UBC Faculty Research and Publications

Validation of the Integrated Biosphere Simulator over Canadian deciduous and coniferous boreal forest… Delire, Christine; Price, David T.; El Maayar, Mustapha; Black, T. Andrew; Foley, Jonathan A.; Bessemoulin, Pierre Jul 31, 2001

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-Black_AGU_2001JD900155.pdf [ 1.31MB ]
JSON: 52383-1.0041932.json
JSON-LD: 52383-1.0041932-ld.json
RDF/XML (Pretty): 52383-1.0041932-rdf.xml
RDF/JSON: 52383-1.0041932-rdf.json
Turtle: 52383-1.0041932-turtle.txt
N-Triples: 52383-1.0041932-rdf-ntriples.txt
Original Record: 52383-1.0041932-source.json
Full Text

Full Text

JOURNAL OF GEOPHYSICAL  RESEARCH, VOL. 106, NO. D13, PAGES 14,339-14,355, JULY 16, 2001  Validation of the Integrated Biosphere Simulator over Canadian deciduous  and coniferous  boreal  forest  stands  MustaphaE1 Maayar,• David T. Price,1 ChristineDelire,2 JonathanA. Foley,2 T. Andrew Black,• and Pierre Bessemoulin4  Abstract. Data collectedduringthe Boreal Ecosystem-Atmosphere Study (BOREAS) at four different forest standswere used to test surfaceenergyand carbon fluxessimulated by the IntegratedBiosphereSimulator(IBIS). These standsincludeddeciduousand conifer speciesand were locatedin both the BOREAS northern and southernstudyareas. Two runswere made: one usingthe original IBIS model and the other usinga version modifiedto consideran organicsoil layer (OSL) coveringthe mineral soil surface.Results showthat the inclusionof the OSL substantiallyimprovedthe simulationof soil heat flux, as well as of temperature and moisturein the topmost soil layer. Simulationsshowthat latent and sensibleheat fluxes,and net ecosystemexchangeof carbon,were not affected appreciablyby the presenceof a thin (10 cm or less)OSL coveringthe forest floor. With a thick (50 cm) OSL, however,simulationof latent heat flux and net ecosystemexchange of carbonwas substantiallyimproved. Considerationof the OSL in the model also led to better simulationof the onsetsof soil thawing. Correct estimationof heat diffusionto deep soil throughthick organiclayersrequiresa parameterizationthat accountsfor the state of the organicmaterial decomposition.Simulationspresentedhere also showthe necessityfor usingdetailed information on soil physicalpropertiesfor better evaluationof model performance. SiB2 of Sellerset al. [1996], the revisedISBA of Calvet et al. [1998], and Dynamic Global Vegetation Models (DGVMs) Characteristicsof vegetated terrestrial surfaces influence suchas the Integrated BiosphereSimulator (IBIS) of Foleyet weather, climate, and atmosphericcompositionthrough their al. [1996]). effectson exchangesof radiation, heat, water, momentum, and The formulationsof suchmodelsare necessarilycomplexyet carbon[Shuklaand Mintz, 1982;Pielkeet al., 1997;Hogget al., mustincludenumerousapproximations and assumptions (be2000].For the lastthree decades,considerablemodelingeffort causemany of the processesare themselvescomplexand rehas been investedin understandingthese influences.During main poorly understood).For this reason, an international thistime, land surfaceparameterizationmodelsdevelopedpri- Projectfor Intercomparisonof Land SurfaceParameterization marily for implementationin AtmosphericGeneral Circula- Schemes(PILPS) was initiated [Henderson-Sellers and Brown, tion Models (AGCMs) haveevolvedfrom simpleaerodynamic 1992], in which different land surfaceparameterizationswere bulk transfer formulae and uniform prescriptionsof surface comparedwith observedflux data. Results involving23 land et al. parameters(e.g., albedo, aerodynamicroughness,soil mois- surface schemeswere reported by Henderson-Sellers ture) [ManabeandBryan,1969;Schneider andDickinson,1974; [1996] and Chen et al. [1997]. As pointed out by Delire and Manabe and Wetheraid,1987] to more realisticrepresentations Foley [1999], however, the PILPS study did not include data of terrestrialbiophysicalprocesses(e.g., SiB of Sellerset al. from any forested sites. It is now fairly clear that the circumpolarboreal forest (of [1986] and CLASS of l/erseghyet al. [1993]). More recently, North America, Scandinavia,and northernAsia) playsan improcess-based biophysicalmodelshave alsotaken into account portant (if not crucial) role in regulating atmosphericCO2 photosynthesisand plant water relations to provide a more internally consistentdescriptionof the coupled fluxes of en- contentby acting as a carbon sink [Tans et al., 1990; Gates, ergy,water, and carbon(e.g., the revisedland surfacescheme 1993;Ciais et al., 1995;Sellerset al., 1997]. It may yet prove to be just as important in controllingthe global climate [e.g., Bonan et al., 1992; Foley et al., 1994; Pielke and Vidale, 1995; •Natural ResourcesCanada,CanadianForest Service,Edmonton, Hogg et al., 2000]. Consequently,it is of considerableinterest Alberta, Canada. whether biosphericmodelssuchas IBIS, designedas dynamic 2Centerfor Sustainability andtheGlobalEnvironment, Institutefor vegetation schemesfor use in AGCMs, can reproduce obEnvironmental Studies,University of Wisconsin-Madison,Madison, servedfluxesin thisparticularregion,in additionto the general Wisconsin. 3Department of Agroecology, University of BritishColumbia, Van- importanceof validating them for different forest ecosystems. couver, British Columbia, Canada. In the studyreportedhere, the performanceof IBIS operating 4M•t6o-France, ServiceCentrald'Exploitation de la M6t6orologie, in a stand-alonemode was investigatedfor different forests Toulouse, France. typeslocatedwithin the Boreal EcosystemAtmosphere Study Copyright2001 by the American GeophysicalUnion. (BOREAS) region [Sellerset al., 1997]. These forest sites,located both in the southernand the northernstudyareas(SSA Paper number 2001JD900155. 0148-0227/01/2001JD900155509.00 and NSA, respectively),are knownwithin the BOREAS liter1.  Introduction  14,339  14,340  EL MAAYAR  ET AL.' VALIDATION  OF IBIS  ET = Tu+Eiu+T•+Ei•+Es P  Tu Eiu  CO•  shortwave  longwave  radiation  radiation  latent heat  sensible heat  CO•  Runoff  snowmelt  ..... .•O cm .......................... 25  so  .nfiltration  ................  •00cm ............................................................................................................................................................................................... heat diffusion...................  • soil moisture 200...cm ................ ! '""'• diffusion  ...... todeep soil SOil aecomposlnon  ..... ............................... '""';,; ...........................................  .... 4..o..o....•.m ............................................................................................................................................................................................................................................................................................................................  Figure 1. Schematicillustrationof the water balance,the carbonbalance,and the energybalancecalculations performed by IBIS. The following terms of the terrestrialwater and energybudgetsare explicitly  simulated'evapotranspiration (ET), whichis the sumof transpiration(T, and T•), evaporationfrom the soil surface(Ex), evaporationof water interceptedby vegetationcanopies(Ei, and Ei•); runoff, soil water infiltration, subsurfacedrainage; latent, sensible,and soil heat fluxes; upward shortwaveand longwave radiation;net carbonassimilation; autotrophic(root) and heterotrophic(microbes)respiration.Precipitation (P' rain + snow), downwardshortwaveand longwaveradiation,and atmosphericCO2 concentrationare prescribed.The subscripts u and l refer to the upper and lower canopies,respectively. ature as the northernold black spruce(NOBS), southernold jack pine (SOJP), southernold-aspen(SOA), and southern youngaspen(SYA) stands.Within the North Americanboreal, forest dominated by black spruce (Picea mariana (Mill.) B.S.P.)is mostcommon[Larson,1980],althoughaspen(Populus tremuloidesMichx.) is the dominantdeciduousspecies.In the BOREAS region,jack pine (PinusbanksianaLamb.) is the secondmostcommonconiferousspecies.The BOREAS data therefore provide an important basisfor validatingIBIS when applied to high-latitudeforest ecosystems. Only when such studieshave confirmedthat the model is able to providerealisticresponses to present-dayclimatefor a rangeof ecosystems will it be appropriateto use it to assesseffects of possible climatic changeson forest processesand distribution.  