CARAIB: a global model of terrestrial biological productivity

GLOBAL BIOGEOCHEMICAL CARAIB: CYCLES, VOL. 8, NO. 3, PAGES 255-270, SEPTEMBER 1994 A global model of terrestrial biological productivity P. Warnant, L. Francois,D. Strivay, and J.-C. G•rard Laboratoirede PhysiqueAtmosph•riqueet Plan•t•ire, Institut d'Astrophysique,Universitede Liege, Liege,Belgium Abstract. CARAIB, a mechanistic modelof carbonassimilationin the biosphere estimatesthe net primaryproductivity(NPP) of the continentalvegetationon a grid of 1ø x 1ø in latitude and longitude. The model considersthe annual and diurnal cycles.It is basedon the couplingof the three followingsubmodels; a leaf assimilationmodel includingestimatesof stomatal conductanceand leaf respiration,a canopymodeldescribingprincipallythe radiativetransferthrough the foliage,and a woodrespirationmodel. Present-dayclimate and vegetation characteristics allow the discriminationbetweenecotypes.In particular, specific informationon vegetationdistributionand propertiesis successfully usedat four levels;the leafphysiological level,the plant level,the ecosystem level,andthe global level. The productivity determinedby the CARAIB model is comparedwith local measurements and empiricalestimatesshowinga good agreementwith a global valueof 65 Gt C yr-1. Thesensitivity ofthemodelto thediurnalcycleandto the abundance of C4 species is alsotested.The productivityslightlydecreases (10%) whenthe diurnalcycleof the temperatureis neglected.By contrast,neglecting the diurnal cycleof solarirradianceproducesunrealisticallyhigh valuesof NPP. Even if the importanceof this increasewouldpresumably be reducedby the coupling of CARAIB with a nutrientcyclemodel,this test emphasizes the key role of the diurnal cycle in a mechanisticmodel of the NPP. Uncertaintieson the abundance and spatialdistributionof Ca plants may causeerrorsin the NPP estimates, however,as demonstrated by two sensitivitytests,theseerrorsare certainlylower than 10% at the globalscaleas shownby two tests. Introduction Continental vegetationplays an important role in the climatic system. Indeed, the hydrologicalcycle is modified by the extraction, transpiration, and storage of soil water in plants. Moreover,photosynthesisof green plants is an important sink of carbon. Atmospheric CO2 is, after water vapor, the most important greenhouse gas, and its assimilation by the biospheremodifiesthe atmosphericreservoirand therefore the heating or coolingof Earth's atmosphere. In particular, globatL climate changesinducedby human activitiesmay be in-. fiuencedby the coupled atmosphere-biosphere system. In this scopethe study of the global carbon uptake by the biosphereis of primary importance. During the last years, global models of carbon as. sirnilationby the biospherebased on empirical para- These modelsestimate biosphericcarbon pool sizesand fluxes. However, the parameterizations they use are calibrated with present-day CO2 levels and climatic conditionsbut may be less appropriate for future conditions. On the other hand, mechanisticmodels have often been applied only at the leaf, plant, or, eventually, the canopylevel. The study describedin this paper ten- tativelyusesmechanistic models[Farquharet al., 1980; Collatzet al., 1992]to predictthe net primaryproductivity (NPP) at a globalscale. Even if this scaling-up method restsupon many simplifications,it is a first step toward the modeling of CO2 assimilation by continental vegetation, and it gives results as realistic as those of previous estimations. Furthermore, improvementsof this model are expectedas physiologicalknowledgeand scaling-upmethodologyprogress.The sensitivity of the model to the diurnal cycle is tested. It is shown that meterizations havebeendeveloped [e.g.,Esser,1991]. ignoring this cycle may introduce important errors in NPP Copyright 1994 by the American Geophysical Union. Paper number 94GB00850. 0886-6236/94/94GB-00850510.00 estimates. Model Description The C ARAIB model has been built to estimate the net primary productivity of continental vegetation at 256 WARNANT ET AL.: GLOBAL MODEL OF TERRESTRIAL BIOLOGICAL PRODUCTIVITY a global scaleusing vegetationinformationand climatic data. A spatial resolution of 1ø x 1ø in latitude CARAIB intendsto be as mechanistic as possible andis basedon the coupling of the threefollowing submodels: a leafassimilation modelincluding estimatesof stomatalconductance andleafrespiration, a and longitudehasbeenchosen because it permitsthe descriptionof relativelyfine spatial variationsof the NPP, while stayingcomputationally manageable.At canopymodeldescribing principallythe radiativetransthisresolution the continents cover15,347gridpoints. ferthroughthe foliage,anda woodrespiration model. The NPP is calculatedindependently at eachof these Photosynthesis is the majorcarbonfluxdetermining grid points. The functioningof the modelis outlined plantgrowth.In thisstudytheemphasis isthusputon in Figure 1, which the diagramillustratesthe various theestimation ofgross primary productivity (GPP).By submodels coupled in CARAIB withtheprocesses they contrast, respiration of leaves and wood is not known consider,the input data they require,andthe timescale accurately. Only a roughestimateof respiration rates (dayor season) to whichtheyapply.In thelowerright willthusbeperformed here,mainlyto provide thevalue ofnetprimaryproductivity andto enable thecomparitext (sections 2.3 and 3.2) for the areaand carbonal- sonofmodel results within situmeasurements (existing of Figure I is a summaryof the symbolsusedin the locationfractionsof the differentvegetationcoversof a for NPP but not for GPP). gridpoint. CARAIB considers the maintwosolarcycles,thediurnalandannualcycles.Because ofthelarge Leaf Assimilation Submodel variationof the photosynthetic rateduringthe day,the Theleafgross assimilation rateA (/zmolm-2 s-1) CO2uptakeiscalculated onanhourlybasis.Thehourly isdescribed bytwoquadratic equations [Collatz et al., values aresubsequently summed upto provide thedaily 1991]' assimilation.However,sincea monthlymeanclimatic data set is used,randomday-to-dayvariationsof wea- therconditions cannotbetakenintoaccount. Thedaily NPP is thusestimatedfor a midmonthdayandmultipliedby the monthlengthto obtainthemonthlyvalue. LEAF LEVEL OAp • - Ap(A1 + A•)+ A1A•= 0 (1) /SA • - A(Ap + Aa)+ A;,Aa =0 (2) CANOPY LEVEL PLANT LEVEL Trendat Wood J Mainten Respiration Construction , I I SOIL HYDROLOGY Wood Biotaare VEGETATION AREA FRACTION ALLOCATION , Airhumicity • water BARE SOIL GROUND Bucket , I -f o C, VEGETATION C4 fo • 0 ( I - f •,) f0• fo, • , TREE LEAVES h o ( 1- { fo ( 1- { ) WOOD (1- h0 )( 1- CARbonAssimilation In the Biosphere ( CARAIB) Figure1. A schematic diagram showing thestructure ofthecarbon assimilation inthebiosphere (CARAIB) model. Theparameters defining theareaandcarbon allocation fractions (see text) ofthedifferent vegetation covers ofa gridpointaresummarized at lowerright. WARNANT ET AL.: GLOBAL MODEL OF TERRESTRIAL where 0 and /• are parameters; A1, A2 and As are functionsdescribinglimitations of the assimilationrate BIOLOGICAL PRODUCTIVITY 257 Finally,the leafnet assimilation rateAn (/zmolm-2 s-1) is givenby (•umol m-2 s-i); andApistheassimilation rateresul& = A- ting fromthe couplingof the firsttwo limitations(/zmol m-2 s-l). These two equations indicate that the assimilation rate is limited by three processes,with a coupling between them representedby the parameters 0 and For Ca species, A1 is the