JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 106,NO. D24, PAGES 33,543-33,550,DECEMBER 27, 2001
Intercomparisonof BOREAS northern and southernstudy area surface fluxes in 1994
AlanG.Barr, • AlanK. Betts, 2T. A. Black, 3J.H. McCaughey, 4andC. D. Smith • Abstract. Sensibleandlatentheatfluxesfrom the BorealEcosystemand Atmosphere Study(BOREAS) towerflux sitesin 1994 are analyzedover bothdiurnaland seasonal cycles.We compareandcontrastthesouthernandnorthernstudyareasandthebehaviorof five differentland covers.For eachland coverthe evaporativefractionsandsurface conductances to watervaporarehigherin the souththanin the north,with the rankingfrom largestto smallest:aspen,fen, blackspruceandjack pine. The coniferand,particularly,the jack pine sitesshowthe greateststomatalcontrolof transpiration,asthe vaporpressure deficit increasesfrom morningto afternoonand as the soil driesduringperiodswith low precipitation.The relationbetweensurfaceconductance andthe Priestley-Taylor coefficient(x is consistentbetweensc)uthern andnorthernstudyareasbut variesamongland covers.The aspenandten siteshave higherc• valuesthanthe landscapemean,andthe matureconifer siteshavelower c• valuesthanthe landscapemean.We attributethe diflbrencesto the impactof spatialheterogeneityat the landscapescale. 1. Introduction
2. Data Setsand Processing
The processes that governthe exchangesof heatandwater vapor betweennaturalecosystems ,andthe atmosphereplay a key role in the global climate system.These processesare partially under the control of surfacevegetationvia canopy composition and structure, leaf area index and plant ecophysiology,specifically,the interactionsbetweenstomata (i.e., leaf or canopy conductance)and the controlling environmental variables, including light, temperature, humidity,and soil moisture.From a climaticperspectivethe surfaceexchangeis mainly controlledby surface available energy (net radiation minus storage) and its partition into
This studyanalyzestower flux data from the 1994 field phaseof BOREAS, collectedat nine sites(Table 1) between May 24 and September19. The BOREAS southernand northernstudyareas(SSA andNSA, respectively)hadpaired sites for /'our land covers:•nature(old) black spruce(wet
sensible and latent heat Ilux over both diurnal
all sites, with subtle site-to-site differences in instrumentation
conifer)(SOBS and NOBS), maturejack pine (dry conifer) (SOJPandNOJP),youngjack pine(SYJPandNYJP) andfen (Sfen and Nfen). In addition,the southernstudyareahad a matureaspen(deciduous)site (SOA). Sensibleand latent heat flux densities(H and AE) were
measuredevery half hour by the eddycovariancemethodat
and seasonal
cycles. Few field experiments haveattemptedto characterize these exchangesat the landscapescale. The Boreal EcosystemAtmosphereStudy (BOREAS) was one such experiment, designedto evaluatethe role of boreal forestsin the global climate systemand in global change [Sellers et al., 1995, 1997]. In this paper,we summarizethe water and heat flux data from the BOREAS 1994 field campaignsover both diurnalandseasonal cycles.We alsocompareandcontrastthe BOREAS northernandsouthernstudyareasandthe five land coversrepresented by theBOREAS towerflux sites.
and signalprocessing[Newcomeret al., 2000]. Supporting meteorologicalmeasurements includedsolar and net radiation (Rs and Rn, respectively), air temperature(T,), relative humidity (RH), vapor pressuredeficit (D, defined as the difference between saturation vapor pressure at air temperatureand ambientvaporpressure),and wind speed(u), again with some site-to-site differences in instrumentation [Newcomer et al., 2000].
2.1. Evaporative Fraction and Priestley-TaylorAlpha
We will usetwo derivedvariablesto describethe partition of availableenergyinto the sensibleandlatentheatfluxes, the evaporativefraction(EF) and the Priestley-Taylorcoefficient •Climate Research Branch, Meteorological Serviceof Canada, a [Priestleya•ut Taylor, 1972]. The valuesfor EF and a were estimated for each half hour as
Saskatoon,Saskatchewan,Canada.
2Atmospheric Research, Pittsford, Vermont. •Faculty of Agricultural Sciences, University of British Columbia,
EF=•
Vancouver, British Columbia, Canada.