2.  Model Description  Version 2.1 of IBIS [Kuchariket al., 2000]was usedin this study.This model is hierarchicallyorganizedto allow for explicit couplingamongecological,biophysical,and physiological processes operatingat differenttimescales[Foleyet al., 1996]. Heat, momentum, and water exchangesare computed using the LSX land surfaceschememodel of Pollardand Thompson [1995].LSX borrowsits main structurefrom the SiB [Sellers et al., 1986]and BATS [Dickinsonet al., 1986]models,simulating two vegetationlayers(corresponding to upper and lower canopies),and sixsoillayerswith horizonthicknesses of 0.1, 0.15,  0.25, 0.50, 1.0, and 2.0 m, respectively.IBIS simulatescarbon exchangeand stomatal regulation of both Cs and C4 plant speciesaccordingto a semimechanisticapproach based on physiological evidence[Farquharet al., 1980;Ball et al., 1986;  Collatzet al., 1991, 1992].Net ecosystem exchange(NEE) is computedas the differencebetween carbon taken up by the vegetationthroughgross(canopy-level)photosynthesis and carbon lost through plant respiration of foliage, wood, and roots and through microbial decompositionof litter and soil organicmatter. For all the simulationsreported here, vegetation was treated as "static" in the IBIS vegetationdynamics module; i.e., no changein stand structurewas assumedto occur during the periods for which simulationswere performed.Figure 1 providesa schematicdescriptionof the LSX land surfaceschemeas coupledwith the photosynthetic model of Collatzet al. [1991]and the soil respirationmodel of Lloyd and Taylor[1994].  3.  Data and Study Sites  Detailed descriptionsof sites, instrumentation,and measurementsare reportedby Black and Nesic[1999]and Chenet al. [1999]for SOA; by Bessemoulin and Puech[1998]for SYA; byBaldocchiet al. [1997a,1997b]for SOJP;andby Gouldenet al. [1997] for NOBS. Measurementsof turbulent fluxes of latent and sensibleheat, and NEE were performedusingthe eddycorrelationtechnique(referred to from here on as EC).  EL MAAYAR  ET AL.: VALIDATION  OF IBIS  14,341  Table 1. VegetationandSoilParameters Usedfor the Simulations Forest  SOA  SYA  Site  SOJP  NOBS  VegetationParameters  Vegetationtype  Fractional vegetation cover  upper  trees  lower  shrubs  upper  0.8  lower  0.8  trees  trees  0.9  0.8  upper  21.50  2.75  13.50  Leafareaindex(LAI)  upper lower upper upper  dailya dailyb 0.4 30.0  5.5  2.4 2.0 0.2 25.0  Stemareaindex(SAI) Vmax c  lower  0.9  0.8  Vegetation height,m  lower  trees  C3 grass  2.00  0.4 30.0  27.5  10.00  0.20  4.5 0.6 25.O  25.0  Soil Parameters  Soil texture  siltyloam  siltyloam  sand  silty clay  Soilsand/silt/clay content,  20/65/15  20/65/15  92/05/03  10/45/45  %  1.00 0.08 12.1/10.6/10.0/ 9.3  1.00 0.08 12.1/12.1/12.1/ 12.1  Initial soilmoisturecontent,  0.58/0.58/0.47/  0.16/0.23/.23/  0.1/0.1/0.1/  0.0  0.37  0.38  0.15  0.0  0.00  0.00  0.00  0.80  mineralsoil  0.50  0.50  0.44  0.48  organicsoil  0.80  0.80  0.80  Wiltingpoint,m3/m 3  mineralsoil organicsoil mineralsoil  0.33 0.60 0.13  0.33 0.60 0.13  0.09 0.60 0.03  Saturated hydraulic  organicsoil mineralsoil  0.20 0.25  0.20 0.48  0.20 1.46  organicsoil mineralsoil  0.25 0.21  0.48 0.21  1.46 0.07  0.46  organicsoil mineralsoil  0.21 4.70 4.70 3.60 1.11  0.21 4.70 4.70 3.60 1.11  0.07 1.70 1.70 1.42 0.67  0.34 7.90  m3/m 3 (bylayer) Initial soilice content,m3/  1.00 0.05 13.4/12.1/11.6/ 11.6  4.00  Soildepth,m Organic layerthickness, m Initialsoiltemperature, øC (bylayer)  0.5  -5.9/-4.3/-3.5/ -2.0/-0.49/-0.5  m3  Soilporosity, m3/m 3 Fieldcapacity, m3/m 3  conductivity,m/d  Saturated air entry potential,m  Soilb parameter  organicsoil  Totalsoilcarbon,kg-C/m 2 Totallittercarbon,kg-C/m 2  0.80  0.39 0.60 0.25 0.20 0.46  0.34  7.90 4.18 0.48  •'LAI..... (upper canopy)is 2.  bLAImax (lowercanopy)is 2.3. CVma x isthemaximum Rubisco capacity at 15øCof thetopleaf.Theirvalues weretakenfromKucharik etal. [2000].  emic, luvisolic,and organicsoil orders.Data reported here hourlydata were used;however,data collectedon the half were obtainedin the period July 17 to August 16, 1994. No hourswere deletedexceptfor precipitationwhere half hourly NEE measurements were available for this site. datawereaddedto obtainhourlydata(mmh- •). Testsshowed The SOJP site was located in the SSA region at 53.55øN, Dominatedbyjackpine,with a meancanopyheight thatno pertinentinformation waslostusinghourlyin placeof 104.41øW.  All observeddata were recordedat half-hourlyintervals.Only  of 13.5 m, the standextendedfor over one kilometrefrom the half-hourly observations. The SOA sitewasan extensive70-yearold stand,dominated instrument tower in all directions.Understory shrubswere  byaspen,locatedin theBOREASSouthernStudyArea (SSA) at--•53.7øN,106.2øW.Averagetree heightwas --•21.5m. The understory vegetation wascomposed mainlyof hazelnut(Corylus comutaMarsh) --•2 m tall. The soil is predominantlyan  sparse,consisting of isolatedpatchesof alder(Alnuscrispa). Ground cover was extensivehowever, including bearberry  (Arctostaphylos uva-ursi), bogcranberry (Vaccinium vitisideae), and lichens(Cladinaspp.).Soil was a coarsetexturedsand,  classified asa degradedEutricBrunisol/Orthic EutricBrunisol. Data used in this study were obtained for the periodJuly 1 to [Blacket al., 1996].Data usedin this studywere collected duringthe 1996fieldcampaign andcoverthe periodJune1 to September16, 1994. Orthic Luvisolwith an 8-10 cm deep surfaceorganiclayer  31. The NOBS sitewas locatedin the BOREAS Northern Study by a black The SYA sitewas alsolocatedin the SSA, with approximate Area (NSA) at 55.88øN,98.48øW,surrounded forestof varyingstatureextendingfor sevcoordinatesof 53.39øN,105.20øW.The standwasdominatedby spruce-dominated  December  aspen, withveryhighstemdensity (•10 stems m-2) anda eral kilometersin all directions[Gouldenet al., 1997]. Mean meantreeheightof 2.75m. Soilsrangedfrom graywoodedto tree heightwas --•10m. Soilsin this area consistof organics, degraded black,classified in the brunisolic, gleysolic, chernoz- clays,and somesandydeposits.Data usedin this studywere  14,342  EL MAAYAR  800  ET AL.' VALIDATION  For the NOBS site, initial soil water content was assumed  RB  equal to porosityand all in the form of ice. Initial soil moisture content  E 600  •  values at the SYA  and SOJP sites were taken  from  Cuencaet al. [1997] andNijssenet al. [1997],respectively. Soil texturesfor SOA and SOJP siteswere classifiedas silty loam and sandy,respectively,basedon data givenby Cuencaet al. [1997](Table 1). For the NOBS site,soiltexturedatawere taken from the globaldata set usedin large-scalesimulations with IBIS [Foleyet al., 1996].Soil at the SYA sitewasassumed  ,.• 400 ,-  OF IBIS  200  0  to have the same texture as that found at the SOA site. For all  0  800  200  400  600  SSA sites,soil hydraulicpropertieswere taken from Nijssenet al. [1997]and Cuencaet al. [1997],with soil depthset to 1 m, followingdatafrom Claytonet al. [1977].Fractionalvegetation coverandleaf areaindex(LAI) for SOJPandNOBS, andstem areaindexvalues,weretakenfromKimballetal. [1997],Nijssen et al. [1997], and Goweret al. [1997],respectively.Leaf area index at SYA was assumedequal to that reported for SOA by Black et al. [1996] for the period of the simulation,as also reportedby R. D. Pyleset al. (personalcommunication,1999). Daily changesin LAI at SOA, asreportedby Chenet al. [1999], were approximatedusing a third-order polynomialfunction. Root profileswere as givenby Nijssenet al. [1997] for SOJP and NOBS and as givenby Jacksonet al. [1996]for SOA and  800  Sn  600 400  200  b  0  0  200  400  600  SYA. Total soil carbon and total litter carbon values, were  800  taken from Goweret al. [1997] and Steeleet al. [1997].  