ribulos-biphosphatecar- boxylaseoxygenase(Rubisco)limited rate, mudA2 is the electrontransportlimitedor light-limitedrate [Farquharet al., 1980].They are givenby A1 = Vcmax Pi - F. + + (9) where Rd, the dark respiration rate, is assumedto be proportional to Vcmax. The proportionality constant usedis 0.015for Ca plants[Sellerset al.,1992]and0.020 for C4 plants(estimatedfrom Gollatzet al.'s [1992]value of the dark respirationrate at 25øC). The CO2 pressurein interee!!ularspacesis related to the atmospheric CO•.pressure, p•tm (Pa), by a diffusion equation Pi Patm -- = P P (3) (10) g where Vcma•is the maximum catalytic capacity of Ru- to bisco(/zmolm-2 s-1); Pi istheintercellular CO2pres- whereg, thetotal conductance is given by sure(10-6 Pa);O2istheintercellular O2pressure (Pa); I I 1 F. is the CO2 compensationpoint in the absenceof - = ] (11) g gst darkrespiration (10-6 Pa);KcistheMichaelis-Menten constant for CO2(10-6 Pa); andKois the Michaelis- where gst is the stomatal conductance;and gbl is the leaf boundary layer conductmace. The stomatal conductance to CO•. is estimated fol- Mentenconstantfor O2 (Pa). pi - F. A2= J4(pi +2r.) (4) lowingBall et al., [1987]' 1 whereJ is the potentialrate of electrontransport(/zEq m-2 s-1). V•m•, F., Kc,andKoarefunctions oftemperature, while J is a function of temperature and of the absorbed irradiance. J saturates at a level Jmax at high irradiance. As is the rate limited by the capacityfor the export or A,•hsP gst --•.-.• (g0 q-glPatm-••) (12) where h8 is the air relative humidity at the leaf surface; andF is the compensation point(10-6 Pa). The factor 1.6 accounts for the ratio of the diffusi- vitiesof CO2 and H20 vaporin the stomates[Collatz the utilizationof the productsof photosynthesis [Collatz et al., 1992];go = 0.01(mol•_om-2 s-1) andgl= 9.2 et al., 1991]' for Cs species[Leuning,1990];mudgo= 0.08 (moln2o &= Ycmax 2 m-2 s-1) and gl= 3.0 for C4 species [Collatzet al., (s) 1992]. For low air relativehumidity (hs _• 0.46), gl is decreasedlinearly with the available soil water frac- Similar equationsdescribethe C4 speciesbehavior tion (wn•_o - wp)/(fc - wp) = (hs- 0.1)/0.9 (seesec[Collatzet al., 1992].In this case,A1 and A2 areinde- tion 3.1). This linear decreasetends to simulatethe pendentof p•, and As is a CO2-1imitedrate proportional to Pi observed behavior of the stomatal conductance at low water availability[Mc Mutttie et al., 1992]. This forA1 = V•max Aa = a I As= k pi (6) (7) mulation of the stomatal conductanceinvolves the hypothesisthat the stomates open and closeto optimize the uptake of CO2, while limiting the H20 losses. In the presentversionof the model a constantvalue g• = 0.0714molco•_ m-2 s-1 isusedasa firstapproximation (8) where V•m• is a "Q10function"(Q10 is the factor by which the rate, i.e. Vcmax,is multiplied for each 10øC increase in +•empera•ure• + of temperasure •orrec•ed to limit the assimilationrate at low or high temperature; a is the slope of the photosyntheticresponseto light for the leaf boundary layer conductmace. Equations(1), (2), and (9)-(12) haveto be solvedsimultaneouslyto calculatethe leaf net assimilationrate An. Since A1, A2, and As are not identical functions of pi for Cs and C4 specie, the mathematical •!ution of these equationswill also differ. For Cs, speciesequa- tions (10)-(12) are combinedto give the intercellular pressureof CO2, pi, as a function of the assimilation (molco•_/mO]photon); I is the irradiance absorbed by the 2 1 leaf (•molphoton m- s- ; k is the initial slopeof pho- rate, An. This functionis then introducedin (3) (retosynthetic C02 response (mo]m-2 s-1) andis a spectively(4)), assumingthat A1 (respectively A2) is functionof temperature; and P is the atmosphericpres- the only limiting rate. This procedure leads to two cu- sure(Pa). bic equations,the solutionsof which yield A1 and A2. 258 WARNANT ET AL.: GLOBAL MODEL OF TERRESTRIAL Finally,(1) and (2) are solvedto provideAp, and A, andthusAN by makinguseof (9). SinceAx and A2 are not functionsof pi for Ca species,(1), (2), (6)-(8), and (10)-(12) can be combinedinto a cubicequationdirectly providingthe leafgrossassimilationrate, A (Collatz et al., 1992). A• is then calculatedfrom (9). BIOLOGICAL PRODUCTIVITY canopy,LAIc (assuming,as mentionedabove,that the layerthicknessin LAI is 0.2). To obtain the net primary productivity of the grid point (per unit areaof groundsurface),the respiration rate R•, of the woody parts of the vegetation must be substracted from the total Canopy Submodel leaf net assimilation NPP - LNA- R• rate (16) The CO2 uptake by the green vegetationis directly Consequently,an estimate of the woody respiration related to the net assimilationof a singleleaf, assuming rate is needed before the net primary productivity of that physiologicalparameters are constant throughout the vegetation can be calculated. Woody respirationis the canopy. The canopy is divided into layers of equal very poorly known quantitatively, and a very crude esthickness in leaf areaindex(LAI). The assimilation rate timate is made here only for completenessof the model. of each layer is determined as describedin section 2.1., The approach adopted here is similar to that used by and layer values are added to provide the canopy assiRaichet al., [1991]and McGuireet al., [1992]in their milation. In order to perform this integrationeasily,the calculation of total plant respiration. Maintenance and LAI is reducedto the nearestmultiple of the layer thickconstructionrespirationrates are calculatedseparately, ness,taken as 0.2 in this study. The temperature, relasincethe former is proportional to the biomassand the tive humidity, and CO2 pressureof the air are supposed latter to the net carbon assimilation. Thus R,, can be to be constantthroughoutthe canopy,while light is ab- written as sorbedwithin the canopy. An exponentialattenuation R• = R• + R•, (17) of the solar flux Ir within the foliageis implemented whereR• is proportionalto woodybiomass,and R•, [Sellers,1985] It. = Io exp(-kLL) (13) is proportionalto the part of net assimilationallocated to wood growth. FollowingMcGuireet al., [1992],the increase of/• whereIo is the irradianceat the top of the canopy;L is with temperature is representedby a "Qm relationthe cumulative LAI; and kr.is the extinction coefficient. ship," in which Q•o is, itself, temperature dependent As leavesare assumedto be sphericallydistributed, and is calculated from a third-order polynomial fit to kLis given by observations. kr. = (1 - •)o.