Nov 27, 1999 - which includes the old jack pine (OJP) and old black spruce. (OBS) tower ...... Dobosy, R. J., T. L. Crawford, J. I. MacPherson, R. L.. Desjardins ...
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 104, NO. D22, PAGES 27,755-27,769,NOVEMBER 27, 1999
Comparison of the spatial and temporal distribution of fluxes of sensibleheat, latent heat and COz from grid flights in BOREAS
1994 and 1996
SegunO. Ogunjemiyoand PeterH. Schuepp Departmentof Natural ResourceSciences,McGill University, Ste-Anne-de-Bellevue, Quebec,and Centrefor ClimateandGlobalChangeResearch,McGill University,Montreal,Canada Ian J. MacPherson
Institutefor Aerospace Research,NationalResearch Councilof Canada,Ottawa
Ray L. Desjardins ResearchBranch,AgricultureandAgri-FoodCanada,Ottawa
Abstract. Analysisof airborneeddy correlationflux measurements of heat (H), moisture (LE) and CO2(C) over two 16 km x 16 km heterogeneous grid sitesin BOREAS 1994 (IFC-2) and 1996 are comparedin order to examinepersistenceand variability in the distributionsof surfacecharacteristics and fluxesbetweenthe two years. The data used were obtainedin grid patternsflown at 30 rn abovegroundlevel, undergenerallyclear sky andthermallyunstableconditions.Maps of fluxesand surfacecharacteristics were constructed by block averagingover 2 km windowsalongthe flight lines, analyzedfor similarities,and usedto quantifyspatialvariability of the fluxes. Sensitivityanalysis suggested minor effectsof boundarylayer variabilityandwindowsizeon the main featuresof the source/sink distributions. Incidentradiationwasmorehighlycorrelated with grid-averagedvaluesof C thanwith H andLE. The dominantrole of surface inhomogeneity,as opposedto local variationsin solarenergyinput, on spatialvariation of flux distributionswas confirmed,and mesoscalemotionwas foundnegligible, probablybecauseof the smallsizesof homogeneous subareaswith sufficientsurface contrastto inducethermallygeneratedmotion.CO2flux andgreenness indexwere highly correlated,but correlationwas site- and time-specific.The previouslyobservedlow correlationbetweensensibleheatflux and surfaceminusair temperaturedifference (T•Ta),primarilyover old blackspruce,wasconfirmed.The highBowenratio over the forestcontributedto the growthand developmentof the observeddeepboundarylayers over the sites,but no clear correlationemergedbetweenboundarylayer depthand observed near-surface
1.
fluxes.
the surfacecover.This doesnot invalidate thepreviously demonstrated potentialof grid flightsto maptemporaland
Introduction
Modelsof biosphere-atmosphere interactionshave been spatialvariationsin surfacefluxesand to relatethe observed developed primarilyforhomogeneous surfaces. Extrapolation flux estimates to surfacecharacteristics [Desjardins et al., of suchmodelsto heterogeneous surfaces requiresvalidation 1992; Schuepp et al., 1992; Mitic et al., 1995], in principle, by datathatintegratemeasurements overareastypicalof the butmakesitsapplication to theboreallandscape muchmore land surfaceand over an adequateperiod of time. Our of a challenge. analysisof the data obtainedby the Twin Otter aircraft
The analysis presented in thispaperusesTwin Otterdata [Ogunjemiyo et al., 1997] duringthe Boreal Ecosystem- fromBOREAS1994second intensive fieldcampaign (IFC-2) Atmosphere Study(BOREAS)1994suggests thattheprospect and the 1996 IFC to comparethe spatialdistributions of of usingboundarylayer modelsto estimatesurfacefluxes fluxesof sensible heat(H), latentheat(LE), andCO2 (C) at dependson the scalarunder consideration and the natureof thetwogridsitesbetween thesetwoobservation periods. This involvesidentification of areasat thesiteswheresignificant Copyright1999by theAmericanGeophysical Union. changes in fluxpatterns occurred andtheirlinkstopatterns of surface characteristics. Ourpaperalsoaddresses thestatistical Papernumber1999JD900052 stability of fluxmapsderivedfromoursampling andanalysis
0148-0227/99/1999JD900052509.00
procedure. 27,755
27,756
OGUNJEMIYO ET AL. ßAIRBORNE FLUX MAPPING IN 1994 AND 1996
years,andall flightsat eachsitewereflownwithin10days, exceptthe1996flightsin theSSA,whichwerespread over 20 days.Theairborne datawerecollected by theCanadian