500  4.  Lup  450  Comparison of Observed and Simulated Data  Two seriesof simulationswere performed for each forest stand.The first serieswas carried out usingonly mineral soil 400 texture classes,as used in IBIS2.1 simulationsat the global 350 scale by Kucharik et al. [2000]. For much of the Canadian 300 borealforestzone,however,particularlythosevegetationcommunities dominated by aspen and spruce, an organic layer 250 coversthe surface[Claytonet al., 1977].Typically,thislayer is 200 deeper in the northern regionsbecausedecompositionrates are normally lower than in the warmer conditionsfarther 200 250 300 350 400 450 500 south.Hence, for our secondseriesof simulations,an organic soil layer was introducedinto the model, with hicknessfixed Observed (W m-2) accordingto observations made by Black et al. [1996]for the SOA site, by Baldocchi et al. [1997a]for SOJP,and by Clayton Figure 2. Observedand simulated:(a) net radiation (R,), et al. [1977] for NOBS (see Table 1). For the SYA sitewe used (b) net shortwaveradiation (Sn), and (c) upward longwave radiation(Lup)at old aspensiteduringtheperiodJune1 to the same thicknessas that reported for SOA. At the SOA, December 31 1996. SYA, and SOJP sites,only the top layer of the sixmodel layers was treated as organic,but at NOBS the top four layerswere all treated as organic,replacingthe correspondingmineral soil layers. Moreover,whenpresenceof organicsoilwassimulated, obtainedfor the periodApril 1 to October9, 1996.No soilheat surfacealbedowas set equal to 0.1 (i.e., it was treated as a flux measurements were available at this site. For eachhourlytime step,IBIS simulatedthe energyand C moist dark soil as defined in Table 6.4 of Brutsaert[1982]). balancesusingthe followingforcingvariables:air temperature, Mineral soil albedoswere computedby the model asa function hourly total precipitation, wind speed, humidity, incoming of sand,silt,andclayfractionsfollowingDickinsonet al. [1986]. Heat capacity of dryorganicmaterial(0.58MJ m-3 K-•) short-waveand long-waveradiation, and surfaceair pressure. (0.06W m-• K-•) weretakenfrom Incoming long-waveradiation, whenever missing,was esti- andthermalconductivity matedusingthe formulaof Satterlund[1979].Occasionaldata Guyot [1997]. Becauseof the lack of informationon the parof the orgapswere filled usingmesonetstationdata [Shewchuk,1997]. tide sizedistribution(or degreeof decomposition) In addition, site-specificregressionequations, derived from ganicmatter layersat the studysites,saturatedhydraulicconmeasureddata, that expressednet radiation as a function of ductivityand the soil b parameter (i.e., the exponentfor the incomingshortwaveradiation,were usedto fill very occasional Campbellmoisturereleaseequation[Campbelland Norman, 1998])were assumedto be the samein both setsof simulations: gapsin shortwaveradiation time series. The data used to initialize the model are shown in Table 1. i.e., with andwithoutthe simulatedsoilorganiclayer.Data for Soil temperaturedata were availablefor all sites.For the SOA saturatedhydraulicconductivitiesat each site were obtained site, measured soil moisture was used to initialize the model. from Cuenca[1997]. i  i  i  i  i  i  i  EL MAAYAR  ET AL.: VALIDATION  OF IBIS  14,343  500 400  •H  IBIS  •  IBIS+  300  _  200  ß.'.'-  1oo  v'  . .•.•.•'•'.,  o  " '  _L  ;..--..- ,-.-..  '  __  '• • '  :  :,,.:  -100 -200 I  -200  I  I  -100  0  100  200  300  400  500  -200  -100  0  100  200  I  300  I  400  I  500  I  0  100  200  300  400  500  600  -100  0  100  200  300  400  500  600  6OO 500  400 300  200  1O0  o -lOO  -100  200 G  leafperiod only  100 •-  '.  total period ofsimulation  e  (leaf+leafless periods) f ß  '...V4:g•t'-.  .-.,....,-  ,.:.-..,?". :.-  .,....,=:..  -100 [ "' I  -20  I  -lo  '. . .. '  T  0  I  I  I  0  •0  20  ß  -20  1  ß  IBIS+osl ßIBIS I  I  30  40  I  I  -20  I  -•0  I  I  I  I  I  0  •0  20  30  40  I  Observed (Wm-2)  -NEEIBIS  -10  ß  :..  Observed (Wm-2)  10  .  ..., :: "•,.'..',.'.,......'..'.. '. . .. '  -  1BIS+osl ,.•_-  '..":" .-_:_':'•"-7 • ß':':?::' •"  -30  -40 -40  -30  -20  - 10  0  10  Observed (gmol-CO 2m-2s-•)  20  -40  -30  -20  - 10  0  10  20  Observed (gmol-CO 2m-2s-•)  Figure 3. Simulatedhourly above-canopy fluxesplotted againstdata observedat the old aspensite during the period June 1 to December31, 1996. (a) Sensibleheat flux assumingpresenceof mineral soil only, (b) mineralsoilwith surfaceorganiclayer.(c) Latent heatflux assuming presenceof mineralsoilonly,(d) mineral soilwith surfaceorganiclayer. (e) Soil heat flux duringthe periodwhen trees are in leaf, (f) when trees are leafless.Dark dotsindicatethat the run assumedthe presenceof mineralsoilonly and light trianglesindicates that the run assumedthe presenceof mineral soil with surfaceorganiclayer. (g) Net ecosystemexchange (NEE) assumingpresenceof mineral soil only, and (h) mineral soilwith surfaceorganiclayer. Carbonflux (NEE) flux is positivetowardthe atmosphere.  14,344  EL MAAYAR ET AL.: VALIDATION OF IBIS  400 t 300  ß  observed  a  IBIS  200 lOO  o - lOO  500 400  300 200  100  50  -50  ' al 0  . *,• e  e  e  [..T..• -30 t0,,• ee •Z;  e  ,  •''  186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 Day of year  Figure 4. Observedand simulateddiurnalvariationof: (a) sensibleheat flux, (b) latentheat flux, (c) soil  heatflux,and(d) net ecosystem exchange at old aspensite.Dotsindicateassumed presence of mineralsoil only;graytrianglesindicateassumed presence of mineralsoilwithsurfaceorganiclayer.Carbonflux(NEE) is positivetoward the atmosphere.  5.  Results  5.1.  SOA Site  Simulated and observed radiation and flux data are shown in  Figures2, 3, and 4. The agreementbetweenobservedand  simulatednet shortwaveradiation (Sn) is high (Figure 2b), suggesting that the parameterizationof surfacealbedowas fairly accurate.Agreementbetweenobservedand simulated fluxesof net radiation(Rn), and upwardlongwaveradiation  (Lup)is alsoreasonable, although R, is slightly underesti-  EL MAAYAR  300  ET AL.: VALIDATION  OF IBIS  14,345  ........  _  a "  !  "  200  :-  •  •  ....  '  ß  ':  !  j  "  ':  i!  observed  IBIS(SAI=0.2) ..................................................... IBIS (SAI=0.4)  IBIS(SAI=0.8)  100 o -100 ..  500  ß  b  :.  4OO  300  200 lOO o -100  --  c  1oo  ?  50  0  ...  -50 I  I  I  I  I  I  I  I  I  I  I  I  I  I  I  186  187  188  189  190  191  192  193  194  195  196  197  198  199  200  Day of year Figure 5. Sensitivityof simulatedenergyfluxesto different input valuesof stem area index (SAI), at old aspensite: (a) sensibleheat flux, (b) latent heat flux, and (c) soil heat flux.  matedowingto overestimation of L up(Figure2c).The latter selectedvalue of SAI was relativelyminor (Figure 5c). A furresultsfrom overestimationof surfacesoil temperature(Fig- ther developmentof this point is pursuedin section5.3 where ures 5a and 5b). measured canopy heat storage data were available for the Agreement between observedand simulatedlatent heat flux SOJP site. (LE) (Figures3c, 3d, and 4b) is higher than for sensibleheat The modelgenerallyoverestimates observedG (Figures3e flux (H) (Figures3a, 3b, and 4a) and soil heat flux (G) (Fig- and 4c), a problem common to many land surface models, ures 3e, 3f, and 4c). This bias obtainedin the simulationof H particularlywhen they are appliedto forestsites[e.g.,Shaoand causesa temporary imbalance,mainly around midday, in the HendersonSellers,1996; Chen el al., 1997; Delire and Foley, total heat flux.Two other simulationstudiesreporteda similar, 1999;Bellisarioel al., 2000]. This evidentlyresultsfrom lack of but largerbiasin the LSX model [Pollardand Thompson,1995; considerationof the role of the organiclayer usuallycovering Delire and Foley, 1999].Analyzingthe performanceof LSX in forest floors.Indeed, after includingan organiclayer, referred tropical