• (14) 2p where p is the cosineof the zenith angle of the solar So we have 10 5 [j•o Tln(Q•o) T] (18) R• m= KrBwexp where B•ois the woody phytomassin standingvegetation; Kr is the respirationrate per unit massof woody Net Primary Productivity of a Grid Point material at 0øC; and T is temperature in øC. Sinceonly the living part of the woodyphytomassrespires,K• Usingthe leaf and canopysubmodelspresentedabove, must be substantiallylower than the valueslisted by the total leaf net assimilation(LNA) rate per unit area McGuire et al., [1992]for averagevegetation(leaves of ground surface at a model grid point is calculated pluswood)in differentecotypes.In the absence of prefrom the relationship cise measurements,the K• value has been chosenhere in sucha way that the globaland annual averageof the LNA= fo , • A•(C4) (15) respirationrate is comparableto the valuereportedby beam;and •v is the scatteringcoefficient(0.175). [f•Lc Harvey[1989]. The woodyphytomass B•ois not cal- L=I + (1- fc,) L=I ] culated explicitly in the model. Rather, it is estimated from a simpleparameterizationlinking the annualmean NPP andthe phytomass [Esser,1984,1991] B•o= 0.59181NPP•, •%o.7v2•6 (19) where thesummation extends overcanopy layers, A,r (C4 andA•(C3) arethe net assimilation of layerL for C4 where •"o is the mean stand age of woody material, and C3 species calculatedfrom (9) with solarirradiance and NPP,, is the annual mean NPP allocated to wood derivedfrom (13); fc4 is the fractionof vegetationus- growth. This annual mean value of B•o is used to ing the C4 photosynthetic pathway at the grid point calculateR•, assumingthat B•o is roughly constant considered;fo is the fraction of the soil surface cove- throughoutthe year. The value of •% dependson the red by vegetation; and Lc, the number of layers in the ecotype.Note that sincethe annualmeannet primary canopy,is determined from the leaf area index of the productivity of the wood NPP,, is not known until the WARNANT ET AL.: GLOBAL MODEL OF TERI•STRIAL calculationis performedoverthe wholeyear, it will be necessaryto use an initial guessof Bw while starting an iterativeprocedure(seebelow). The NPP allocated to woodgrowthis relatedto total NPP by the simple relationship NPP• = (1 - H) Nee (20) where the herbaceous factor H is also a characteristic of BIOLOGICAL PRODUCTIVITY 259 adoptedbecauseit avoidsan explicitcalculationof the biomass from mass conservationequations. This sim- plifiedmethodneglectsthe seasonality of carbonallocation (parameters H and h0) associated with the phenologicalchanges,but it, nevertheless, allowsa correction of the net assimilationfor wood respiration, so that the model NPP can be compared with average measurements in the major ecotypesof the world. the ecotypeand is calculatedhere from the fraction• of the vegetatedsurfacecoveredby groundvegetation Input Data (asopposedto trees)as follows: Z = •+(1-•)ho (21) where h0 is the Ëaction of tree NPP allocated to leaf growth. Usinga similarapproach to that of Raichet a/.,[1991], we assumethat the rate of constructionrespiration/• is given by R• = 0.2NA,, NA,, >0 R•,=0 NA,, _<0 (22) NA• - LNA - H x NPP - R•TM (23) where is the net assimilation(NA) left for woodgrowthand wood constructionrespirationwhen the amountsallo- catedto growth(H x NPP) of groundvegetationand tree leavesand to woodmaintenancerespiration havebeensubstracted fromleafnet assimilation (LNA). By substitutingNPP with its valuederivedfrom (16) and (17) and rearanging,(23) canbe rewrittenin the equivalent form NA• -- (1 - H) (LNA- R•) + H R•, (24) Introducingthis valueof NA• into (22) and solvingfor /•w, it becomes R•= (•-•) (LNAR•) (LNAR•)>0 R• = 0 (LNA- R•) _<0 (25) Whenfo, fc,, LAIc, •, andh0(characteristics of the ecotype)are known,(15)-(18)togetherwith (21) and (25) canbe solvedsimultaneously to yieldthe net primary productivity(NPP) on an hourly,diurnal,monthly, or annualbasis(dependingon whichtime interval A• and /• are calculated), providedthat an initial value of B•o is known. As mentioned earlier, in this first versionof the model the woody phytomassB•ois assumedconstant over the year and is estimated from (19) and (20), that is, fromthe annualmeanNPP. As a result, the calculationsof NPP and B•omust be repeated iteratively until convergence.In this way the model is capableof estimatingthe NPP from hourly to annual timescales. The method used to calculate the woodrespirationrate is very preliminaryand has been The model requirestwo kinds of inputs, climatic data usedto estimate the photosyntheticrate and vegetation data used to differentiate the ecosystems. Climatic Data All climatic inputs are monthly mean data. Mean temperature T,• and precipitationP,• are given by the International Institute for Applied Systems Analysis database [Leeroans andCromer,1991]on a regulargrid of 0.5ø x 0.5ø and are averagedover the 1ø x 1ø grid elementsof the model. The monthly mean maximum and minimum temperatures,Truaxand Trainare given by spatial interpolation of observationsat about 3000 stations situated all over the world. These observa- tion filesare thoseusedby May et al., [1992],and a linear interpolationis carriedout, ecosystem by ecosystem. The surfaceirradiancesare the monthlymean values for 1989 and come from the International Satellite CloudClimatologyProject (ISCCP) databaseavailable at Goddard Institute for Space Studies in New York [BishopandRossow, 1991].Theseinputsaretreatedto providethe mean diurnal air relative humidity and the hourly valuesof temperature and irradiance. Air Relative Humidity. A simplebucket model of the surfacehydrologicalcycleconsidering one single soilpoolis usedto estimatethesoilwatercontentW•ao. In this model the precipitatingwater which is not evaporatedfills in the soilpool until the field capacityfc is reached. Excessprecipitation leavesthe site as runoff. The potential evapotranspirationrate is calculatedwith a parameterization, dependingon temperature and so- lax irradiance(Turc formula).The thicknessof the soil layer (rootingdepth) dependson the vegetationtype and the soil texture. It variesbetween1.0 and 2.5 m, except for lithosols, for which a value of 0.1 m is assumed. The soil water content is limited to a minimum equalto the wiltingpoint,wp. The fieldcapacityand the wilting point are fianctionsof soil texture. The air relativehumidity h8 is estimatedwith a formulaadapted from Sellers[1983] hs= 0.1+0.9(wH, O (fc - -wp) (26) In this formula,wH,Ois limitedto topto be consistent with the hydrologicalmodel. 260 WARNANT ET AL.: GLOBAL MODEL OF TERRESTRIAL BIOLOGICAL PRODUCTIVITY Temperature. The diurnal variationof tempera- ture is introduced in the model using a rough estimate = (h- 14) parameters arethoseof Collatzet al. [1992]. Canopy or plant characteristics. At this stage 24 ) + AT cos(2•r the leaf area index of the canopyLAIc is fixed by a parameterizationdependingon the monthlytemperature where h is the local solar hour, and A T = Tm.x- Train. Surface Irradiance. in Table1. For C4 species the valuesof the physiological The hourly irradiance /surf (W m-2) is deduced fromdailymeanvaluesandfrom the irradiance calculated at the top of the atmosphere, I,o, (W m-2) Tm[Pitmanet al., 1991] LAIc = LAIm,x- ALAI[1 - f(T•,)] = o <_ooc f(Tm) = 1-0.0016(25-Tin) f(Tm)= I Is,rf(h) = Itoa(h)exp[-kd(h)] (31) 0øC<Tm < 25øC; Tm _>25øC. (28) whered(h), the lengthof the path of the solarbeam throughthe atmosphere(km), is calculatedfrom the solar zenith angle at hour h, assuminga sphericalatmosphere. The monthly mean atmosphericextinction where LAIm,x is the maximum leaf area index, and ALAI is the amplitude of the seasonalvariation of the leaf area index. LAIc is minimum for temperatureslower than 0øC, increasesquadraticallywith temperature, coefficient k (km-1) is determined by comparing the and reaches its maximum ISCCP monthly means of daily irradiancesat the sur- lue LAImax and the amplitude of the seasonalvariation ALAI of the leaf area index differ betweenecotypesdependingon their seasonalbehaviorand their growthpotentiality. Owing to this rough parameterization, some canopy layers may have negative annual NPP. These layersare not considered,and the canopyLAI is lowered accordingly. The valuesused here for LAIm•x and face,Isurf(W m-2), withthe dailymeanirradiances at the top of the atmospherecalculatedfor the middle of the samemonth. Its averagedaily value,usedin (28), is obtained from linear interpolation of the monthly valueso The direct and diffuseparts of the irradiance/air