2. Data Acquisitionand Processing 2.1.
Sites
The two grid sites,eachwith a dimensionof 16 km x 16 km, are located in the northern (NSA) and southern(SSA) studyareasof BOREAS, respectively[Ogunjemiyoet al., 1997]. The sites are heterogeneous,with different size patchesof surfacecoverandplant speciesat differentstages of growth.The land classificationimageryand the percentagesof the surfacecover typesare shownby Ogunjemiyoet al.[ 1997]. The NSA grid site is boundedby latitude55.80øN and 55.94øNand longitude98.40øW and 98.65øW, and the SSA grid by latitude53.78øN and 53.92øN and longitude 104.56øWand104.80øW.The northernhalf of theNSA grid, which includesthe old jack pine (OJP) andold black spruce (OBS) towersites,is coveredby wet anddry conifersandthe southernhalf by a former burn area in various stagesof regeneration,with onecentrallylocatedunburned"pocket". The SSA grid, which includesthe OJP, youngjack pine (YJP), and fen sites,is dominatedby wet conifermixed with fen, belowthe NW-SE diagonal,with drier conifer, logged, and regeneratingareasaboveit.
TwinOtter(TO) atmospheric research aircraft[MacPherson,
1990,1996] in a gridpattern,flownat approximately 30 m abovegroundlevel(agl),at a meanair speedof 60 m s4. Eachgrid flightconsisted of nineparallelstraightlines, spaced 2 kmapart,witheachlinesampled twicein opposite directions, in a sequence thatassured thatall gridlineswere sampled atthesamemeantime.Flighttrajectories (east-west ornorth-south) werechosen forclosest approach tocrosswind conditions.
Data were digitizedat 16 Hz. Observeddatausedin this
study include airtemperature (Ta),surface temperature (Ts) , incidentsolarradiation,reflectedred (R) andnear-infrared
(NIR)radiation, andmixingratiosof CO2andH20.Parametersderivedfrommeasured variablesincludenetradiation, Rn, greenness index, GI (from the Simpleratio SR = NIR/R), difference betweensurfaceand air temperature (AT= Ts-Ta)andpotential temperature (0). Moreinformation abouttheinstrumentation is givenby MacPherson [1996]. Thegeneral weather conditions duringtheflightsaresummarizedin Tables1 and2. All flightswere flownaround
solarnoon.Except forJulianday211in 1996,whenthesky wasovercast, theflightperiods werecharacterized bymainly The 1994 (IFC-2) data set consistedof eightgrid flights, sunny conditions. ThewindattheSSAwasgenerally stronger
2.2. Samplingand Weather Conditions
four at eachsite, flown duringJuly andAugust,andthe 1996 setconsisted of nine grid flights, four in the SSA and five in the NSA, alsoflown duringJuly and August.As seenfrom Table 1, the SSA flights precededthe NSA flights in both
in 1994thanin I996, withaveragevaluesof 4.4 and3.5 m
s'1 respectively. Thereverseis thecasein theNSAwhere averagewindspeedwas3.5 m s-• in 1994and4.1 m s4 in i996. The averageair temperatures in 1994 and in 1996
Table 1. Twin Otter Grid Flight Summary Flight
Time, Site
General Weather Conditions
1704-1859
SSA
clear, smoke from fire, north winds
N-S
1700-1854
SSA
mostlyclear, few cumulus,SW winds
26
E-W
1653-1844
SSA
clear, north winds
207
30
N-S
1705-1856
SSA
mostlyclear, few cumulus,SW-W winds
July 28
209
33
N-S
1617-1810
NSA