(1)
(H +XE)
4Department ofGeography, Queens University, Kingston, Ontario, and Canada.
or= Copyright2001 by theAmericanGeophysical Union. Papernumber2001JD900070. 0148-0227/01/2001 JD900070509.00
(s+7)9.E s(H + 2E)
,
(2)
where 2'is the psychrometric constantand s is the derivative of saturationvaporpressurewith respectto temperature.Our formulationand use of a dif/Ers from that of Priestleyand 33,543
33,544
BARR ET AL.' NORTHERN AND SOUTHERN STUDY AREA FLUXES IN 1994
Table 1. BOREASTowerFluxSitesUsedIn ThisStudy Site
Latitude, Longitude, BOREASPrincipal øN øW Investigator a
Reference
Southernmatureaspen(SOA)
53.63
Black etal. [1996]
-106.20
T.A. Black (TF-1);
G. denHartog(TF-2) Southernmatureblackspruce(SOBS) Southernmaturejack pine(SOJP) Southernyoungjack pine(SYJP)
53.98 53.92
-105.12 - 104.69
P.J. Jarvis(TF-9) D D. Baldocchi (TF-5)
53.88
-104.65
D E. Anderson (TF-4)
Southern fen (Sfen)
53.80
-104.62
S B. Vem•a (TF-11)
Northernmatureblackspruce(NOBS) Northernmaturejack pine(NOJP) Northernyoungjack pine½NYIP)
55.88 55.93 55.90
-98.48 -98.62 -98.29
S C. Wofsy(TF-3) DR. Fitzjarrald(TF-8) J.H. McCaughey(TF-10)
Suykeret aL [1997] Gouldenet al. [ 1997] Fit•jarrald etal. [ 1995 McCaugheyet at [ 1997]
Northern fen (Nfen)
55.91
-98.42
D.E. Jdinski (TF- 10)
LafleuretaL [1997]
Jarvis et at [1997] Baldocchietal. ['1997] Andersonetal. [1995]
a'IF denotes theBOREASTowerFluxgroup.
Tco'lor[1972] in two respects. First, to avoidthe issueof energybalancenonclosure we havesubstituted (H + AE)for
from near zero at SOBS to 44% at NOBS, +7% at S YJP,
+9% at SOJP, +15% at SOA, +19% at NYJP, +27% at Sfen,
(R,•- Q,), where Q, is the sum of the minor energybalance +31% at Nfen, and +38% at NOJP. The closureadjustments terms (section2.2). If we assumethat the measuredBowen were appliedonly whenthe Bowenratio was well defined were appliedconsistently ratio (H/XE) is correct,thissubstitution in both(1) and(2) has [Ohmura, 1982]. The adjustments the sameeffectasadjustingH andAE to forceenergybalance at all sites and, in the absence of additional information, solutionto theclosureproblem closure(section2.2). Second,unlike Priestleyand Taylor's providethe mostreasonable originalenergy-limited conceptanddefinition,our useof a, [Twine et al., 2(}00]. Still, the large, unexplaineddift•rences whichis diagnosticratherthanpredictive,encompasses both in closureamongsitesare a causetbr concern.The large energyimbalances at somesites(thefens,in particular)•nay energy-limited andsoilwater-limitedconditions.
be duein partto theunderestimation of Q•. 2.2. Energy BalanceClosure 2.3.
The surfaceenergybalancemaybe writtenas
Rn-Qs=H +AE
(3)
and
Qs=Qs +QI, +Qa +Qv +Qc,
(4)
where Q.• is the groundheat flux, Qt, is the rate of
Surface Conductance
The surfaceconductance to watervaporg• was calculated
fromthe closure-adjusted valuefor AE (gE') usingthe invertedform of the Penman-Monteithcombinationequation [Monteith, 1981]'
gs=
'
,
(5)
[s(R•-Qs- AE*)-rAE*]/ ga+pcpD aboveground biomassheatstorage,Qa andQ•.aretheratesof sensibleand latentheat storage,respectively,in the air layer conductance to heatand water below the eddy flux measurement level, and Q½is the whereg• is the aerodynamic of air,cvis thespecific heatof air,and photosynthetic energyflux. Four of the ninetowerflux sites vapor,p isthedensity deficit.The valuefor s in (5) was hadcompletemeasurements of thetermsin (4): SOA, SOJP, D is the vaporpressure NYJP and Nfen. At siteswhere Qs was not measured,Qswas estimated at the mean of surface temperatureand air wheresurface temperature (T,) wascomputed as estimatedas f(R,), using a fourth-orderpolynomialwith temperature, coefficients fit to data from the most similar site where Qs H* Ts = Ta +•. (6) wasmeasured (e.g.,Nfenfor Sfen,NYJP for SYJP,SOJPfor t9Cp g a
NOJP, SOJP for SOBS, and SOJP for NOBS). These
estimatesare only approximatebut are the best available. The valuefor g,•wascalculated, followingThom[1972]and However, even with the measuredor estimatedQ,, the surface
energy balance (equation 3) does not close. Energy imbalances are commonin eddycovariancestudies[see,e.g., Bart etal.,
1994; Twine etal.,
2000], but their cause is
Verma [ 1989], as 2
ga=
" , It+ B-1lt,
(7)
uncertainand may vary amongsites.Possiblecausesinclude velocity andB4, thedimensionless eddycovariancemeasurement errorsor limitations,violation whereu, is thefriction of eddycovariance assumptions [Mahrt, 1998] anderrorsin sublayerStantonnumber,was set to 2 (forest)or 4 (fen) the measurementof R,•- Qs.