and at 25øC. The maximum va- Iaig are estimated,assuggested, by Gruntet al. [1989] ALAI aretakenfromPitman et al. [1991]andaregiven in Table Iaig = 1.2 {1-exp[-kd(h)]) I•urf = (29) - Finally, similarly to the formulationof Raich et al. [1991], it issupposed that45%ofthedirectand65%of the diffuseradiation are photosyntheticallyactiveradia- 1. The stand age of woody material rw is listed in Table I as a function of the ecotype; these values are de- rivedfrom Esser[1991]. The Ëactionh0 of tree NPP allocated to leaf growth has been estimated from data for some speciesof tropical, temperate deciduous and coniferousforests reported by Bray and Gorham tion (PAR),while0.825(= 1-•) ofthe PARis absorbed [1964],Dajoz[1982],Duvignaud [1971],andSchlesinger by the vegetation[Sellers,1985].The averageenergyof [1991]. The average values obtained are0.30fortropical oneabsorbed PARphotonis 3.6 x 10-x9 J. species,0.27 for temperate deciduoustrees, and 0.20 for Vegetation Characteristics coniferous trees. These values are assumed to be con- stant over the year. This hypothesis of constancy is The ecotypeclassificationby Wilsonand Hendersonmade for the sakeof simplicity, although it is obviously Sellers[1985]is used.It givesthe distributionof con- not correct. For instance, cold deciduousleavesgrow tinental ecotypeson a 1ø x 1ø grid. For each ecoessentiallyduring spring. However,the consequence of type, severalpiecesof informationsaboutthe vegetation this hypothesison the model results discussedin this arerequiredincludingphysiological parameters( V•max, paper are minor. Indeed, h0 is used only to calculate Jm,x),canopyor plant characteristics (LAI, standage of the woodymaterial, the fraction h0 of tree NPP al- locatedto leaf growth),and spatialdistributions(the fraction f0 of the soil surface coveredby vegetation, the herbaceous factor H usedin (19) and (25) to estimate B•, and R•,. The first of these two variables, B•,, is calculated on a yearly averaged basis and thus the mean annual value of h0 must be used, while the the fractionfc• of vegetationusingC4 photosynthe- secondvariable,R•,, is a flux of secondaryimportance tic pathway,and the relative areascoveredby ground for the estimate of the NPP. vegetation(•) and by trees(1-•)). Physiological parameters. Measurements of the maximum rate of carboxylation and of the maximum rate of electrontransport havebeen listed and grouped intobroadcategories by Wullschleger [1993].Thesedata are used to estimate the value of Vcm•xand Jm•x Spatial Distribution. The seasonalvariation of the Ëaction f0 of the surface covered by vegetation is describedwith a parameterization similar to that used for LAIc [Pitmanet al., 1991] fo -- f0,max-- Afo (1 - f(T.,)) (32) appropriateto eachecotype. To do this, valuesrepresentativeof groundvegetationand trees are averaged wherefO,m•xis the maximumfo valueoverthe year. with the weightfactor •. V•m•x,Jm•x,and • are listed The amplitudeA fo of the seasonalvariationof fo is 262 WARNANT ET AL.: GLOBAL MODEL OF TERRESTRIAL generally small, except for cultivated ecotypes,where it simulatesthe cycleof sowingand harvesting. As for LAImaxand ALAI, the valuesusedhere for fo and Afo are taken from Pitman et al. [1991]and are givenin BIOLOGICAL PRODUCTIVITY areas(Sahara,Australianlow-productivity regions,Himalaya)are clearlyvisible;andevensmallergeographical entities suchas the Alps or the Ural Mountains can be seen. Very high NPP values in the United States Table 1. Values of • listed in Table I have been de- and in centralEuropeare due to the presenceof maize crops.Sincemaizeusesthe C4 photosynthetic pathway, (21) are comparable to thosegivenby Esser[1991]for its productivity may be much higherthan that of other similar ecotypes. The fraction • is assumed constant culturesand eventhan that of natural specieswheninthroughout the year. tensivelycultivated.The deserticareasappearto be too For agricultural vegetationthe fraction of C4 species extensive, as, for example, in South Africa. This feais fc4 = 0.5 when the ecotypeis maize or cane sugar ture can be attributed to the hydrologicalmodel which (assuming only 50%of the 1ø x 1ø grid elementis cove- underestimates the soil water content of semidesertic arred with corn or cane sugar, the rest of the cell being eas. Presumably,this low soilwater contentis, at least coveredwith natural grassland)and is fc4 = 0.0 for the partly,linkedto the assumption madein the hydrologiother agricultural ecosystems. cal modelthat the monthlytotal precipitationof a site In the absenceof global data for natural vegetation is distributed uniformly over time within the month. the C4 speciesare very roughly distributed in natural This assumption,obviouslynot correctin semidesertic ecosystemsas follows:fca = 0.1 • between15øSand areas,leadsto too lowdailyprecipitation,implyingthat 15øN latitude; fc, -- 0.1 • to 0, decreasinglinearly be- precipitatedwater can be readily reevaporatedwithout tween 15ø and 30ø; and fca -- 0.0 at latitudes higher filling the soil pool. than 30 ø . The modelglobalnet primary productivityis 65 Gt C fined so that the herbaceous factors H calculated from yr-•, slightly higher thanotherestimates. Forexample, TinkerandIncson[1990]giveNPP rangingbetween45 and 62 Gt C yr-• with a mostprobablevalueof 60 Gt C yr-•. This relativelyhighglobalNPP valueis Results The spatial distribution of the annual NPP calculated by the CARAIB model is presentedin Plate 1. The general features appear realistic; the productivi- ty is high in tropical regions(Amazionianand tropical forests)and decreasesat higher latitudes; desertic certainly due, in large part, to the lack of couplingwith a model of nutrient cycle in the vegetation and soils. This couplingwould certainly limit the productivity of certain areas and therefore decreasethe global annual estimate. , 0 100 - ! 1 O0 - 200 œ'L'• - 300 40O - 50O 500 - 7O0 700 - 900 IlOO - !300 t500 -. ] 50O 900- AnnualNet PrimaryProductivity(gO m-2 y-l) Plate 1. Globalmap of the annualmeannet primaryproductivity(NPP) calculatedwith the C ARAIB model. Note that the color scale is nonlinear. WARNANT ET AL.: GLOBAL MODEL OF TEtLRESTRIAL In Figure 2 the annual valuesof the NPP calculated by our model are comparedwith the measuredlocal NPP used to calibrate the terrestrial ecosystemmodel BIOLOGICAL PRODUCTIVITY 263 [Houghton et al., 1990]scientific assessment of climatic change(Table2). To makesucha comparison possible, we groupedthe ecosystemsof Wilson and Henderson- [Raichet al., 1991;McGuireet al., 1992]coveringa Sellers[1985]into 13broadvegetation types,definedas wide range of natural ecotypes.The climatic data used in this test are thoseof the grid point coveringthe measurement site. The representativeecotypesare chosen to be as closeas possibleto the vegetation types of the measurements. Two different behaviors appear clear- follows(the numbersin parentheses referto Wilsonand Henderson-Sellers ecotypes):desert(70,71,73),tundra (61,62), grassland(30,31,34,36), savanna (32,33,37), shrubland(16,24,27,28,35,39), needleleafforest(10,18), borealand temperatewoodland(11,13,14,17,21),temperatebroadleafand mixedforest(12,15,19,20),tropical woodland(23,26),tropicaldeciduous forest(25,52), equatorialrain forest(50,51),wetland(2,5), and cultivation(4,40-49,80).Threeotherecotypesfor whichwe do not calculatethe NPP (inlandwater, semidesert,human area) are addedto providea list compatiblewith the IPCC publishedone. Unfortunately,eventhe broad vegetationtypes differ between