forests,Pollard and Thompson[1995] suggestedthat to here as the IBIS + osl run (Figures3e, 3f, and 4c), mean the heat imbalancewas causedby exaggeratedheat storagein simulatedG was much closerto the observedvalues,although trunks and stems(estimatedfrom stem area index, SAT) and it remainedhigherfor daytimeandlowerfor nighttime(Figure releasedmainly by evaporationduring the night. To test this 4c). Poor numerical resolution of soil temperature near the hypothesis,we ran the model for different valuesof stem area surface is a common problem with land surface models, as (Figure5). Doublingthe SATincreasedthe middaybiasin both explainedby Bellsel al. [1993] (workingon surfaceparametersimulatedsensibleand latent heat fluxes,while halvingthe SAT izationsfor the European Centre for Medium-RangeWeather led to a relative improvementin the simulation of LE as well Forecasts)and by Chenel al. [1997] (who carriedout a PTLPS asH (Figures5a and 5b). Sensitivityof soilheat flux (G) to the study involving23 land surfaceschemesusing data from the  14,346  EL MAAYAR  ET AL.: VALIDATION  OF IBIS  Table 2. Averagesand StandardDeviationsof Observedand SimulatedFluxesa Standard  Average Observed  IBIS  IBIS + osl  Deviation  Observed  IBIS  IBIS + osl  nb of Observations  SOA Forest-SSA  Rn, W m-2  84.31  44.45  47.39  187.74  175.94  175.66  5039  H, W m-2 LE, W m-2  21.89 48.11  8.43 53.28  8.28 55.00  77.21 82.22  36.84 93.89  40.14 102.93  4632 4268  Bowen ratio, H/LE  0.46  0.16  0.15  0.94  0.39  0.39  4268  G, W m-2  0.82  -11.10  -9.53  6.51  40.00  25.41  4897  -4.01 2.52 -1.66  7.92 1.44 6.72  10.22 0.60 8.72  10.20 0.58 8.75  2591 1672 4263  223.24 49.46 116.49  236.56 39.49 107.71  235.25 41.41 121.36  696 676 666  NEE-l,/•mol-CO2 m-2 s-• NEE-2,/•mol-CO2 m-2 s-• NEE-3,/•mol-CO2m-2 s-•  -2.95 1.38 -1.39  -3.70 2.56 -1.45  Rn, W m-2 H, W m-2 LE, W m-2  153.57 13.81 94.48  136.54 29.12 100.92  138.35 29.21 104.05  Bowen ratio, H/LE  0.15  0.29  0.28  0.42  0.37  0.34  666  G, W m-2  2.44  6.41  5.07  15.07  40.31  19.91  696  Rn, W m-2  121.73  88.03  87.84  194.63  204.71  200.33  1755  H, W m-2 LE, W m-2  61.32 47.84  52.08 32.75  56.85 29.44  110.59 64.61  85.98 66.20  92.16 73.72  1729 1722  SYA Forest-SSA  SOJP Forest-SSA  Bowen ratio, H/LE  1.28  1.59  1.93  1.71  1.30  1.25  1722  G, W m-2  -1.40  2.14  1.51  18.20  33.62  14.79  1765  NEE,/xmol-CO2 m-2 s-•  -0.43  -0.40  4.59  1.68  1.93  1716  Rn, W m-2 H, W m-2 LE, W m-2  124.03 63.13 31.42  87.09 50.54 26.23  87.23 50.33 30.03  202.93 126.05 47.38  216.25 84.63 34.32  215.58 85.79 44.91  4607 4344 4322  2.66  -0.78 NOBS  Bowen ratio, H/LE  2.01  G, W m-2  NEE,/•mol-CO2 m-2 s-•  -0.45  Forest-NSA  1.93  1.68  7.60  4.09  0.21  -0.57  3.03  2.47  1.91  31.40  14.90  1.16  1.44  4322  4350  aIBIS + osl indicatesthat the first soil layer in IBIS wasconsideredas organicfor YA, OA, and OJP sites.Idem for OBS sitebut the four first soillayerswasset as organicin this case.NEE-1 is the NEE for the periodwhentreesare in leaf; NEE-2 is the NEE for the periodwhen trees are leafless;and NEE-3 is the total NEE (NEE1 + NEE2). Depth of the soil organiclayer usedin the IBIS + osl simulationis: 8 cm for SOA and SYA, 5 cm for SOJP, and 50 cm for NOBS.  Cabauwgrasslandsite). This causesan amplificationof soil heat fluxesand a delay of turbulent fluxes. Simulatedand observedNEE data are shownin Figures3g, 3h, and 4d. We adopted the conventionthat NEE is positive when the net flux is toward the atmosphere.Agreement between the simulation  and measurements  is lower around noon  then at other times of day.Delire and Foley [1999] reporteda similarproblemwhen comparingresultsof their simulationfor a grasslandsitewith data from the FIFE experiment[Vermaet al., 1992].Figure 4d alsoshowsthat the model generallyoverestimatesnighttimerespiration.Recent analysesby BaMocchi et al. [1998] and Lee [1998] on the reliabilityof EC measurements of carbon exchangebetween the surfaceand the atmospherein forestsmay help to explaintheseresults.By analyzing the conservation equationof CO2 [BaMocchietal., 1998,Equation (1)]. Theseauthorssuggestthat the conventionalanalysis of EC data may neglectan important advectiveeffect in forests under stable conditions,occurring particularly at night. Although there is some controversysurroundingLee's explanation, it is evident that the flux of CO2 from the air column below the instrumentmay be underestimatedand hence measured NEE is overestimated.Hence the model'sslightunderestimatesof observednighttimeNEE seenin Figure 4d maybe reasonable. Conversely, the model evidently overestimates daytimeNEE, which implieseither that ecosystem respiration is underestimatedor that net photosynthesis is also exagger-  ated. The simulationof net photosynthesis is derivedfrom the Farquhar-Collatz model, however, which assumessimultaneousstomatalregulationof CO2 exchangeand transpiration. The strong agreement between measured and modeled LE indicatesthat the photosyntheticflux is approximatelycorrect, whichthereforesuggests more investigationof respirationprocesses within  IBIS  is warranted.  Another source of the nighttime respiration discrepancy could originatefrom the soil carbondensitydata reported by Goweret al. [1997],that were usedto initializethe model, and whichaffectsoilrespiration.Thesedata (seeTable 5 of Gower et al. [1997]), are considerablylower than thosereportedby other investigators (e.g.,T. Nerbusand D. Anderson,Universityof Saskatchewan, unpublisheddata, 1996;H. Veldius,Universityof Manitoba,unpublisheddata,1996).A highervalueof initial soilcarboncontentwould resultin greatersimulatedsoil respiratoryfluxes during the growing season,assumingall other factors are unchanged.A sensitivitytest showed,however, that even if the higher value of initial soil carbon was used,it could not provide a completeexplanationfor the discrepancyobtained between measuredand simulatedmidday NEE.  Over the entire simulationperiod (NEE-3 in Table 2), the model explained---96% of NEE measuredat SOA (original IBIS run). This overallagreementhidesfairly seriousdiscrepancies, however, in both summer and winter data sets.For the  EL MAAY•  ET AL.: VALIDATION  i  30  i  OF IBIS  14,347  i  i _  15  _  a  -15  IBIS  '•..............  IBIS+osl  30  8-25 cm -  15 0 -  _  _  _  15 _  k__r•••_ I  I  250  300  I  15 10 5 0I  200  150  350  Day of year 24  -  b  soil 0-8cm-  16 12  8  l0 0  211  2O?  215  219  223  227  Day of year  0.8  C  '  0-8'cm  0.6  t  0.4 0.2 I  -•...':...............: .... ...... .........::•  0.4  2•-•0cm t  0.2 I 0.5  I 50-100  -  cm  -  0.4 _ 10.. o  0.3  0.2  1,[.....  153  !  ,  .....  200  i.  . _,.,.. i  ,:  5 • 0  •  247  Day of year Figure6. Observed andsimulated: (a) dailysoiltemperature, (b) hourlysoilandstemtemperatures, and(c) dailysoil moisturecontent,at old aspensite.Note that the horizontalscalesdiffer significantly between Figures 6a, 6b, and 6c.  period when trees are in leaf and NEE indicatesnet carbon Table 2). SimulatedwintertimeCO2 effiuxis generallymuch uptake(NEE-1 in Table 2), the ratio of observedto modeled higherthan it shouldbe, suggesting that the temperaturereNEE is -0.81, comparedto -0.54 for the period when trees sponseof the soil respirationmodel is too shallow. are leaflessand NEE indicatesnet carbonrelease(NEE-2 in While includingthe OSL substantially improvedthe simu-  14,348  EL MAAYAR  ET AL.: VALIDATION  lation of G, turbulent fluxes(H and LE) were virtually unchanged(Figures3a-3d, 4a, 4b, and Table 2). AverageNEE increasedby -10% in responseto the simulated OSL, althoughthisresponsewasrestrictedalmostentirelyto the growing season(Table 2). The explanationfor this appearsto be that the organiclayer increasedwater infiltrationand reduced runoff, therebyincreasingsoil surfacewater availability(Figure 6c) andreducingvegetationdroughtstress.Thisin turn led  OF IBIS  midday H by the model. Figure 9 showsclearly that when energy received by the surfacewas close to maximum, the  simulated canopy