Therefore,before calculatingthe surfaceconductance to watervapor(equation5), we resolvedthe energyimbalance in (3) by adjustingH andAE to forceenergybalanceclosure, assuming,as in the calculationof EF and a above,that the measuredBowen ratio was correct [Barr etal., 1994; Blanken etal., 1997; Twine etal., 2000]. We will denotethe closure-
[Thom, 1972; Wu etal., 2000]. The value for u. wasestimated on thebasisof the stabili:ycorrectedlogarithmicwind profile [see,e.g., Brutsaert,1984],with the zero-planedisplacement
andtheroughness lengthfor momentum setto 64% and13% of the canopyheight,respectively.At the six siteswhere measurementsof u. were available, there was general
agreement betweenthe measuredand derivedvaluesfor u.,
of 0.11m s4 anda root-mean-square adjusted values ofH andAEasH* andAE*,respectively. The witha meandifference
meanadjustment to half-hourlyH andXE variedamongsites, differenceof 0.21 m s4. We usedthe derivedvaluesfor u, in
BARR ET AL.' NORTHERN AND SOUTHERN STUDY AREA FLUXES IN 1994
(7) to be consistentamong sites, but the effect of using derived or measured u, was small.
It is difficult to estimateerrorlimits for EF (equation1), or (equation2) andgs(equation5), because we lackindependent estimates of theseparate errorsin H andAE.It is possible that the errorsin EF, or,and gs are smallesttbr siteswith small energy imbalances(in sequencefrom smallestto largest: SOBS, NOBS, SYJP, SOJP,SOA, andNYJP) andlargestfor siteswith largeenergyimbalances (the fens andNOJP),but this is only tentativebecausethe actualcauseof the energy imbalanceis not known. It is equally possiblethat by using (la) and (2) to skirt the closure issue, we have effectively
33,545
3. Comparison of SSA and NSA Sitesin the 1994 Growing Season 3.1. SeasonalCycle of EvaporativeFraction Figure 1 showsthe seasonalcyclesof (1) air temperature and relative humidity (10-day averages), (2) evaporative traction (10-day averages),and (3) precipitation(10-day totals) for the SSA and NSA flux sites.Values of Ta and RH
are averagedoverall sites.For precipitation we showa single averagein the NSA and two valuesin the south;one for the SSA aspensite and the other for an averagerepresentative of the other SSA sites, which are in a cluster ~100 km to the
minimizedthe errorsin EF and a relatedto energybalance east-north-east of the aspensite.Note the differencein early nonclosure. surmnerprecipitationbetweenthe two. There is a cool wet period, with high RH in early summer (with more precipitationin the SSA than in the NSA), followed by a 2.4. Filling Gaps in Data warmer, drier period, with little rainfall, toward the end of AppendixA gives•noredetailson how we filled gapsin
July. Amongthe conifersites,EF at thejack pine sites,which have permeablesandysoils,is more sensitiveto precipitation, were filled usingequations(3) and (A3), with missingvalues while EF at the black sprucesites,where the water table is for gsin (A3) estimatedusingan empiricalfunction(equation high and the organic soil has high water retention, varies A 1) of Rs,D, and T•,.Gapsin meteorological data were filled much less with precipitationand has a slight upward trend using linear regression •nodels and data from nearby over the season.EF is much higher at the deciduousaspen BOREAS mesonetor towerflux sites.Gapsin Qswere filled site, increa•sing rapidly with leaf out in late May, and at the with modeledvalues,estimatedasf(R,,) usinga fourth-order fen sites,where the vegetationis •so mostlydeciduous.The polynomial with site-specific coefficients fit to the entire aspenand I•n sitesshow a tnidsmnmerEF peak, /bllowing period. the early summer rains. Figure 1 also shows that mean
thefluxandconductance data.Briefly, gapsin H* andAE*
18
o... 16
70 -•
14
60 a:
12
5O
0.8
tu 0.6 0.4
m 0 n JI
g SepJunJulAugSop
Figure1. Seasonal variationin (top)air temperature T• (solidlines)andrelativehumidityRH (dashed lines), (middle)evaporative fractionEF, and(botto•n)rainfall1¾om BOREAStowerflux sitesin the southern study areaSSA andnorthern studyareaNSA duringthe 1994growingseason. The dataare 1O-dayaverages tbr T•, RH, andEF and 10-daytotalstbr rainfall.Beforecalculating EF, we firstaveraged the l{)-dayfluxes.The EF line style denotedtbr each land cover type is the samefor sitesin the SSA and NSA. Abbreviationsare as
follows:OA, matureaspen; OBS,matureblackspruce; OJP,maturejack pine;andYJP,youngjack pine.