clear, NE winds
1994
Aug. 01
213
35
N-S
1547-1752
NSA
some cirrus, south winds
1994
Aug. 04
216
37
N-S
1546-1749
NSA
clear with some smoke, NW winds
1994
Aug. 08
220
39
N-S
1526-1748
NSA
few stratuscumulus, west winds
1996
July 09
191
58
E-W
1705-1906 SSA
1996
July 20
202
68
E-W
1605-1810
SSA
mostlyclear, few cumulus,SE winds
1996
July 27
209
71
N-S
1722-1908
SSA
overcast skies, SE winds
1996
July 29
211
72
E-W
1631-1818
SSA
clear,smallcumulusacrossgrid, SSE winds
1996
July 31
213
76
E-W
1456 1645
NSA
generallyclear, southwinds
1996
Aug. 02
215
79
N-S
1447-1638
NSA
clear, SW winds
1996
Aug. 03
216
80
N-S
1615-1801
NSA
clear with some smoke, east winds
1996
Aug. 05
218
81
E-W
1615-1810
NSA
few stratuscumulus, south winds
1996 Au•. 08
221
83
E-W
1450-1640 NSA
Year
Date
Day
Flight
Direction
1994
July 20
201
21
E-W
1994
July 21
202
22
1994
July 24
205
! 994
July 26
1994
GMT
generallyclear and sunny,SSW winds
clear, north winds
OGUNJEMIYO
ET AL. : AIRBORNE
FLUX MAPPING
IN 1994 AND 1996
27,757
Table 2a. Northernstudyareagrid averageparameters.
1994
1996
Julian
Wind,
Ta
Ts*
R,,
-L,
Flight
Day
m s-1
øC
øC
Gi*
W m'2
m
33
209
3.4
21.7
23.9
2.73
549.3
64.5
35
213
1.6
22.1
22.5
2.74
465.2
22.4
37
216
4.0
18.5
19.5
2.75
496.5
125.6
39
220
3.1
16.1
16.7
2.74
482.8
68.7
76
213
4.0
23.8
26.5
2.49
474.7
129.7
79
215
5.0
24.2
25.2
2.49
539.2
325.6
80
216
3.4
25.5
29.8
2.49
585.2
72.8
81
218
4.1
22.6
26.1
2.39
486.5
116.4
83
221
4.1
9.6
15.3
2.19
327.4
134.3
* Shiftsin instrument sensitivitymighthavecausedminorchangesin absolute values between,but not in relativedistributions within, the two years.
were 19.7øCand 21.1øC,respectively,at the NSA siteand 22.5øC and 18.7øC, respectively,at the SSA. Thermal stratification,as definedby z/L, where L is the Obukhov length,wasunstable(negativez/L) in all cases.Valuesof L are given in Tables2a and 2b.
were estimatedalongeachof the flight lines and averaged over all the flight lines to obtainthe grid averagefluxes. The mappingprocedureadoptedfor creatingtheflux maps wasthe sameas thatusedby Ogunjerniyo et al. [1997]. Flux estimateswere averaged over 2 km segments,and the estimatesfor the two repeatedpassesaveragedinto 2 km 2.3. Data Processing windows,with 1 km overlapbetweenadjoiningwindows. Vertical turbulent fluxes of sensibleheat, latent heat, and This generates a matrix of 135 datapointsper grid, with 4CO2attheflightlevelwerecomputed fromthetime-averagedkm averagesamplingper data point. Using cubic spline covariance(w'a') betweenfluctuations of verticalwind (w') interpolation,the griddeddata were smoothedand usedas andthe scalarof interest(a'). Potentialtemperature (0) was inputto GIS-basedIDRISI softwareto producethe flux maps usedto compute H, andtheuseof mixingratiosfor H20 (q) of the grid sites. The same procedureswere adoptedto (surfacetemperature and CO2 (c) obviatedthe need for densitycorrection producemapsof surfacecharacteristics indexGI). [MacPherson,1990].Turbulentfluctuations werecomputed excessover air temperature,AT, and greenness as the difference between instantaneous and mean scalar Localvariationsin surfaceemissivitywere notaccounted for.