heatstorage couldexceed100W m-2. Con-  versely,duringthe nightthe modelsimulatedgreaterreleaseof storedenergythan observed(Figure 9), causingH to be overestimated(Figures8c and 8d). Introducingthe OSL produced no significantimprovementin simulatedH and LE fluxes, althoughG wassubstantiallyimprovedand to a greaterextent to a reduction in the seasonal constraints on vegetation- than at the deciduoussites(Figures8g and 8h). atmosphereexchangesof water and carbon. The largest discrepancybetween measuredand simulated The model was able to reproducestemtemperaturesuccess- NEE was also obtained around midday (Figures 8i and 8j; fully (Figure6b), whereassoiltemperaturewasgenerallyover- diurnal cyclenot shown).Nighttime respirationwas correctly estimated,but particularlyin the topmostsoil layer. This was simulated,however,which supportsthe hypothesisthat the due partly to the inconsistencies in the numericalsolutionof simulationof temperatureeffectson canopyrespiration,durskin temperaturementionedpreviously.Adding the OSL to ing daylightperiodsin both coniferousand deciduousecosysthe originalmodel resultedin greater soil moisturecontentin tems,requiresfurther modellingeffort. the top soillayer,however,whichled to increasedheatcapacity Soil temperaturewasgenerallyexaggeratedby the model at and decreasedsoil surfacesoil temperature(Figures6a and this site (Figure 10). Furthermore,the discrepancy between 6b). It shouldbe borne in mind, however,that the results observed and simulated soil temperatures increased with obtainedfrom the IBIS + osl run are dependenton the value depth,suggesting that soil thermal diffusivitywassignificantly estimatedfor the saturated hydraulic conductivityof the or- lower in reality than implied by the valuesusedin the model. ganiclayer (seesection4). Moreover,the observations did not This conclusionwas confirmedby simulatingthe OSL, which showany significantsoil moistureincreasein responseto the improvedagreementfor the upper soil layers(depthsto 25 high rainfall (-60 mm) recordedon Juliandays188 and 189 cm),but stillresultedin appreciableoverestimation of average (Figure 6c), whichsuggests a problemin data acquisitiondur- temperatures. ing this period. Temperatureof the 50-100 cm soil layerwas alsooveresti- 5.4. NOBS Site mated by the simulationbetweendays250 and 350. This sugIn the northern boreal region, soil organic layers are typigestspoor representationof heat diffusionto deep soil after cally deeperthan 50 cm [Claytonet al., 1977]. Hence, for the the latter reachedits peak temperaturein the summer. NOBS site located in the BOREAS-NSA, thicknessof the OSL 5.2.  SYA Site  Reasonable agreement between simulated and observed fluxesof R• and LE was obtained, althoughR• was slightly underestimatedduringnighttime (Figures7a, 7b, 7e, and 7f). In general,H wasoverestimated, but particularlyat night(Figures7c and 7d; seealsoTable 2). This nighttimeexaggeration couldbe due to simulatedreleaseof heat storedin the canopy during the day. Daily stored energyis also likely to be partly releasedin the form of evaporationof water at night (Figure 7f). These nighttime releaseswould then cause,in turn, a systematicunderestimationof nighttimeR• for the SYA site (Figure7b) aswell for the other sites(Table 2). Figure7d also showsthat the diurnal courseof simulatedH generallylagged the observationaldata. As reported above,this originatesfrom the delayedtransfer of soil temperatureinformationinto the atmosphericboundarylayer calculations.Nevertheless,aswith the SOA site, simulatingan organic surfacelayer causeda significantimprovementin modeledsoil heat flux, both tightening the distribution of points and shifting the regression slopecloserto the 1:1 line (Figures7g and 7h).  was set to 50 cm for the IBIS + osl run, correspondingto the top three soillayersin IBIS. Porosity,field capacity,andwilting point data (in termsof volumetricwater content)were taken fromNijssenet al. [1997](Table 1). The standardIBIS run, i.e., assumingonly mineral soil at the surfacewascarriedout using a silty claysoil texture,correspondingto that usedin IBIS for globalsimulations(seesection3). As with the other sites,observedR• wassimulatedbetter for daytimethan nighttime(Figures11a and 11b). Includingthe organic layer resulted in almost no changesin simulatedH  (Figures1lc and 1ld), but producedsignificantly better prediction of LE (Figures11e and 11f). The major contributing factor for the improvementin LE was the reduction of G (Table 2; note that measurements of G were not availablefor this site), which resultedfrom the greater moisturestorage capacityin the 50 cm OSL, comparedto the lower porosityof the mineral layers. Comparisonof observedand simulated mean Bowen ratios for this site and for OJP shows that both  are greaterthan 1.0 (Table 2), whereasthosefor the deciduous siteswere typicallymuch lower. Accordingto the Collatz [1991, 1992] model used in IBIS, changesin stomatalconductanceto water vapourlead directly 5.3. SOJP Site to changesin COx uptake.The improvementsin LE resulting R• was simulated correctly during the day but underesti- from includingthe OSL in the NOBS simulationstherefore mated at night (Figures7a and 7b; diurnal cycleis not shown alsoappearto contributeto improvedsimulationof NEE, with (see section5.2. for illustration)).As for the deciduousforest both a highermean and a wider rangebetweenminimum and sites(SOA and SYA), LE wasgenerallybetter simulatedthan maximumvalues(Figures11gand 11h). As with the SOA and H (Figures8c-8f). The disagreementbetweenobservedand SOJP sites,however, NEE values around midday were oversimulatedH aroundmiddaywas smaller,however,than at the estimated.When integratedover the whole period, the total deciduous forest sites. The existence of measured data on simulatedNEE indicatesa slight carbon sourcewhen no orcanopy heat storagefor this site allows verification of the ganiclayeris included(Table 2), comparedto an observednet explanationgivenin section5.1 concerningunderestimationof sinkfor the measurementperiod. In the IBIS + oslsimulation,  EL MAAYAR  ET AL.' VALIDATION  OF IBIS  14,349  800  b  RB  800  600  600  •  400  400  ,-•  200  200  •  ,•,•  r•  o  0  0  200  400  _H  800  204  i  oo_ .  -100 -100  600  205  206  207  208  209  210  d 200  oø {o  I  I  0  100  200  5oo  211  I  I  I  I  I  i  I  i  204  205  206  207  208  209  210  211  •  - LE .  ß  • .•.'-'wM.y•;;': .  1O0 •Y•J.' •'••.• • .  -100 •l -100  0  5oo  •  300 -  200-  ß  200  ..• •:1' f - 400 ..  •  10  i i i i ii  100  300  400  100  SO0  204  20S  I i i i i i--100  206  207  208  209  210  211  120  120  G ?  80  80  B: 40 '•  0  •  -40 -80  40  -40  -I  I  I  i  I  I  -40  -20  0  20  40  60  Observed (Wm-2)  -  I  I  I  i  I  I  i  204  205  206  207  208  209  210  I - -80 211  Dayofyear  Figure 7. Simulatedhourlyabove-canopy fluxesplottedagainstdataobservedat the youngaspensiteduring theperiodJuly17 to August16, 1994.(a, b) Net radiation.(c, d) Sensibleheatflux.(e, f) Latentheatflux.(g, h) Soil heat flux. Dark dotsindicateassumedpresenceof mineralsoil only;light trianglesindicateassumed presenceof mineral soil with surfaceorganiclayer.  however,the inclusionof an organiclayer resultedin an average net sink, comparablein magnitudeto the observedvalue. Agreementbetweenobservedand simulatedsoil temperatures for the top layer alsoimprovedwhen the OSL was con-  sidered(Figure 12), but with increasingdepth,this agreement became progressivelypoorer. Inclusion of the OSL also improvedthe timing of the onsetof soil thawingat the topmost layer.  14,350  EL MAAYAR  800  - Rn  ET AL.: VALIDATION  IBIS  600  OF IBIS  IBIS+osl  -  400  2oo  o 0  .•, ,7,  600  200  - H  400  600  IBIS . .  800  .  0  200  -  400  600  IBIS+osl. ß  800  .  -  400  •  200  m r.•  ?  ,•,  o  C  -200  600  I  I  - 100  0  100  -  I  ß  •  400  ,-•  200  I  300  ' ß  I  400  500  IBIS  I  I  I  - 100  0  100  I  -  d  I  200  300  .  --  I  I  400  500  IBIS+osl  -  ß  _  ß  ...,t  øß  .' ß ß  :-...: .