33,546
BARRET AL.' NORTHERNAND SOUTHERNSTUDYAREAFLUXESIN 1994
2O
o,.,
15 lO
(b) 0.8
m 0.6 0.4
(c)
400
200 ............... .........
ß
6
12
18
' ...
,
-OBS -OJP ____ yJp ..................... Fer•
(d)
,
24
UTC
Figure2. Diurnalvariationin (a) air temperature T, (solidlines)andrelativehumidityRH (dashed lines),(b) evaporative fractionEF, (c) solarradiation Rs(solidlines)andnetradiation R, (dashed lines),and(d) surface conductance to watervaporgst¾om BOREAStowerfluxsitesin theSSAandNSA,averaged between May 24 andSeptember 19, 1994.We firstaveraged thefluxesby timeof daybetbrecalculating EF. The mean surfaceconductance was calculatedfrom measureddataonly (i.e., with no gapfilling) afterexcludingvalues belowthetenthandabovetheninetiethpercentiles. The linestyledenotedfor eachlandcovertypeis thesame for Figures2b and2d. Localtimein theSSAandNSA is 6 hourslessthanUTC.
summerEF is higherin the SSA thanthe NSA for all four pairedsites,fen, old blackspruce,old jack pine and young jack pine,althoughthedifferenceis a little lessclearthanin Figure2 (seesection3.2).
3.2. Diurnal Cycleof EvaporativeFractionand Surface Conductance
Figure2 comparesthe diumal cycles(an averagefrom May 24 to September 19, 1994,corresponding essentially to the growing season)Of (1) air temperatureand relative humidity,(2) evaporative fraction,(3) solarandnetradiation, and(4) surfaceconductance for the BOREAS flux sitesin the SSA andNSA. As in Figure 1, the meteorological variables
are averaged overall sites.Figure2a showsan afternoon maximum of temperature anda _minimum of RH. Figure2b shows a daytime minimum of EF at all sites,withEF at the
SSAexceeding EFattheNSAforeachpaired landcover. In addition, Figure2b shows characteristically different diurnal patterns of EF for eachlandcovertype,independent of geographic location. At thefensites, EFincreases asRHfalls from an early morningmaximumto a midafternoon minimum.At the aspenand old black sprucesites,EF is relatively constant duringthemiddayhours.At thejackpine sites, EF falls the most and reaches the lowest afternoon minimum. These differences reflect the decreasing
availabilityof waterfor evaporation andtranspiration andthe strongest stomatalcontrolon transpiration at the jack pine
BARR ET AL.: NORTHERN AND SOUTHERN STUDY AREA FLUXES IN 1994
33,547
1.0
0.5
1
10
0.1
1
10
0.1
gs(mm s-1) Figure 3, Halt:hourlyvaluesof the Priestley-Taylorcoefficient(x as a functionof the surfaceconductance to
water vapor g.•fortheBOREAS tower fluxsites onMay24toSeptember 19,1994. The[}lots show measured
data only (i.e., with no gap filling). l)ata are excludedtbr solar irradianccs< 500 Wm". The solid line is a landscapemcanfrom equation(8), fit to the data from all sites.
sites. Mean RH is a little lower for the NSA than the SSA,
significantlyinfluencedby othcr land coverswith higher
consistent with the unitbrmlylower EF.
cval:x)rativcIYactions.