quantities, with meansdetermined by applyingdetrending Given the fact that the PRT-5 sensor has internal calibration techniques developed by Ogunjemiyo et al. [1997]. Fluxes assuminggreen vegetationcanopieswith emissivitiesap-
Table 2b. SouthernStudyArea Grid AverageParameters Year
1994
1996
Julian
Wind,
Ta
Ts*
R,,
-L,
Flight
Day
m s-1
øC
øC
Gi*
Wm-2
m
21
201
5.7
21.5
19.3
2.71
525.9
414.2
22
202
5.1
26.5
24.7
2.70
622.9
204.5
26
205
4.9
19.2
20.1
2.84
611.4
146.3
30
207
1.8
22.9
22.7
2.77
553.8
128.4
58
191
5.0
19.6
25.8
2.3
691.5
151.5
68
202
2.9
18.4
22.9
2.28
601.1
41.6
71
209
4.2
17.0
19.6
2.20
291.2
412.2
72
211
1.9
19.9
23.4
2.26
573.9
14.6
27,758
OGUNJEMIYO
ET AL. ßAIRBORNE FLUX MAPPING IN 1994 AND 1996
2OO
ß
a
15o
E
100 50
0 185
•
I
,
190
I
•
195
I
,
200
I
,
205
I
I
210
t
215
I
220
225
Julian Day 30O
25O
•' 200
..• 150
100
,
50
19o
185
19•'
200
205
I
I
210
215
,
I
22O
225
Julian day -0.1
-0.2
• -0.3 E E
0 -0.4
-0.5
-0.6 185
,
I 190
•
I 195
•
I 200
,
I 205
•
I 210
,
I 215
,
I 220
, 225
Julian day
Figure1. Plotsof thegrid-averaged fluxes of (a)sensible heat,(b)latent heatand(c)CO2.
proaching unity(>0.98), the associated potential errorin Thesignificance of anygivenfactor,suchasradiation, in emissivity of 1 to2% isnotlikelytobesignificant. Also,no thecomplex relationships thataffectflux magnitudes, can atmospheric corrections weremadeonthesurface tempera-onlybededuced indirectly bycomparing thevariability inits turemeasurements, sincePerryandMoran[ 1994]foundsuch estimates tothatinthefluxestimates. Weusedthisapproach
corrections to be generallylessthan0.5ø for aircraftdata onthebasis ofthenormalized variances inthefluxmaps, Vr, acquired fortheir(higher) flightlevelsat 100m agl. andin theradiation maps,Vr, thatis,
OGUNJEMIYO
ET AL. : AIRBORNE FLUX MAPPING IN 1994 AND 1996
1./•i=•v _•]2
27,759
Table 3.Boundary Layer Parameters
VfNM j='-j E•[F(i d) i=l
zi
(1)
Azi,
W.
t.,
Flight rn rn rn s:• s
Vr -NMj=• ••[R(id)-•]2
o.,
q.,
K
g kg4
21 7461801.26360.06 0.08 22 1180 170 1.7 688
/=1
777
30
150
O.O5
1.6
497
0.10
0.05
1045 710
1.7
633
0.08
0.05
33
928
1.6
577
0.08
0.04
transformed into asetofvalues Z•jbysubtracting themean
35
800 820 1.4
579
0.07
0.03
andnormalizingthestandard deviationof thedifferences. The similaritybetweenthetransformed mapsisthencomputed on
37
1116
674
0.07
0.03
thebasisof similarity in thesignof Zijpairsas
39
1227 1034 1.8
657
0.08
0.03
where F(i,j) and R(i,j) are the grid data matrix in the two maps,withM rowsandN columns,andtheoverbarindicates averageof all the map data. To determinesimilaritybetween1994and 1996mapsof a givenscalaror flux, the datapointson the mapsare first
26
O.O8
644
937
1.6
58 121465
1.9 576
0.09
0.04
68
1.8
0.08
0.04
number ofZijpairs with negative, positive, ormixed signs. 71 714 252 1.0 750
O.O4
O.O5
Values ofCsvaryfromzero(nosimilarity) to 1 (absolute
72
785 614 1.5
546
0.08
0.05
Cs= rnm+n +n +p
(2)
1191
973
666
whereCs is the similaritycoefficient;m, n, andp are the similarity).