authors. This is obviously emphasizedby the differencesin global areas coveredby each ecotype(Table 3). In particular, it ly in Figure 2; in somecases(twelvegrid points)the estimated NPP is equal to or higher than the measure- ments,in the other cases(sevengrid points)the NPP is underestimatedby the model. In the first category we find the boreal and temperate ecosystems.Only the lessproductivevegetation(arid shrublandor tundra) shows a large discrepancy. If we do not considerthe three grid points that belong to these latter ecotypes, then the mean error is 33%. Regardingthe secondcategory for which carbon assimilationis underestimated,a meanerror of 36% is obtained. Local conditions,in particular, regional climatic ones,can explain, in part, the differences between estimates and measurements. The test of the model results at a global scalewill, for example, show that the calculated NPP of tropical savanna is relatively high, while it is underestimatedin this first test. However, even at a global scale the tundra seems to be less productive than expected from the model, and equatorial forestsmay have a higher productivity than the calculated ones. A further test of our model is performed by comparing our results with empirical estimatesgathered in the Intergovernmental Panelon Climate Control(IPCC) is clear from Table 3 that the ecotypesnamed here as woodlandsare consideredforestsby other authors. This discrepancymay induce some differencesbetweenthe variousestimates.Moreover,it is not alwayspossibleto find a correspondence betweenWilson and HendersonSellers and IPCC's ecotypes,so that we compare our calculated NPP for needle leaf forest with the boreal forest estimates. As already pointed out, the global NPP estimatedby CARAIB is rather high. Consistently, for almostall ecotypesthe calculatedNPP is in the higher part of the spectrumof empirical values. However, the relative NPP variationsbetweenecotypesare similar to empirical ones. Moreover, in almost all cases,the CARAIB NPP is within the rangeof the IPCC valuesand even very closeto the valuesof Atjay et al. [1979].In fact, our estimates differfromAtjay et al.'s valuesby lessthan 20% for sevenecotypes(grassland, savanna,shrub land, needleleaf forest, temperate forest, tropical deciduousand equatorialforest),this differencebeinglessthan 10% for five of them. For deserts 15oo the difference is 3 gC m-2 yr-1. Forfourecotypes (inland water, semidesert,boreal temperate, and tropical 1 ooo woodland)the comparison is impossible,and for woodlands our estimates are relatively closeto the valuesof correspondingforests. Only two ecotypesshowimportant discrepancies,tundra and wetland. The reasons for the differencesin the NPP of tundra may be multi- 5O0 ple. First of all Atjay et al. [1979](asdo Whittakerand Likens[1975])obviously usea definitionof tundramore restrictivethan ours, as can be seenby comparingthe globalarea coveredby tundras(Table 3). Someof our 0 500 1000 1500 meosuredNPP (gC m-' yr-') Figure 2. Comparisonof calculated annual mean net primary productivitieswith local measurementswhich were used to calibrate the terrestrial ecosystemmodel [Raichet al., 1991;McGuireet al., 1992]. tundras are more southern onesand are thus more productive. However,this is certainly not the only reason for the discrepancy,and uncertaintiesin the hydrological cycle,the absenceof nutrient cycles,or the inaccu- racy of physiologicalparametersmay be responsiblefor someof the difference. The problem causedby wetlands is not surprising,sinceCARAIB doesnot take into accountthe specificitiesof suchan ecotype. Nevertheless, 264 WARNANT ET AL.: GLOBAL MODEL OF TERRESTRIAL BIOLOGICAL PRODUCTIVITY Table 2. Net Primary Productivity(NPP) NPP Ecotype Inland 0 1 water 2 3 200 200 125 7 0 4 25 Semidesert Tundra 213 39 63 62 95 69 132 Grassland 341 267 352 388 Tropical savanna Shrub land and interrupted wood 649 458 407 318 787 489 435 359 358 379 487 558 -657 598 720 988 711 1019 550 775 Desert Needle leaf forest 419 Boreal and temperate woodland Temperate broadleaf deciduousand mixed forest Tropical woodland Tropical deciduousforest Equatorial rain forest 558 659 728 711 925 Wetland (bog,mangrove) 483 1350 1143 1037 Cultivation 771 293 425 761 481 354 100 401 401 65 53 60 61 Human Mean area Global values NPP valuesare in gC m-2 yr-1, and globalvaluesare in GtC yr-1. The resultsof the CARAIB model (column0) are compared with estimates citedin Houghtonet al., [1990];[Whittakerand Likens,1975](column 1); [Atjayet 199] (column2); and [Olsonet al., 1985](column3). It is assumed that the needleleafforests are representativeof the boreal forests. Table 3. Surface Area as a Function of the Ecosystem SurfaceArea,106km2 Ecotype Inland water Desert Semi desert Tundra Grassland Tropical savanna Shrubland and interrupted wood Needleleaf forest Boreal and temperate woodland Temperate broadleaf deciduousans mixed forest Tropical woodland Tropical deciduousforest Equatorial rain forest Wetland (bog, mangrove) Cultivation Human area 0 I 2 3 12.0 23.3 2.0 24.0 18.0 8.0 9.0 2.0 24.5 21.0 9.5 12.5 3.2 20.4 13.0 13.6 6.7 20.0 3.8 15.0 8.5 22.5 4.5 24.6 12.8 4.0 12.0 9.5 11.7 12.0 7.0 8.2 7.5 17.0 4.5 10.3 6.0 12.0 20.2 12.4 2.4 7.4 1.4 12.8 1.0 13.4 2.0 14.0 3.5 16.0 2.0 2.9 15.9 [Olsonet al., 1985](column3). It is assumed that the needleleaf forestsare representative of the borealforests. Notethat desertandsemidesert categories of columns 1, 2, and3 includeAntarctica(-• 14x 106km2),a continent not consideredby CARAIB. WARNANT ET AL.: GLOBAL MODEL OF TER_RIgSTRIAL BIOLOGICAL PRODUCTIVITY 265 the mean NPP of paddy rice (actually consideredas correlated with the seasonalvariation of the tempera- cultivation and not as wetland)is 1024gC m-2 yr-1 ture (Figure4). However,in the formulationof Pitman et al. [1991],the leafareaindexis nonzeroin winterfor which is not far from IPCC valuesfor wetland. A part of the error on the estimate of the NPP of wetlandsmay thus be causedby inappropriate vegetation characteristics rather than by a misunderstandingof the specific physiologicalprocesses.This problem applies only to a small part of the continental surface but has to be studied in more detail in the future. The productivity of cultivated areas also appearsslightly too high. Cultivated plants often need addition of nutrients to grow, since they are not naturally the best adapted to climatic and soil conditions. Thus in regionswhere fertilizers are not intensivelyused, CARAIB overestimates the productivity of cultivated areas. Figure 3 illustrates the seasonalvariability of the all nondeserticecotypes. The seasonalvariation of the LAI is illustrated in Figure 5 for the three test sitesjust described. Since in winter the LAI is nonzero, leaves can assimilateCO2, provided that the daily maximum temperature is positive, even when the averagetemperature of the day is negative. For example, in Figure 3 the NPP of the broadleaf deciduous forest in March is low but nonzero, owing to a positive daily maximum temperature and despitethe negative monthly average temperature shown in Figure 4. Sensitivity Tests NPP. The solidline corresponds to a tundra (111.5øW The diurnal variation of CO2 assimilation by plants longitudeand 77.5øN latitude), the dotted line to a broadleafseasonalforest(131.5øElongitude,47.5øNla- is important but difficult to take into accountin a getitude), and the dashedline to an equatorialrain forest neral model of the carbon cycle in the biospheredue to (65.5øWlongitude,1.5øNlatitude). Differentseasonal limitations in computationalcapacities. Neglectingthis behaviors can easily be observed;a zero productivity, except for 3 or 4 monthsfor the tundra, a seasonalproductivity occuringduring morethan half of the year for deciduousforest, and a quasi-constanthigh productivity with small oscillationsfor the equatorial forest. The photosyntheticrate highly dependson temperature and is even equal to zero for temperatures lower than 0øC. The seasonalvariation of the productivity is therefore cycle by the use of daily mean climatic data as input to the model would substantiallyincreasethe computational efficiency. The influenceof such a simplification on the productivity has been tested by comparing the results of four runs summarized in Table 4 and Figure 6. The first run is the standard one discussed in the section4 (run 0, solid line in Figure 6). In the secondrun the hourly temperature is replaced by the 150 • 125- "' 100 - 75- o o•,• 5o- i• 25- Oz -25 0 I I I I I I I I I I I 30 60 90 120 150 180 210 240 270 300 330 365 Day Number Figure 3. Seasonal variationof the net primaryproductivity (gC m-2 month-1).The solid line corresponds to a tundra (111.5øW,77.5øN);the dottedline, to a broadleafseasonalforest (131.5øE,47.5øN);and the dashedline is an equatorialrain forest(65.5øW,1.5øN). 266 WARNANT ET AL.: GLOBAL MODEL OF TERRESTRIAL BIOLOGICAL PRODUCTIVITY 3O 20- o -10 - -:zo-:30 - -4O 0 I I I I I I I I I I I 30 60 90 120 150 180 210 240 270 300 330 365 Day Number Figure 4. Seasonalvariationof the temperature(øC). The solidline is a tundra (111.5øW, 77.5øN);the dottedline is a broadleafseasonal forest(131.5ø E, 47.5øN);and the dashedline is an equatorialrain forest(65.5øW,1.5øN). _ _ _ _ 0 I I I I I I I 30 60 90 120 150 180 210 { 240 I I I 270 300 330 3( Day Number Figure 5. Seasonalvariationof the leaf area index. The solidline is a tundra (111.5øW, 77.5øN),the dottedlineis a broadleaf seasonal forest(131.5øE,47.5øN);andthe dashedline is an equatorialrain forest(65.5øW,1.5øN). WARNANT ET AL.' GLOBAL MODEL OF TERRESTRIAL BIOLOGICAL PRODUCTIVITY 267 Table 4. Calculated Net Primary Productivity for variousRuns of the CARAIB Model Standard Ecotype Diurnal Cycle Sensitivity C4 Distribution Sensitivity Run 0 Run I Run 2 Run 3 Run 4 Run 5 Tundra Grassland 7 213 341 7 191 303 8 253 410 35 333 967 2 213 317 15 215 447 Tropical savanna Shrub land and interrupted wood 649 458 614 403 838 546 2071 1249 463 440 692 528 Desert Needle leaf forest 419 344 424 680 419 423 Boreal and temperate woodland Temperatebroadleafdeciduousand mixed forest Tropical woodland Tropical deciduousforest Equatorial rain forest 558 659 728 711 925 473 569 681 666 880 580 713 924 911 1186 953 1250 2171 2135 2555 557 658 650 684 900 573 666 743 717 926 Wetland (bog, mangrove) 483 442 594 1145 478 492 Cultivation Mean Global values 771 481 65 706 440 59 946 586 79 1889 1260 170 699 433 59 787 513 69 Valuesfor net primaryproductivity are in gC m-2 yr-•, and globalvaluesare in GtC yr-•. Testsof the sensitivityto diurnalcycle(runs1, 2, and 3) and the spatialdistributionof Ca species(runs4 and 5) are shown. In run I the temperature T is set constantduring the diurnal cycle;run 2 usesa 24-hour constanttemperature and a constant nonzero irradiance during the daylight hours; run 3 uses constant values of temperature and irradiance throughout the 24 hours of the day; in run 4 no Ca speciesis introduced; and in run 5 the amount of Ca vegetationin natural ecotypesis highestfrom 0 to 30ø of latitude and decreaseslinearly between30ø and 60ø of latitude. 5 I Lat' I 50.50øN Long' 5.50'E Belgium _ -1 0 I I I I I I I 3 6 9 12 15 18 21 24 SolarTime (h) Figure 6. Diurnal variation in June of the calculatedleaf assimilationfor the upper canopy layer of a mixed vegetationpixel centeredover Belgium. The four followingtests are presented; the standardrun (run 0, solidline), a run without diurnalvariationof the temperature(run 1, dotted line), a run with a 24-hour constanttemperatureand a constantnonzeroirradiance duringdaylighthours(run 2, dashedline), and a run with constanttemperatureand irradiance throughoutthe 24 hoursof the day (run 3, dash-dottedline). 268 WARNANT ET AL.: GLOBAL MODEL OF TERRESTRIAL daily mean temperature(run 1, dotted line in Figure 6). In the third run the daily meantemperatureis used with the solar irradiance evenly distributed over day- light hours(run 2, dashedline in Figure6). Finally,in the fourth run the temperature and the surfaceirradiance are constant throughout the 24 hours of the day BIOLOGICAL PRODUCTIVITY that the Ca speciesextend to higher latitudes, and the fractionof the grid point areacoveredby Ca grassfc4 is givenby fc4 = 0.1 • between30øSand 30øNlatitude;fc• - 0.1 • to 0, decreasinglinearly between30ø and 60ø; fc• = 0.0 at latitudes higher than 60ø. The estimated globalNPP reaches 69 Gt C yr-•, a valueincreased by and are equalto their daily meanvalues(run 3, dash- 4 Gt C yr-• with respectto the standardrun. dottedline in Figure 6). Thesetestshavebeenchosen becauseirradiance and temperature are the main parameters influencing the daily variation of the leaf assimilation. Unfortunately, the diurnal variation of the hydrologicalcycle is too complexto be incorporatedin the model and cannot be tested. Estimated global pro- ductivity is loweredby about 10% whenthe diurnal cycleof temperatureis turnedoff (Table4, run 1). When the solar irradiance is evenly distributed over daylight hours(run 2), the globalNPP increases by morethan 20%. The increasebecomesvery strong(about 160%) whenthe diurnalcycleis completelyeliminated(run 3). Figure 6, showingthe leaf net assimilationof the upper layer of the canopy, helps to explain these important variations. In the standard case, leavesrespire during the night; photosynthesisbegins at sunrisebut is still limited by light availability; and finally, the assimilation rate saturates due to other limitations for midday hours. The reverseprocessoccursduring the afternoon. The saturation level of midday hoursvariesslightly due to temperature variations. When diurnal temperature variationsare neglected(run 1, dotted line in Figure 6), the saturationlevel is constantand somewhatlower. This behavior explains the decreaseof the daily assimilationrate in this case. As the daily mean irradiance is usually high enough, the assimilation of the upper canopy layer is generally not limited by light in runs 2 and 3 (dashedand dash-dottedlinesin Figure 6). The leaf assimilationreachesits maximumduring the whole daylight period in run 2 or during the whole 24 hours of the day in run 3, resultingin an important overestimateof the daily value. For lower canopy layers, light may becomelimiting even at midday,thereby reducingthe differencesamongthe four tests. However, theselayershave only a small contributionto the plant NPP, and the abovediscussionis still valid at plant level. The unrealistic importance of the NPP increasein run 3 will be reduced when CARAIB is coupledwith a nutrient cycle model. However,the generaltendency and the conclusionsreachedhere will remain, showing the needto includethe diurnal cyclein NPP calculation when physiologicalmodelingof photosynthesisis used. The influenceof C4 speciesis oftenneglectedin global modelsof NPP, introducinga subtantialerror (about 10%at a globalscaleforrun4, Table4). Thiserrormay Discussion CARAIB is a new global model of terrestrial biological productivity. It is built to estimate the influenceof possibleclimate changesdue to human activities on carbon exchangesbetween the atmosphereand the biosphere. From this perspective the emphasisis put on photosynthesis,the major carbon flux determining plant growth. An underlying hypothesisof this mo- del is that at any scalestudied(leaf, canopy,plant, grid point,ecotype,or global)the response of photosynthesis to environmental conditionsis largely dependenton its responseat the leaf level. The first needof sucha model is thus to relate correctly the leaf CO2 assimilationto given environmentalconditions. The mechanisticmo- del developed by Farquharet a/.