmentofEnergy(DOE-DE-
ysis that several improvement for ISAM are needed to enhance SC0006706).
the model capability for the global simulations, such as adding
morebiometorepresentheterogeneityinlandscapes(e.g.woody AppendixA.
savanna),andimprovingthelitterproductionschemesfortropical
TableA1
Modelparametervaluesusedinthisstudy.
Function Units Tropicalevergreenforest Savanna Pasture Reference
Structure Vegfrct – 1 0.80a 0.85a Thisstudy
cnleaf – 42 25 50 White et al. (2000)
cnstem – 300 200 200 White et al. (2000)
Photosynthesis cVncmroaoxt –(cid:2) mole m−2s−1 5985 5904 5804 WDohmitien egtu aels. e(2t 0a0l.0 ()2007)
Plant respiration kkkknlsreoteaofmt –gggCCC ggg NNN−−−111ddd−−−111 0000....5000914 aaa 0000....5000446 aaa 0000a...50066 aa ATTThhhroiiisssr asss tttauuunddddyyy Boer (2005)
Phenology daylen h 12 12 12 Thisstudy
biotr ◦C 12 12 11 This study
Biotrmin ◦C 7 2 2 This study
LAI max m− 2m–1−2 6 4 .50 3 .29 This study
Allocation ω – 0.6 0.8 0.5 Thisstudy
εεεlsre oteaofmt ––– 000...5040a 919aa 000 ..46 0 000 ..46 0 TTThhhiiisss ssstttuuudddyyy
k – 1.6 1.6 1.6 AroraandBoer(2005)
Litter production rrYwtlmemaafaxx ydde−−a11rs 200.. 02005 000...7030505 100.. 01055 TAThhroiissr ass ttauundddyy Boer (2005)
bw – 3.0 3.0 3.0 Arora and Boer (2005)
Turnover rate YbTYctsfrotoeldomt/Ycroot –yy◦Cee aarrss 3552..0.000a /8.0a 3522....0050 / 5.0 3511....0000 /2.0 AATThhrrooiissrr aass ttaauunnddddyy BBooeerr ((22000055))
a Calibratedparameters.
TableA2
ISAMequationsforthecarbonmodules.
Function Equations Reference
Plant autotrophic respiration Rleaf= kleaf×cCnleleaaff × ∅ × gt (S1) Sitch et al. (2003), Arora (2003)
Rstem= kstem×cCnsstteemm × gt (S2)
Rroot= kroot×cCnrroooott × ∅ × gt (S3)
QQgt1100===Q 33(T..v22eg22− −−20 )00/..10004466fo ××r lTTevsaoeigvlesffooarrn lrdeoaosvttees ms asnd stems (((SSS456)))
10
B.El-Masrietal./AgriculturalandForestMeteorology182–183 (2013) 156–167 165
TableA2(Continued)
Function Equations Reference
gt=Q(Tsoil−20)/10 forroots (S7)
Phenology FRRRomGa rutt=doo te a=0lc1.0=i 2dR 5muR otl×eouat afm⎧sl++tar xR eR(seGt0es,m: G +P PR r−oo Rtmtotal) (S((1SS980)))
⎪⎨ julian day > 200
daylength>daylen
leaf onset=⎪⎩ Tr> b io tr (S11)
normalgr⎧owth={GLAPIP≥ −( R0l.e5af×>L A0I(cid:6)max) (S1⎫2)
⎪⎨ j ul iand ay >2 00 (cid:6)(cid:6) ⎪⎬
dayleng th> d aylen(cid:6)
leaf fall =⎪⎩ T r< boior tr min (cid:6)(cid:6)(cid:6)or LAI< (0.4 × LAI max)⎪⎭
GPP(cid:10) − Rleaf> 0
daylength<daylen
dormant.noleaf= (S14)
Forherbaceo u(cid:11)s: Tr< bio trmin
lea fonset= GPP − Rleaf> 0 (S15)
norm algr o⎧wth={TLrA>I> 27(05.5×LAI max) (S16)
⎨GP P − Rle af < 0
leaf fall =⎩ Tr< 275 (S17)
ws<0.4
Allocation NPP = GPP − Rmtotal − RG (S18) Arora
Fortropicalevergreenforest: and
IfNPP>0: Boer
CCC lsreoteaofmts= == N NNPPPPPP × ×× A AAllllolloo f affaacctclttesratofeomts (((SSS122910))) (Fe2rt i0ae0ld.5li)n,gstein
IfNPP<0: (1999)
CCC lsreoteaofmts= == G GGPPPPPP × ×× A AAllllolloo f affaacctclttesratofeom−t− −R l ReRarsoftoetm (((SSS222234)))
Forsavannaandpasture:
Maximumgrowth:
Cleaf= NPP (S25)
Normalgrowth:
Sameastropicalforest
Leafoff
CCClsreoteaofm t=== ( (G(GGPPPPPP −− − RR RGGG−−− RR Rlresoatofe)tm) × )× ×A A lAllollol o faf afcacttclertaosfotet−m− R− Rle raRofosttem (((SSS222678)))
Allocationfactorfortrees:
Allofactst em=1 ε+steωm (+2 −ω(L1− −w Ls)) (S29)
Allo factroot=1εr+ ooωt +(2 ω− (1L −− ww ss)) (S30)
Allo factleaf= 1 − A llo fac t ste m− allo factroot (S31)
Allocationfactorforherbaceous:
Allo factro ot=1ε r+ooωt +(1 ω+(1L −− wwss)) (S32)
Allo(cid:12) factleaf=1 + ωεl(e1af ++ Lω −L ws) (S33)
exp(−kn × LAI) f or tr ees and crops
L = max(0,1 − LAI) forgrasses (S34)
4.5
Litter production ˇrrrLnwttit===t=e r(cid:12)r r(t Ywpml rmae(oxaTafdx×a×1ilr×e (a−31 f(6 1=(−5T − 5 )Ccˇ1.o l0wtl ed)absf−t)×b 5w (.r0n)+) rwTc+old rt>) TTaaiirr>> (TTcocoldld− 5.0) ((((SSSS33335786)))) AaB(2nor0ode0rra5)
Forstemsandroo0t s: (cid:13) Tair≤ (cid:14)(Tcold− 5.0)
1
Litter prodstem= Cstem×(cid:13) Ystem× 365(cid:14) (S40)
1
Litter prodroot= Croot× Yroot× 365 (S41)
Description:variability, the Amazon basin is a net source during the drier and .. site (PDG) is at the lower end of the site measurement. Croot cnroot. × ∅ × gt. (S3). Q10 = 3.22 − 0.046 × Tveg for leaves and stems. (S4) day length > daylen.