The SSA-NSA
difference
in EF is consistent with
the
earlier resultsof Barr and Betts [1997], who analyzed the
boundm'ylayer budgetsof the BOREAS radiosondcs. They reportedmeanlinddayBowenratiostBr the BOREAS NSA
Figure2c showsthe •ncandiurnalcyclesof solarea•dncl radiation, and Figure 2d shows the derived surface conductanceto water vapor. The SSA aspensite has the highestsurfaceconductance, and thejack pine siteshavethe
and SSA during the 1994 intensive field ca•npaignsthat
lowest.Unlike the fen sites,whereconductance is nearly
correspondto evaporativetYactionsof I).53 trod 0.45, respectively. Thesevaluesare intermediate betweenthe lower conifervaluesandthehigheraspenandfen valuesin Figure2
symmetric with radiation (more st) in the north than in the
south), the diurnal pattern of conductanceis markedly asymmetricat the tbrestsites.The high forestconductances in and are ~30% higherthanthe meanmiddayvaluestbr mature the early morning to midmorning reflect the maximum black spruce(0.43 in the SSA and0.34 in the NSA). These daytimestomatalopening•[sa resultof low-m-magnitude leaf differencesillustratethat althoughthe borealit)restlandscape water potentials,high RH, and low D [Mac•olis and Ryan, is dominatedby conifers,particularlyblack spruce[Bettset 19971.At somesitesand ti•ncsthey may also reflect the al., thisissueI, its energybalerace at the landscapescaleis also presenceof early morningdew on the canopy.The fall of
33,548
BARR ET AL.' NORTHERN AND SOUTHERN STUDY AREA FLUXES IN 1994
conductance betweenmidmorningand late afternoonat the forestsitesreflectsstomatalcontrolas D increases[Margolis and Ryan, 1997]. As was observedwith EF, the pairedsites(fen, old black
extensive fen. For patches like conifer with lower than averagevalues for gs, the horizontaladvectionof cooler, moisterair t¾omthe surrounding wetterpatchescausesD to be lowerthanthe equilibriumvalue.This diminishes gE and
spruce,oldjack pine,andyoungjack pine)eachhavehigher
causes a to be lower than that of an extensive coniferous
mean conductance in the SSA than in the NSA. There is little
landscape. However,the resultantAE and a are alsoaffected differencein incomingshort-waveradiationor in temperature by the presenceof a strongfeedbackfrom D to gs in betweenthe NSA and SSA, and the 5% lower middayRH in coniferousspecies.This t•edback increasesgs above its the NSA is not sufficient to accountfor the significantly equilibriumvalue anddampensthe reductions in/1E and a. at the landscapescale lowergs.The higherrainfallin 1994in theSSA mayexplain The net effect of surfaceheterogeneity part of the higherconductance, particularlyat thejack pine is to heightenthedifferencesin EF and a amongcontrasting land covers.
sites. However, Betts et al. [this issue] showed a similar
differencebetweenthe black sprucesites in 1996, when rainfall was similar in both SSA and NSA, so we doubt that
3.4. Impact of Energy BalanceClosure the seasonalatmospheric and soil waterconstraints between The analysisof section3.3 hasonecaveat,whichis related southandnoah are entirelyresponsible tbr the lowersurface to the energybalanceclosureadjustments in section2.2. With conductance in the north. Other possibilities include the exceptionof NOJP the sites with the largestenergy differences in nitrogenavailabilityandleaf areaindex. closureadjustments to gE and gs (the fens) also have the largestpositivedeviationsof a from (8), whereasthe sites 3.3. Couplingof Alpha to SurfaceConductance with thesmallestenergyclosureadjustments to AE andgs(the Figure3 showshalf-h,.mrly valuesof thePriestley-Taylor matureconifers)havethelargestnegativedeviationsfrom (8). coefficienta (equation2) as a functionof gs(equation 5) for If we repeat the analysisin Figure 3 but substitutethe each of the nine sites.The scalefor gs is logarithmic.The unadjustedvalues tbr a and g•, the contrastsbetweenland solidline in eachplot showsthe meanrelationship between• coversdiminishbut do not disappear.The averagedepartures and gsfor all sites,fit to a functionsuggested by Monteith of a t¾omthe landscapemeanbecome+7% for fen, +2% for [19951: mature aspen,+1% for youngjack pine, 0% for old black a = a m(1- exp[-gs/ gc1), (8) spruce,and-14% for old jack pine. Note that thepairedsites continueto show consistentdeparturesfrom the landscape
where amis the asymptoticlimit for a and gc is a scaling meanandthat theold black sprucesitesnow typify theboreal The revisedvaluesfor an andg½(equation 8) are conductance. Note the tightcouplingof a to gs.For the data landscape. 0.94 and 5.8 mm s '•, respectively. This caveat tempers our acceptance threshold ot Rs> 500 W m in Figure3 themean of horizontaladvectionin a (all site)estimates for amandgc(equation8) are 1.05and6.0 conclusionaboutthe significance
mms'•, respectively, inreasonable agreement withMonteith's patchworklandscape,but it does not invalidateit. It also of carefulmeasurement of Qsand [1995]estimates of 1.1to 1.4foramand5.0 mms'• forgo. highlightsthe importance understanding of theenergy The estimatefor g, (butnot an) is sensitive to theRsthreshold theneedfor a morefundamental usedtoscreen thedataanddrops to4.6mms'• forRs> 250 imbalancein eddycovariancestudies.