•'•' ,.  '.  .  o  •....-.,,;.,. •.• -200  150  -. _  _  I  I  I  I  I  - 100  0  100  200  300  - G  I  I  400  1BIS  lOO  -100  I  I  I  I  0  100  200  300  I  400  -  1BIS+osl  -  -  .._.•..•  -  ß ' 4,•,:_.'..i,.•.:•.'_. coß  50  •  I  200  LE  -•  .,,•e..,-,.:..,j:. -. •  _  I  ß •,"•',••=.,•,-•..•  o ß  ß  -  '  '  -40  -20  .-. ¾.:. . g  ß  m  ' "0 4'0 2 Observed (Wm-2) 0  6'0' -40  '  -20  h  ' 40' Observed (Wm-2)  _  '  0  60  15  lO 5 o  -5  •NEE IBi•• • _ ß ....a:.-.•.:>',•:•.7 :. --  _  -2o -15  -20  '  ".: ,'•:'.',..1tlI11•,. '-.".."'•..:.". '.': ß  IBIS+•osl • .•  . •: .,,,•.<•,:•...v •.'. -  +  ß..4:'.?,'7'_."•'.?'•.::'-:"' ñ : ".::'"'"•' •  :'  -15  ' "•" ' '  .....  -10  -5  0  5  i -  10  Observed (gmol-CO 2m-2s-1)  "  i,: ' 15  -20  -15  ',,'"••";" •.: ' -  '..... "::"•"': '  -10  -5  0  5  :-I 10  j15  Observed (gmol-CO2 m-•s-1)  Figure8. Simulated hourlyabove-canopy fluxesplottedagainst dataobserved at theoldjackpinesiteduring theperiodJuly1 to September 16,1994.(a) Net radiationassuming presence of mineralsoilonly,(b) mineral soilwithsurface organiclayer.(c) Sensible heatfluxassuming presence of mineralsoilonly,(d) mineralsoil withsurfaceorganiclayer.(e) Latentheatfluxassuming presence of mineralsoilonly,(f) mineralsoilwith surface organiclayer.(g) Soilheatfluxassuming presence of mineralsoilonly,(h) mineralsoilwithsurface organic layer.(i) Net ecosystem exchange (NEE) assuming presence of mineralsoilonly,(j) mineralsoilwith surfaceorganiclayer.Carbonflux (NEE) is positivetowardthe atmosphere. 6.  Discussion  and Conclusions  hasincreasedsignificantly over the last 30 years.As this understanding hasdeveloped, so too hasthe desireto encapsuCollectiveunderstanding of the physical,physiological, and late it in simulationmodelsof increasing complexity. Currentecologicalprocesses governingterrestrialbiosphericbehavior generation biospheric models such as IBIS are highly  EL MAAYAR ET AL.' VALIDATION OF IBIS  250  14,351  .....  observed IBIS  200 150  100  50  0  -50  -100 1  217  218  219  220  221  222  223  224  225  226  227  Day of year Figure9. Illustration of diurnalvariationof observed (dottedline)andsimulated (originalIBIS;solidline) canopyheat storageat old jack pine site duringsummer.  sophisticated, but this sophistication also makesthem very validatethem againstobserved data at the local (site-level) difficult to test. In order to have confidencein suchmodels, scaleaswell as at the continentaland globalscales.Hence, as e.g.,when usingthem to investigateeffectsof future environ- a part of a projectto exploreclimatechangeeffectson Canamental changeson the biosphere,it is essentialto test and da's forests,this studycomparedIBIS version 2.1 with data  25  ........  observed  0-5 cm  IBIS  ........................... IBIS+osl ,, /• t•  20 ',  15  \  '  ',  •  I'. ',  ,  ,  •  '•  \  10 ..  ..  _ _  i  i  i  i  i 5-15 cm -  I  I  I  I  I  20  15-25 cm -  15  10  ß ' .  237  . -. . ß. .."' ..... . -'-...ß .''. '".  241  245  249  .ß .....  253  Day of year Figure 10. Observed (dottedline) andsimulated diurnalvariationof soiltemperature at oldjackpinesite. Boldsolidlineindicates assumed presence of mineralsoilonly;thinsolidlineindicates assumed presence of mineral soilwith surfaceorganiclayer.  14,352  EL MAAYAR  800  Rn  ET AL.: VALIDATION  OF IBIS  IBIS  IBIS+osl  600  400 200 o  a  ß  b  -200  -200  6OO  ?  ,•,  •  500  0  200  •H  400  600  800-200  0  200  IBIS  400  600  IBIS+osl  800  ,  400  •:  300  x:•  200  '•  100  E  o  r.•  -tOO  d  -200  _  -200-tOO  400  -  0  - LE  tOO  200  IB  o  300  400  500  600-200-tOO  0  -  ß  tOO  200  300  400  500  600  IBIS+o. sl  300  200  - . •?•: :-.••.i •-•  100  o  '  -tOO  - t 00  ' ' 0  t 00  e -.ß.:: ,.•5" ':•:'  200  300  400  - t 00  0  Observed (W m-2) t0  NEE  _  t 00  200  300  400  Observed (W m-2)  IBIS  IBIS+osl  5 0  -5  -15 g  -20 -2o  -•5  -lo  -5  o  s  lo  15  Observed (gmol-CO 2m-2s-1)  -2o  -15  -lo  -5  o  5  •o  •5  Observed (gmol-CO 2m-2s-1)  Figure 11. Simulated hourly above-canopyfluxesplotted againstdata observedat the old jack pine site duringthe periodApril 1 to October9, 1996.(a) Net radiationassuming presenceof mineralsoil only, (b) mineral soil with surfaceorganiclayer. (c) Sensibleheat flux assumingpresenceof mineral soil only, (d) mineralsoilwith surfaceorganiclayer.(e) Latentheatfluxassuming presenceof mineralsoilonly,(f) mineral soilwith surfaceorganiclayer.(g) Net ecosystem exchange(NEE) assuming presenceof mineralsoilonly,(h) mineral soilwith surfaceorganiclayer. Carbonflux (NEE) is positivetowardthe atmosphere.  from different boreal forest types obtained in the BOREAS experiment. The results showed both successes and deficiencies  in IBIS'  ability to simulate exchangesof massand energyover boreal  ecosystems. Although evapotranspiration was generallysimulated reasonablywell, sensibleheat fluxeswere systematically underestimated.The latter problem was more evident at a mature deciduousforest site (SOA) than at coniferoussites  EL MAAYAR  ET AL.: VALIDATION  OF IBIS  14,353  20 10  -10  20  10 0 -113 I  20  t  25-50  cm  10 5  0 -5 50-100  10  cm  ..  -5 I  90  110  130  150  170  190  I  I  210  230  I  I  250  270  Day of year Figure 12. Observed(dotted line) and simulateddaily soil temperatureat old black sprucesite.Thick solid line indicatesassumedpresenceof mineral soil only; thin solidline indicatesassumedpresenceof mineral soil with surfaceorganiclayer.  (SOJP and NOBS). A sensitivitytest and comparisonwith measuredcanopyheat storagedata showedthat this problem could be related to the physicalparameterizationsused to represent heat transfer between vegetation and atmosphere. Evidently,further work is required to improve the representation of canopyheat storagein the model. Net ecosystemexchangeof carbon (NEE) was generally simulatedacceptably,althoughfor the SOA site, summerand winter estimates showed appreciable bias compared to the meansof all data (see Table 2). Furthermore, midday assimilation wasgenerallyoverestimated,which appearsto be due to underestimationof daytimerespirationrates.The origin of this problem is not easyto determine:it could result from underestimationof soil respirationtemperaturecoefficients(we do not have measuredsoil respirationdata), or it couldbe due to undereslimatingsoil carboncontent(see above,section5.1.). Future analysesbased on field measurementsare required. Careful validation of the soil biogeochemistrycomponentof IBIS againstextensivefield measurementsof soil respiration shouldto help to resolvethis problem. Simulated ecosystemrespirationwas also evidentlyoverestimated duringthe winter period at SOA. This, combinedwith a generaltendencyto overestimatenighttime respiration,may suggestthat the temperatureresponsesfor respiration(of soil and/or vegetation)are too shallow.Furthermore, for a better evaluationof NEE simulatedby suchDGVMs, more complete  understandingis needed of the problemsin usingeddy correlation over tall canopiesat nighttime when wind speedsare low, availableenergyis small and atmosphericstabilityis high [e.g., Lee, 1998; Baldocchiet al., 1998]. Hence it is to be expected that the model can be further improved as technical measurementproblemsare resolvedor alternative techniques are developed. Both observed  and simulated  mean  Bowen  ratios for conif-  eroussites(SOJP and NOBS) were greater than 1.0, whereas those for the deciduoussiteswere typicallymuch lower. This demonstratesthat IBIS is able to reproduce correctly one of the fundamental  differences  between  deciduous  and conifer-  dominated boreal ecosystems.The importance of predicting these differencesin the Bowen ratio successfullyis related to the influence  of H  and LE  on mesoscale  and local weather  conditions(seethe reviewof Pielkeet al. [1998]andHogget al. [2000]). This implies that IBIS simulationsof changesin vegetation, e.g., due to globalwarming effectson speciescompetition and physiology,will lead to changesin the surfacefluxes used as lower boundary conditionswhen IBIS is coupled to a GCM.  