[1980]hasbeenchosen for Cs plants, owing to its ability to calculate the CO2 assimilationof a single leaf from a detailed description of the major physiologicalprocesses.The influenceof C4 speciescannot be neglectedat the global scale. The modelproposed by Collatzet al. [1992],comparable to Farquharet al's [1980]model,is thususedto studythe behaviorof C4 plants.A conductance formulation[Ball et al., 1987]is coupledto the leafassimilation modelsto describe the modulation of the CO2 flow into the leaf in responseto stomatal openingor closure. This combination of models allows calculationsof instantaneousCO2 fluxesbetweena leaf and the surroundingair. C ARAIB can thus study photosynthesisat all timescaleslonger than the characteristictime of the leaf-air exchanges, provided climatic and irradiance conditionsare known and enough computing time is available. These sub- modelsrequirethe knowledge of manyparameters(1) to (12). In its presentversion,CARAIB uses,generally, for theseparametersthe valuesfound in the original papers. Nevertheless,the valuesused for the maximum catalytic capacity of Rubisco Vcma•and the maximum rate of electron transport Jmax differ for the various vegetation types considered. The uncertainties on the valuesof Vcmax and Jmaxare important[Wullscb•eger, 1993]. However,the useof ecotype-dependent values allowsthe implicit considerationof differencesin physiology,morphology,or even leaf nutrient content among be particularly important for tropical and equatorial the major plant types. An improved model would use herbaceousecosystems suchas savannas.The accuracy similar ecotype-dependentlists for the other parameters and would calculate explicitly the influence of factors of the spatial distribution of the Ca speciesdoes not allow an exact estimate of the quantitative importance such as nutrient availability. of this error, as shown by the results of the last test At its presentstage of development,CARAIB scales presentedin Table 4 (run 5). In this test we assume up from the leaf to the canopy,plant, ecotype,and glo- WARNANT ET AL.: GLOBAL MODEL OF TER/tESTRIAL BIOLOGICAL PRODUCTIVITY 269 bal levels to provide global NPP values. This method, though it may be crude, permits the descriptionof all spatial scalesand, in particular, also allows CARAIB to be applied locally or regionally. At the canopylevel it is supposedthat leavesare distributed in a discrete number of layers.Environmental timated from more empirical models. This somewhat high value can be at least partly explainedby the fact that nutrient limitation of NPP is not yet implemented in our model. Nevertheless,the distribution of the NPP overthe continentsor amongthe world major ecotypes appears realistic and also comparableto. the djstribu- conditions(only the solar irradiancein the present study) may vary from one layer to the other, but leaf physiologicalcharacteristicsare kept constants. This tion established latter assumptionis certainly not always verified, and the influence of leaf characteristic variations within the canopyon C ARAIB resultshave to be tested in future studies. The radiative transfer module is very simple but certainly accurate enough for global applications. However, improvementswould result from a better estimation of the extinction coefficientkL, dependingon plant type. The total leaf area index is presently evaluated from a simple parameterization. This estimate will be improved by the addition of an explicit calculation of green and woody phytomassin the model. This calculation of phytomass, consideringthe allocation of carbon to leaves, roots, and wood during the various phenologicalstages,will also be important for the evaluation of the wood respiration rate. Nevertheless,experimental estimatesof this respiration rate are needed to determinemore accuratelyits K•value per unit mass of woody material. The hypothesisused to scale up from the plant to the grid point or ecotype level is that the vegetation cover of a grid point is homogeneous.For grid point coveredwith mixed vegetation the ecotype-dependent from observational data. One characteristic of CARAIB is its consideration of both the seasonaland the diurnal cycles. The latter cycle is often neglectedin existing biosphericmodels for computationalefficiency. This approximationis, of course,justified in empiricalmodelsbasedon parameterizationsvalidated on a monthly or annual basis. More mechanisticcalculations,however,cannota priori neglectthe diurnalcyclewhenthey consider(nonlinear) physiologicalprocessesat the leaf level with characteristic times much shorter than I day. For this reason, CARAIB takes into account the diurnal variation of the surfacesolarirradianceand temperature.A sensitivity test has thus been performed in which the diurnal oscillation of one or both of these environmental variables has been neglected. The results indicate that the model showsa high sensitivityto the diurnal cycle. This sensitivity is especiallyhigh for the variation of surface irradiance,sincethe assimilationof top of canopylayers saturates under relatively low irradiances reached during most of daylight hours. In summary,CARAIB, an initial attempt to scaleup from the leaf level to the globalscale,is rather successful in predictingNPP. Many not well known parameters are, however, needed by the model such as aveparameters(Table 1) are assumedto be representative rage physiologicalparametersof ecotypes,someplant of a "mean plant type." This simplificationis partialcharacteristics, or the spatialdistributionof Ca species. ly corrected by making separate productivity calcula- These parameterswere very roughly estimated in this tions for Ca and C4 species. The geographicdistribustudy. For future applicationsan improvedknowledge tion of the ecotypesusedhere (as are other onessuch of thesecharacteristics is expected. A better descripasthoseproposed by Matthews[1983]and Olsonet al. tion of many processessuch as carbon allocation, for [1985])are realistic.However,a better distributionof instance, is also necessary. In this regard it must be C4 grasseswill complementusefully these data and enremindedthat CARAIB emphasizes the leaf level prohanceone major characteristicof CARAIB, which is the cesses usingrather detailedleaf photosyntheticmodules use of a specificmodel for C4 species.The net CO2 flufor both Cs and Ca plants, but still treats higher lexes between the atmosphereand the continentswill be velsin a very preliminary way, and totally neglectsthe obtained by couplingCARAIB with litter production- role of nutrient cyclesand microbialprocesses.Another decompositionand soil respiration modules. Summary and Conclusions CARAIB is an attempt to scale up from a leaf level, physiological modelto a globalvegetation,primary productivity model usingcanopyand wood respiration submodelsas well as vegetationparameter characteristic of