W m'2.Ourestimate for ammaybelowerthanMonteith's simplybecausehis analysisincludeddata from productive 4. Summary and Conclusions agriculturalsiteswith highervaluesfor a and gs thanthe This paper summarizesthe sensibleand latent heat flux
borealsitesin this study.
The general applicabilityof the mean (landscape) relationship t¾om(8) at all sitesshowsthe broadutility of relationships like (8). The pairedsites(old blackspruce,old jack pine, youngjack pine, and fen) havevery similara-gs relationships,independentof their geographiclocation.
data from the 1994 BOREAS
However,
dift•rences.The rankingof the evaporative fractionsby land cover is, in order of largestto smallest,aspen,fen, black spruce, and jack pine. Land cover differencesin EF are
the individual land covers show subtle but
consistentdeparturesfrom the mean a-gs relation.The averagedepartures of a from the landscape mean(equation (8), shownas the solid linesin Figure3) are +18% for fen, +4% for matureaspen,-1% for youngjack pine,-9% for old black spruce, and -11% for old jack pine. If the relationshipis indeeduniversalas, for example,de Bruin [ 1983],McNaughton andSpriggs[1989]andMonteith[ 1995] argue,thenthesedeparttires may showthe impactof spatial heterogeneity on evapotranspirationfrom contrasting elementsin a patchworklandscape. For patcheslike fen that are relativelywet and haveabove-average valuesfor gsand weak feedbackfrom D to gs,the surrounding drierpatchesact as
sensible
heat
sources
via
horizontal
advection
[McNaughton,1976]. This causesdisequilibrium between AE and D. The above-equilibrium D enhances XE from the fen patchesand causescx to be higher than that of an
tower flux sites and contrasts
thesouthernandnorthernstudyareasandthebehaviorof five dift•rent land covers.The data show consistentlyhigher evaporativefractionsand stirfaceconductancesin the south than
the
north,
with
no
obvious
relation
to climatic
attributedin part to differencesin surfaceconductanceand stomatal control and in part to the impact of spatial
heterogeneity on evapotranspirationfrom contrasting elementsin a patchworklandscape.The coniferousand, especially,the jack pine ecosystemsshow the strongest stomatalcontrolof transpiration. Both southernand northern studyareashavea similarrelationship betweenthe Priestley Taylor coefficient a and surface conductance,but the relationshipvaries subtlyamongland covers.The variation mayshowhow spatialheterogeneity in landcoverandsurface conductanceat the landscapescale influencesa among contrasting landscape elements. Spatial heterogeneity increasesa from patcheslike fen and aspen with higher conductance and decreasesa from coniferouspatcheswith
BARR ET AL.: NORTHERN AND SOUTHERN STUDY AREA FLUXES IN 1994
lower conductance.We conclude that although conifers dominatethe borealforestlandscape,other land coverswith higherevaporativefractionsalsoinfluenceits energybalance at the landscapescale.In evaluatingthe role of borealforest in the global climatesystem,it will be importantto consider thepatchy,mosaiccharacterof theboreallandscape.
33,549
[hreeanonymous reviewers wereconstmctive andhelpful. Financial supportwas providedto Alan Bettsby NASA undergrantNAG57377 andby NSF undergrantATM-9988618.CraigSmithreceived supportfrom the ClimateResearchBranchof the Meteorological Service of Canada and Environment
Canada's
Science Horizons
program.
References
Appendix A' FillingGapsinH*, J.E,and Data
To fill gapsin thetimeseries of H*, AE*,and$s,we modeledgs as ,•R, Ta, D) at each site, roughly following Jarvis [ 1976]:
gs= mgx,f(Rs)f (Ta)f (D) ,
(A1)
Anderson,D.E., R. Striegl, D. Baldocchi,and D. Stannard,The fluxesof CO2 andwatervapormeasured aboveandwithinyoung andmaturejack pineforestsof centralCanada,Paperpresented at InteractiveEnvironmental Effectson ForestStandsWorkshop,Int. Unionof For.Res.Organ.,New Zealand,January1995. Baldocchi,D.D., C.A. Vogel, and B. Hall, Seasonalvariationof energyandwatervaporexchange ratesaboveandbelowa boreal jack pine forestcanopy,J. Geophys. Res.,102, 28,939-28,952, 1997.
with
Ban',A.G. and A.K. Betts,Radiosonde boundary-layer budgetsabove a borealtorest,J. Geophys. Res.,102,29,205-29,212,1997. Ban', A.G., K.M. King, T.J. Gillespie,G. den Hartog, and H.H. Neumann,A comparisonof Bowen ratio and eddy correlation
,f(Rs)= Rs
(A2a)
R+ns
sensible and latent
heat Ilux
measurements
above
deciduous
forest, Boundary Laver Meteorol., 71, 21-41, 1994.