The results of this study also showed the importance of includingan organiclayer in simulatingsurfaceenergycarbon fluxesfor boreal forest stands.The presenceof surfaceorganic material affectssoil moisture and temperature, as well as soil heat flux, soil thawing, surface energy partitioning, and sur-  14,354  EL MAAYAR  ET AL.: VALIDATION  face-atmospherecarbon exchange.For instance,inclusionof the OSL in the model led to appreciableimprovementsin simulatedsoil heat flux and surfacetemperature,providingan additionalexplanationfor why existingland surfacemodels tend to overestimatesoil "skin" temperature and downward soil heat flux. Bettset al. [1993] and Chenet al. [1997] related this problemto the numericalalgorithmsadoptedto solvethe surfaceenergybalance(see alsoBeljaars[1991]for numerical details). Resultsfor soil temperatureat the NOBS site also showedthat the inclusionof a surfaceorganiclayer delaysthe start of thawing,whichin turn affectssoilmoistureavailability, althoughit is clearthat with increasingdepth,the simulationof thisdelayis muchlesssatisfactory (seeFigure12d).Thiseffect is of considerable  interest  OF IBIS  ond,theyprovidea verygoodbasisfor estimatingthe probable errorsin simulationsof canopyexchangeprocessesat regional and global scales.  Acknowledgments. We thank ENergy from the FORest (ENFOR), fundedby the CanadianFederalPanelon EnergyResearchand Development (PERD), who provided the financial support of this study.Brian Amiro, Robert Grant, Ted Hogg, and three anonymous reviewersprovidedcarefulreviewsof an earlier versionof this manuscript.AMS mesonetdata were providedby the SaskatchewanResearch Council.  We wish to thank M. Goulden  and D. Baldocchi  for  making,respectively,NOBS and SOJP data availableto the BOREAS researchcommunity.  because several studies have dem-  onstratedthe importanceof correctestimationof soilmoisture content on the Bowen ratio, which in turn affectsthe development of the atmosphericboundarylayer, and henceboth local and regionalatmosphericconditions(an extensivereviewcan be foundin the work of Bettset al. [1996]).Moreover,Goulden et al. [1998]showedthe importanceof soildepthand thaw on boreal forest carbon balance.It will be particularlyimportant in spatialmodelingtherefore that the physicalcharacteristics of both mineral and organicsoil layersbe properlyparameterized. In order to achieve this difficult goal a starting point could be to test land surfaceparameterizationsat a range of sites,usingmeasureddata for the vertical distributionsof soil texture, soil thermal and hydraulicproperties,soil carboncontent, and root biomass.Unfortunately, extensivedata setscontaining reliable values for these variablesare not generally available. Future field campaignsshould attempt to report suchinformation,as suggested alsoby Delireand Foley[1999]. For instance,someof the disagreementbetweenobservedand simulatedsoil temperature and moisture, and hence energy and carbonfluxes,obviouslyoriginatesfrom the use of a constantsoiltextureprofile in our simulations.Concerningsurface organic material, Enrique et al. [1999] attempted to quantify the effect of plant-residuemulchon energyexchangebetween a grasslandecosystemand the atmosphere.A comparable, physicallybased,approachcouldbe developedfor forestcanopies,but for organicmaterial in varyingstatesof decomposition.  References Baldocchi, D. D., C. A. Vogel, and B. Hall, Seasonalvariation of carbondioxideexchangerates above and below a boreal jack pine forest,Agr/c.For. Meterol.,83, 147-170, 1997a. Baldocchi, D. D., C. A. Vogel, and B. Hall, Seasonalvariation of energyand water vapor exchangerates above and below a boreal jack pine forest,J. Geophys. Res.,102(D24), 28,939-28,952,1997b. Baldocchi,D., K. Wilson, and K. T. Paw U, On measuringnet ecosystem carbonexchangein complexterrain over tall vegetation,paper  presented at 23rdConference onAgricultural andForestMeteorology,Nov. 2-6, Am. Meteorol. Soc.,Albuquerque,N.M., 1998. Ball, J. T., I. E. Woodrow, and J. A. Berry, A model predictingstomatal conductanceand its contributionto the control of photosynthesisunder different environmentalconditions,in Progressin PhotosynthesisResearch,vol. 4, edited by J. Biggins, pp. 221-224, Martinus-Nijhoff, Zoetermeer, Netherlands, 1986. Beljaars,A. C. M., Numericalschemesfor parameterizations, in NumericalMethodsin Atmospheric Models,vol. 11, pp. 1-42, Eur. Cent. for Medium Range Weather Forecasts,Reading, England, Sept. 1991.  Bellisario,L. M., L. D. Boudreau,D. L. Verseghy,W. R. Rouse and P. D. Blanken, Comparingthe performanceof the CanadianLand Surface-Scheme(CLASS) for two subarcticterrain types,Atmos. Ocean,38(1), 181-204, 2000. Bessemoulin, P., and D. Puech, BOREAS TF-06 SSA-YA Surface  energy flux and meteorologicaldata, http://www-eodis.ornl.gov/, Distrib. Active Archive Cent., Oak Ridge Laboratory, Oak Ridge, Tenn., 1998.  Betts, A. K., J. H. Ball, and A. C. Beljaars, Comparisonbetween the land surfaceresponseof the European Centre Model and the FIFE1987 data, Q.J.R. Meteorol. Soc., 119, 975-1001, 1993. Betts,A. K., J. H. Ball, A. C. Beljaars,M. J. Miller, and P. Viterbo, The land-surface-atmosphere interaction:A review based on observational and globalmodelingperspectives, J. Geophys. Res.,101(D3),  The importanceof theseeffectsof the organicsurfacelayer on simulated exchangesof water and carbon is, of course, dependentupon its thickness.This studyshowsthat consider7209-7225, 1996. ation of the OSL in the model has only a minor effect on Black, T. A., and Z. Nesic, Boreal forest atmosphereinteractions: simulated water and carbon fluxes, for both deciduous and  Exchangesof energy,water vapor and tracesgases,Distrib. Active  when this was taken into consideration  fluxesin and above a boreal aspenforest, Global ChangeBiol., 2,  Archive Cent., Oak Ridge Natl. Lab., Oak Ridge, Tenn., 1998. coniferoussiteswhere the organicmaterial thicknessdoesnot (Available at http://www_eodis.ornl.gov) exceeda few centimetres.For the northern black sprucesite, Black, T. A., G. Den Hartog, H. H. Neumann, P. D. Blanken, P. C. however,wherethe organiclayeris 50 cm thick,the simulation Yang, C. Russell,Z. Nesic, X. Lee, S. G. Chen, R. Staebler, and M.D. Novak, Annual cyclesof water vapour and carbon dioxide of both latent heat flux and NEE were substantiallyimproved in the model.  This studyconfirmsand buildsupon similarwork by Delire and Foley [1999] who applied IBIS to other vegetationtypes includinggrasslandand tropical forest ecosystems. Suchstudies help to confirmthe validity of the processcomponentsof IBIS and hence its value as a global ecosystemsimulator. Although other aspectsof the model still require validation, e.g.,competitionand phenology,validationssuchas ourshave two major benefits.First, theyincreaseconfidencethat the use of dynamicvegetationsimulatorswill improvethe representation of land surfaceeffectswhen coupledto GCMs, and hence allow climate modelersto improvepredictionsof how vegetation feedbacksmay affect climate as the climate changes.Sec-  159-168, 1996.  Bonan,G. B., D. Pollard,and S. L. Thompson,Effectsof borealforest vegetationon global climate,Nature, 359, 716-718, 1992. Brutsaert,W., Evaporationinto the Atmosphere,299 pp., D. Reidel, Norwell, Mass., 1982.  Calvet, J. C., J. Noilhan, J.-L. Roujean, P. Bessemoulin,M. Cabelguenne,A. Olioso, and J.-P. Wigneron,An interactivevegetation SVAT model tested againstdata from six contrastingsites,Agric. For. Meteorol., 92, 73-95, 1998.  Campbell,G. S., and J. M. Norman,An Introductionto Environmental Biophysics, 286 pp., Springer-Verlag,New York, 1998. Chen, T. H., et al., CabauwExperimentalresultsfrom the Projectfor Intercomparison of Land-surface Parameterization Schemes, J. Clim., 10, 1194-1215, 1997.  Chen, W. J., et al., Effects of climaticvariability on the annualcarbon  EL MAAYAR  ET AL.: VALIDATION  sequestrationby a boreal aspenforest,Global ChangeBiol., 5, 41-53, 1999.  