ecosystems.This approachis a first steptoward a mechanisticmodel of th4 terrestrial biosphere,capable of predicting possible future sequestrationof anthropogenic carbon in continental vegetation or soils. The globalmean net primary productivity calculatedby the modelis 65 Gt C yr-•, in the upperrangeof that es- key aspectconcernsthe validation of CARAIB. Here we principally tested the averagebehaviorof the model in all the major world ecotypes,by comparingmodelmean NPP with a representativeecotypeaverage.However,a completevalidationshouldbe performedat every ecologicallevel, from the leaf to the biome, on the basisof both in situ and remote sensingdata. Thesevariousimprovementsare scheduledin our future plans. They are the necessarysteps in the climb toward the construction of a detailed mechanistic model of the terrestrial biosphere, a model to be used to predict the fate of anthropogenicCO2 releasedinto the atmosphere. Acknowledgments. Support for this work was provided by the Belgian National Impulse Program 270 WARNANT ET AL.' GLOBAL MODEL OF TERRESTRIAL "Global Change,"contractGC/12/017 (P. Warnant) and by the Belgian National Fund for Scientific Research(J.-C. G•rard and L. Francis). Fundingfor this researchwas also provided by the European Communi- ty Environmentprogram(contractEV5V-CT92-0119). W. Knorr and W. May, who provided us data on the observedmaximum and minimum temperature, are gratefully acknowledged.We thank F. Robinet, C. Delire, T. Wautelet, A. Della Vecchia, and all of the team at the Laboratory of Planetary and AtmosphericPhysicsfor helpful discussionsand encouragement. References BIOLOGICAL PRODUCTIVITY Leuning, R., modeling stomatal behaviour and photosynthesis of eucalyptus grandis, Aust. J. Plant Physiol., 17, 159-175, 1990. Matthews, E., Global vegetation and land use: New high resolution data basesfor climate studies, J. Clim. Appl. Meteorol., œœ,474-487, 1983. May, W., D. J. Shea, and R. A. Madden, The annual variation of surface temperatures over the World, Tech. Note NCAR/TN-37œ+STR, Nati. Cent. for Atmos. Res., Boulder, Colo., 1992. McGuire A.D., J. M. Melillo, L. A. Joyce, D. W. Kicklighter, A. L. Grae, B. Moore III, and C. J. VSr'dsmarty, Interactionsbetweencarbon and nitrogen dynasnicsin estimating net primary productivity for potential vegetation in north America, GlobalBiogeochern.Cycles, 6, 101-124, 1992. Atjay, G. L., P. Ketner, and P. Duvignaud, Terrestrial primary production and phytomass, in The Global Carbon Cycle, SCOPE 13, edited by B. Bolin, E. Degens, S. Kempe, and P. Ketner, pp. 129-182, John Wiley, New York, 1979. Ball, J. T., I.E. Woodrow, and J. A. Berry, A model predicting stomatal conductance and its contribution to the control of photosynthesis under different environmental conditions, in Progressin PhotosynthesisResearch,vol. 4, edited by J. Biggins, pp. 221-224, Nijhoff, Dordrecht, Netherlands, 1987. Bishop, J., and W. Rossow,Spatial and temporal variability of global surface solar irradiance, J. Geophys. Res., 96, 16,839-16,858, 1991. Bray, R. J. and E. Gorham, Litter production in forests of the world, Adv. Eco. Res., 2, 101-157, 1964. Collatz, G. J., J. T. Ball, C. Grivet, and J. A. Berry, Physiological and environmental regulation of stomatal conductance, photosynthesisand transpiration: A model that includes a larninar boundary layer, Agric. For. Metcorol., 5.•, 107-136, 1991. Collatz, G. J., M. Ribas-Carbo, and J. A. Berry, Coupled photosynthesis-Stomatalconductancemodel for leavesof Ca plants, Aust. J. Plant Physiol., 19, 519-538, 1992. Dajoz, R., Pr6cis d'Ecologie, Gauthier-Villars, Paris, 1982. Duvignaud, P., Productivit• des 6cosystatuesforestiers, in Acres du Colloquede Brv.xelles Organis• par l'Unesco et le Programme BiologiqueInternational, Imprimeries Populaires, Geneva, 1971. Esser, G., The significanceof biospheric carbon pools and fluxes for the atmospheric CO2: A proposedmodel structure, Prog. Biometeorol., 3, 253-294, 1984. Esser, G., Osnabrfick biospheremodel, in Modern Ecology, Basic and Applied Aspects, edited by G. Esser and D. Overdieck, pp. 679-709, Elsevier, New York, 1991. Farquhar, G. D., S. von Caemmerer, and J. A. Berry, A biochemical model of photosynthetic CO2 assimilation in leavesof Cs species,Planta, 1.•9, 78-90, 1980. Grant, R. F., D. B. Peters, E. M. Larson, and M. G. Huck, Simulation of canopy photosynthesis in maize and soy= bean, Agric. For. Meteorol., •8, 75-92, 1989. Harvey, L. D. D., Effect of model structure on the response of terrestrial biosphere models to CO• and temperature increases, Global Biogeochem.Cycles, 3, 137-153, 1989. Houghton, J.T., G.J. Jenkins, and J.J. Ephraums, eds., Climate change,the IPCC ScientificAssessment,Cambridge Unversity Press.,New York, U_nt_•4_ States, 1990, Leemans, R., and W. P. Cramer, The IIASA database for mean monthly values of temperature, precipitation and cloudinesson a global terrestrial grid, technical report, Int. Inst. for Appl. Syst. Anal., Laxenburg, Austria, 1991. McMurtrie, R. E., R. Leuning, W. A. Thompson, and A.M. Wheeler, A model of canopy photosynthesis and water use incorporating a mechanistic formulation of leaf CO2 exchange, For. Ecol. Manage., 5œ,261-278, 1992. Olson, J. S., J. A. Watts, and L. J. Allison, Major world ecosytem complexesranked by carbon in live vegetation: A database, Rep. NDP-017, Oak Ridge Nati. Lab. and U.S. Dep. of Energy, Oak Ridge, Tenn., 1985. Pitman, A. J., Z.-L. Yang, J. G. Cogley, and A. HendersonSellers, Description of bare essentialsof surface transfer for the Bureau of Meteorology ResearchCentre AGCM, Res. Rep. N ø 3œ, Bur. of Meteorol. Res. Cent., Melbourne, Australia, 1991. Raich, J. W., E. B. Rastetter, J. M. Melillo, D. W. Kicklighter, P. A. Steudler, B. J. Peterson, A. L. Grace, B. Moore III, and C. J. VSr'6smarty,Potential net primary productivity in South America: Application of a global model, Ecol. Appl., 1, 399-429, 1991. Schlesinger,W. H., Biogeochemistry:An Analysis of Global Change, Academic, San Diego, Calif., 1991. Sellers, P. J., Canopy reflectance,photosynthesisand transpiration, Int. J. Remote Sens., 6, 1335-1372, 1985. Sellers,p. J., J. A. Berry, G. J. Collatz, C. B. Field, and F. G. Hall, Canopy refiecta.qce,photosynthesisand transpiration. III. A reanalysisusing improved leaf models and a new canopy integration scheme,Remote Sens. Environ., Jœ, 187-216, 1992. Sellers,W. D., A quasi-three-dimensionalclimate model, J. Clim. Appl. Meteorol., œœ,1557-1574, 1983. Tinker, P. B., and P. Ineson, Soil organic matter and biology in relation to climate change,in Soils on a Warmer Earth, edited by H. W. Scharpenseel,M. Schomakerand A. Ayoud, pp. 71-87, Elsevier, New York, 1990. Whittaker, R. H., and G. E. Likens, The biosphere and man, in Primary Productivity of the Biosphere, Ecol. Stud.,vol.14, edited by H. Leith and R. H. Whittaker, pp. 305-328, Springer-Verlag, New York, 1975. Wilson M. F., and A. Henderson-Sellers,A global archive of land cover and soils data for use in general circulation ................,,,,.,,.,=l,:,, J r,u..., ,5,1 WullschlegerS. D., Biochemical Limitations to Carbon Assimilation in Ca Plants -- A retrospective analysis of the A/Ci curvesfrom 109 species,J. Exp. Bot., J4, 907-920, 1993. P. Warnant, L. Francois, D. Strivay, and J.-C. G•rard, Laboratoire de Physique Atmosph•rique et Planetaire, Institut d'Astrophysique, Universit• de Liege, 5 avenue de Cointe, B-4000 Liege, Belgium. (ReceivedDecember20, 1993;revisedMarch 10, 1994; acceptedMarch 29, 1994.)