(A2b)
and using f(D) t¾omLohammeret al. [1980] (cited by Massnmnand Kaufmann[ 1991]): 1
.f(D) =•
(A2c)
I +bDD
Betts,A.K., J.H. Ball, and J.H. McCaughey,Near-surfaceclimatein the boreallbrest,J. Geophys.Res.,thisissue. Black, T.A., et al., Annualfluxesof water vaporand carbondioxide lluxesin andabovea borealaspenforest,Global ChangeBiol., 2, 101-11 l, 1996.
Blanken,P.D., T.A. Black,P.C. Yang, H.H. Neumann,Z. Nesic,R. Staebler,G. denHattog,M.D. Novak,andX. Lee, Energybalance and canopyconductance of a borealaspenforest:partitioning overstoryand understorycomtxments, J. Geophys.Res.. 102, 28,915-28,927, 1997.
299 pp., D. Reidel, We addedthe parameternt in (A1) to accounttbr seasonal Brutsaert,W., Evaporationinto the Atmosphere, Nora'ell, Mass., 1984. variationsin g•..The valuestbr T,,and T• in (A2b) were fixed
Thc parmneters in (A2a)-(A2c) were fit for eachsiteusing measuredR, D, T•,, and g, where g.•was derived from (5)
de Brain, H.A.R., A model for the Priestley-Taylorparametera, J. Clim.Appl.Meteorol.,22• 572-580, 1983. Fit•iarrald. D.R., K.E. Moore, R.K. Sakai, and J. M. Freedman, Assessingthe impactof cloud cover on carbonuptakein the northernborealforest,Trans.AGU, 76{17), SpringMeet. S125,
basedon AE*,theenergyclosure adjusted valuetbr A/z •
Goulden, M.L., B.C. l)aube, S.-M. t:an, D.J. Sutton,A. Bazz•, J.W.
at 0øand40 øC,respectively. Thevalue.f(/)•)in (A2b)wasset to zero when Ta< T• or 7; > T•.
1995.
(section2.2). The parameterfittingwasdone(usingMatLab,) Munger, and S.C. Wot•y, Physiologicalresponsesof a black in two steps.First, g, b•, bD, and bT in (A2a)-(A2c) were sprucelorest to weather,J. Geophys.Res., 102, 28,987-28,996, 1997. estimated for each site based on all data from May to September1994, with m set to 1.0. The regressionexcluded Jarvis, P. G., J. M. Massheder, S. E. Hale, J. B. Moncrieff, M. Raytnent,and S. I•. Scott, Seasonalvariationof carbondioxide, g• data below the 1stand abovethe 99th percentlies.Second, watervapor,andenergyexchanges of a borealblackspruceforest, the seasonalvariation in m was estimated daily using a J. Geophys.Res..102, 28,953-28,966, 1997. •novingwindowof 10 daysin lengthby linearregressionof g• Jarvis,P.J,The interpretation of the variations in leaf waterpotential and stomatalconductance lbund in canopiesin the field, Philos. versus g•f(ROf(T,•)./rD) The regression line was tbrced Trans. R. Soc. London, Set'. B. 273, 593-610, 1976. throughthe origin. Usingf(Ta) in the regressiongave only I,afleur, P.M., J. H. McCaughey,D.W. Joiner,P.A. Bartlett,and •ninimal benefit, as also reportedby Wu et al. [20001 and D.E. Jelinski, Seasonaltrends in energy, water, and carbon Massmanand Kaufmann[1991)],butf(Ta) wasretained. dioxidefluxesat a northernborealwetland,J. Geophys.Res.,102,
Estimates for missing valuesof H* andAE*werethen
29,(X)9-29,02(), 1997.
calculatedusing (3) and the Penman-Monteithcombination Lohammer, T., S. Larsson, S. Linder, and Falk, O, FASTsinrelationmodelsof gaseous exchangein Scotspine,Ecol. Bull. equationlMonteith,19811: 32, 505-523, 1980.
AE* =.s(Rn ,.Qs) +gaPCp D .
s +2'(1+ ga/gs)
I... l:quxsatnplingerrorsfrom aircraftand towers,J. Atmos. (A3)Mahrt, Oceanic Technol. 15. 416-429, 1998.
Note thatthe estimatesof g•,(equationA l) that were usedto
Margolis,H.A. andM.G. Ryan,A physiological basisfor biosphereatmosphereinteractionsin the boreal Ibrest: an overview, Tree Phvsiol., 17, 491-499, 1997.
estimate missing values of AE*in (A3) wereenergy closure Massman,W.J. and M.R adjustedvalues, becausethe parametersin (A1) and (A2a)(A2c) had been fit usingvaluesof g• that were calculated
fromclosure-adjusted •neasure•nents (2E*,sections 2.2 and 2.3).