OF IBIS  14,355  Larson, J. A., The Boreal Ecosystem,500 pp., Academic, San Diego, Calif., 1980.  Ciais, P., P. Tans, M. Trolier, J. White, and R. Francey, A large  Lee, X., On micrometeorologicalobservationsof surface-airexchange NorthernHemisphere terrestrial CO2sinkindicated by•3C/•2Cratio over tall vegetation,Agric. For. Meteorol.,91, 39-49, 1998. of atmosphericCO2, Science,269, 1098-1102, 1995. Loyd, J., and J. A. Taylor, On the temperature dependenceof soil Clayton,J. S., W. A. Ehrlich, D. B. Cann, J. H. Day, and I. B. Marshall, respiration,FunctionalEcol., 8, 315-323, 1994. Soilsof Canada,vol. 2, Inventory,239 pp., Res. Branch,Can. Dep. of Manabe, S., and K. Bryan, Climate calculations with a combined Agric., Ottawa, Ontario, 1977. ocean-atmospheremodel, J. Atmos. Sci., 26, 786-789, 1969. Collatz, G. J., J. T. Ball, C. Grivet and J. A. Berry, Physiologicaland Manabe, S., and R. T. Wetheraid, Large-scalechangesof soil wetness environmentalregulationof stomatalconductance,photosynthesis inducedby an increasein atmosphericcarbondioxide,J.Atmos. Sci., 44, 1211-1235, 1987. and transpiration:A model that includesa laminar boundarylayer, Agric. For. Meteorol., 53, 107-136, 1991. Nijssen,B., I. Haddeland, and D. P. Lettenmaier, Point evaluation of Collatz, G. J., M. Ribas-Carbo, and J. A. Berry, Coupled photosyna surfacehydrologymodelfor BOREAS, J. Geophys. Res.,102(D24), 29,367-29,378, 1997. thesis-stomatalconductancemodel for leavesof C4 plants,Aust. J. Plant Physiol.,19, 519-538, 1992. Pielke, R. A., and P. L. Vidale, The boreal forest and the polar front, CuencaR. H., D. E. Strangel,and S. F. Kelly, Soil water balancein a J. Geophys.Res.,100(D12), 25,755-25,758,1995. boreal forest,J. Geophys.Res.,102(D24), 29,355-29,367,1997. Pielke, R. A., T. J. Lee, J. H. Copeland,J. L. Eastman, C. L. Ziegler, Delire, C., and J. A. Foley, Evaluating the performance of a land and C. A. Finley, Use of USGS-provideddata to improveweather surface/ecosystem model with biophysicalmeasurementsfrom conand climate simulations,Ecol. Appl., 7, 1053-1069, 1997. trasting environments,J. Geophys.Res., 104(D4), 16,895-16,909, Pielke, R. A., R. Avissar,M. Raupach,A. J. Dolman, X. Zeng, and 1999. A. S. Denning, Interactionsbetweenthe atmosphereand terrestrial Dickinson,R. E., A. Henderson-Sellers,P. J. Kennedy,M. F. Wilson, ecosystems: Influence on weather and climate, Global ChangeBiol., 4, 461-475, 1998. Biosphere-Atmosphere Transfer Scheme(BATS) for the NCAR CCM, Tech. Note NCAR/TN-275-STR, Natl. Cent. for Atmos. Res., Pollard, D., and S. L. Thompson, Use of a land-surface-transfer 69 pp., Boulder, Colo., 1986. scheme(LSX) in a globalclimatemodel:The responseto doubling Enrique, G.-S., I. Braud,J.-L. Thony,M. Vauclin, P. Bessemoulin,and stomatalresistance,Global Planet. Change,10, 129-161, 1995. J.-C. Calvet, Modelling heat and water exchangesof fallow land Satterlund, D. R., An improved equation for estimatinglong-wave coveredwith plant-residuemulch.Agric.For. Meteorol.,97, 151-169, radiationfrom the atmosphere,•5(6), 1649-1650, 1979. 1999. Schneider,S. H., and R. E. Dickinson,Climate modeling,Rev. GeoFarquhar, G. D., S. von Caemmerer, and J. A. Berry, A biochemical phys., 12, 447-493, 1974. model of photosyntheticCO2 assimilationin leavesof C3 species, Sellers,P. J., Y. Mintz, Y. C. Sud, and A. Dalcher, A Simple Biospher Planta, 149, 78-90, 1980. model (SiB) for usewithin generalcirculationmodels,J. Atmos.Sci., 43, 505-531, 1986. Foley, J. A., J. E. Kutzbach, M. T. Coe, and S. Levis, Feedbacks betweenclimate and boreal forestsduringthe holoceneepoch,NaSellers,P. J., D. A. Randall, G. J. Collatz, J. A. Berry, C. B. Field, D. A. ture, 371, 52-54, 1994. Dazlich, C. Zhang, G. D. Collelo, and L. Bounoua, A revisedland Foley, J. A., I. C. Prentice, N. Ramankutty, S. Levis, D. Pollard, S. surface parameterization(SiB2) for atmosphericGCMs, part I, Sitch, and A. Haxeltine, An integrated biosphere model of land Model formulation,J. Clim., 9(4), 676-705, 1996. surfaceprocesses,terrestrial carbonbalance,and vegetationdynam- Sellers,P. J., et al., BOREAS in 1997:Experiment overview,scientific ics,GlobalBiogeochem. Cycles,•0(4), 603-623, 1996. results,and future directions,J. Geophys.Res., 102(D24), 28,73128,769, 1997. Gates, D. M., Plant-Atmosphere Relationships,92 pp., Chapman and Hall, New York, 1993. Shao,Y., and A. HendersonSellers,Modeling soil moisture:A project Goulden, M. L., B.C. Daube, S. M. Fan, D. J. Sutton, A. Bazzaz, J. W. for Intercomparison of Land Surface Parameterization Schemes Munger, and S.C. Wofsy, Physiologicalresponsesof a black spruce Phase2(b), J. Geophys. Res.,101(D3), 7227-7250, 1996. forestto weather,J. Geophy.Res.,102(D24), 28,987-28,996,1997. Shewchuk,S. R., Surface mesonet for BOREAS, J. Geophys.Res., Goulden, M. L., et al., Sensitivityof boreal forest carbon balance to 102(D24), 29,077-29,082, 1997. soil thaw, Science, 279, 214-217, 1998. Shukla,J., and Y. Mintz, Influenceof land-surfaceevapotranspiration on the Earth's climate, Science, 215, 1498-1501, 1982. Gower, S. T., J. G. Vogel, J. M. Norman, C. J. Kucharik, S. J. Steele, and T. K. Stow, Carbon distributionand abovegroundnet primary Steele, S. J., S. T. Gower, J. G. Vogel, and J. M. Norman, Root mass, productionin aspen,jack pine, andblacksprucestandsin Saskatchenet primary productionand turnover in aspen,jack pine and black spruceforestsin Saskatchewanand Manitoba, Canada, TreePhysiol., wan and Manitoba, Canada,J. Geophy.Res., 102(D24), 29,02929,041, 1997.  Guyot, G., Climatologiede l'Environnement,505 pp., Masson,Paris, 1997.  Henderson-Sellers, A., and V. B. Brown,Project of the Intercomparison of Land-surfaceParameterizationSchemes(PILPS): First science plan, IGPO PuN. Set., vol. 5, 53 pp., 1992. Henderson-Sellers, A., K. McGuffie, andA. J. Pitman,The projectfor Intercomparison of Land-surface Parameterization Schemes (PILPS): 1992 to 1995,Clim. Dyn., 12, 849-859, 1996. Hogg, E. H., D. T. Price, and T. A. Black, Postulated feedbacksof deciduousforest phenologyon seasonalpatters in the western Canadianinterior,J. Clim., •3(24), 4229-4243, 2000. Jackson,R. B., J. Canadell,J. R. Ehleringer,H. A. Mooney, O. E. Sala, and E. D. Schulze,A global analysisof root distributionfor terrestrial biomes, Oecologia,108, 389-411, 1996. Kimball, J. S., P. E. Thornton, M. A. White, and S. W. Running, Simulatingforest productivityand surface-atmospherecarbon exchangein the BOREAS study region, Tree Physiol.,17, 589-599, 1997.  17, 577-587, 1997.  Tans, P. P., I. Y. Fung, and T. Takahachi,Observationalconstraintson the global atmosphericCO2 budget,Science,247, 1431-1438, 1990. Verma, S. B., J. Kim, and R. J. Clement,Momentum,water vapor, and carbon dioxide exchangeat a centrally located prairie site during FIFE, J. Geophys.Res., 97, 18,629-18,639, 1992. Verseghy,D. L., N. A. McFarlane, and M. Lazare, CLASS-A Canadian land surface scheme for GCMs, II, Vegetation model and coupledruns, Int. J. Climatol., 13, 347-370, 1993. P. Bessemoulin,Maltrio-France,ServiceCentral d'Exploitation de la Mdtdorologie,42 Avenue G. Coriolis, 31057 Toulouse Cddex,France. A. Black, Department of Agroecology,Universityof British Columbia, Vancouver, BC, Canada V6T 124.  C. Delire and J. A. Foley, Center for Sustainabilityand the Global Environment, Institute for Environmental Studies,University of Wisconsin, 1225 West Dayton Street, Madison, WI 53706. M. E1Maayar and D. T. Price, Natural ResourcesCanada, Canadian Forest Service, 5320-122 Street, Edmonton, AB, Canada T6H 3S5.  Kucharik, C. J., J. A. Foley, C. Delire, V. A. Fisher, M. T. Coe, J. (elmaayar@nrcan.gc. ca) Lenters, C. Young-Molling, N. Ramankutty, J. M. Norman, and S. T. Gower, Testingthe performanceof a dynamicglobalecosystem model: Water balance, carbon balance and vegetation structure, (ReceivedMay 31, 2000; revisedFebruary20, 2001; GlobalBiogeochem. Cycles,•4(3), 795-825, 2000. acceptedFebruary27, 2001.)  


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items