Kaufinann,Stomatalresponseto certain envimmnentalI•ctors: A comparison of modelsfor subalpine treesin the RockyMountains,Agric. Pbr. Meteorol.,54. 155-167, 1991.
McCaughey,J.H., P.M. I,afleur, D.W. Joiner,P.A. Bartlett, A.M. Costello,D.E. Jelinski,andM.G. Ryan, Magninidesandseasonal patterns of energy,water,andcarbonexchanges at a borealyoung
Acknowledgmen[• We acknowledge the BOREASinvestigators jack pineforestin theBOREASnorthernstudyarea.,J. Geophys Res., 102, 28997-29{)07, 1997. and their colleagueswho operatedthe tower flux sitesand collected the flux data {Table 1). We also acknowledgethe efforts of the McNaughton,K.G., Evaw,ration and advectionIt: evaporation downwind of a boundaryseparatingregionshaving different personnel fromtheBOREASInlbrmationSystem.The suggestions of
33,550
BARR ET AL.: NORTHERN AND SOUTHERN STUDY AREA FLUXES IN 1994
surfaceresistances andavailableenergies.Q. J. R. Meteorol.Soc.,
Twine, T.E., W.P. Kustas, J.M. Norman, D.R. Cook, P.R. Houser,
10, 193-202, 1976.
T.P. Meyers, J.H. Preuger, P.J. Starks, and M.L. Wesely, Correctingeddy-covariance flux underestimates overa grassland, Agric. For. Meteorol.,103, 279-300,2000. Verma, S.B., Aerodynamicresistances to transfersof heat,massand momentum,Estimationof Areal Evapotranspiration, IAHS Publ.,
McNaughton, K.G. andT.W. Spriggs,An evaluation of thePriestley and Taylor equation,Estimationof Areal Evapotranspiration, IAHS Publ., 177, 89-104, 1989.
Monteith, J.L., Evaporationand surfacetemperature,Q. J. R. Meteorol. Soc., 107, 1-27, 1981.
177, 13-20, 1989.
Monteith,J.L., Accommodation betweentranspiring vegetation and Wu, A., T.A. Black,D.L. Verseghy,P.D. Blanken,M.D. Novak,W. theconvective boundary layer,J. Hydrol.,166,251-263,1995. Chen, and P.C. Yang, A comparisonof parameterizations of NewcomerJ., et aL (Eds,),CollectedData of TheBorealEcoastemcanopyconductance of aspenand Douglas-firforestsby CLASS, Atmos. Ocean, 38, 81-112, 2000. Atmosphere Study,[CD-ROM], NASA, Greenbelt,Md., 2000. Ohmura,A., Objectivecriteriafor rejectingdatafor Bowenratioflux calculations, J. AppLMeteorol.,21, 595-598, 1982. A. G. BarrandC. D. Smith,Meteorological Serviceof Canada,11 Priestley,C.H.B. andR.J. Taylor,On the assessment of surfaceheat Innovation Blvd., Saskatoon, Saskatchewan,Canada S7N 3H5. flux andevaporation usinglarge-scaleparameters, Mon. Weather (
[email protected];
[email protected].) Rev., 100, 81-92, 1972. A. K. Betts,Atmospheric Research,RR 3 Box 3125, 58 Hendee Sellers,P.J., et al., The Boreal Ecosystem-Atmosphere Study Lane, Pittsford,VT 05763. (
[email protected].) (BOREAS):an overviewand early resultsfrom the 1994 field T. A. Black, Facultyof AgriculturalSciences,Universityof year, Bull Am. Meteorol. Soc., 77, 1549-1577, 1995. Sellers,P.J., et al., BOREAS in 1997, Experimentoverview, British Columbia, 139-2357 Main Mall, Vancouver, British scientificresults,and futuredirections,J. Geophys.Res., 102. Columbia,CanadaV6T 1Z4.(
[email protected].) J. H. McCaughey, Department of Geography, Queen'sUniversity, 28,731-28,769, 1997. Ontario,CanadaK7L 3N6. (
[email protected]. ca.) Suyker, A.E., S.B. Verma, and T.J. Arkebauer,Season-long Kingston, measurementof carbon dioxide exchangein a boreal fen, J. Geophys.Res.,102, 29,021-29,028,1997.
Thom,A., Momentum, massandheatexchange of vegetation, Q. J. R. Meteorol. Soc., 98, 124-134, 1972.
(ReceivedSeptember 22, 2000;revisedJanuary 9, 2001; acceptedJanuary26, 2001.)