Data from serialupperair soundings [Barr and Betts,
76
1227
851
1.6
790
0.06
0.04
1994, 1997]releasedon eachof the flight dayswereusedfor boundarylayeranalysis.Because the releasetime wasabout 5-10 min beforeor after eachflight, with a 2 hour spacing,
79
678
688
1.2
591
0.06
0.05
twosuccessive soundings, coinciding approximately withstart
80
977
725
1.5
616
0.08
0.04
theboundary layer conditions during each flight. Estimated 81
1660
684 1.8 907
0.06
0.04
parameters include themixedlayerdepthzi, takenasthe
872
248
0.07
0.03
andendtimeof thegrid flightcanbe considered to document
depthover which the potentialtemperatureis essentially constant,averageboundarylayer moistureand changesin boundarylayer heightAz, moisturemixing ratio Aq, and potentialtemperature A0, whichoccurredduringthe flight period.We alsoestimatedthe convective velocityscalew,, convection timescalet,, temperature scale0,, andhumidity scaleq, in themixedlayeraccordingto Stull[1988]as
z, w,=[g_Z•(w/O/)]m,t,=•, 0,-(wO%,q,0
W,
W,
W,
(3)
83
1.4
607
NSA site.The observedtemporalvariabilityin the fluxeswas mainly due to variation in incidentsolar radiation and to a probablylesserdegreeto changesin flux footprintsand boundarylayer structure. The net radiationg nwas generallyhigherat the SSA than at the NSA, with lessthan6 % differencebetweenthe years at eachsite. It is interestingto observethatthehighestvalues of C are associatedwith lowest solar insolation,resulting from haze or cloud cover, and the lowest values of C associ-
ated with highestinsolationfrom clear skies. This is tentatively associatedwith physiologicalresponseto high vapor pressuredeficitsof coniferoustrees. Insofar as correlations 3.1. Grid-Averaged Fluxes couldbe established betweenRnandfluxesin suchsmalldata Figure 1 illustratesthe temporalvariationsof the grid sets, they were higher for C than for H and LE, and the with results averaged fluxesat thesitesfor the1994and1996observation observedtrendsbetweenC andg nare consistent periods.Each value represents samplingover a 288 km from otherstudies[e.g., Hollinger et al., 1994; Baldocchiet uptakeof CO2enhanced distance,which is expectedto providearea-averaged flux al., 1997],whichshowedecosystem estimateswith a relative error of < 5% [Lenschowand undercloudysky when the incomingsunlightis isotropic. Examinationof theboundarylayerprofilesof mixingratio Stankov, 1986; Mahrt, 1998]. Temporal variability was (O) at thestartof thegrid flightsshowed greaterattheSSAthanat theNSA; it depended onthenature (q) andtemperature of the scalarandreflectssite-dependent weathereffects.The thatthetimingof theflightscoincidedwith the growingphase composite, grid-averaged valueof H was the samein both of the boundarylayer when the residuallayer decaysas a yearsat theSSAbut24% lowerin 1996thanin 1994at the resultof turbulentmixingtriggeredby surfacewarming.The q with NSA. AveragedLE wasthe samein bothyearsat the NSA shapeof theprofiles,nearlyuniformO, anddecreasing signatures of but 19% lower in 1996 at the SSA. AveragedC was 18% heightwithinthe mixedlayer are characteristic higherin 1996thanin 1994at the SSAbut4 % lowerat the convectivemixed layer turbulence.The averagedvaluesof
3.
Results and Discussion
27,760
OGUNJEMIYO
ET AL. : AIRBORNE Ts -Ta
a
FLUX MAPPING
IN 1994 AND 1996
c
1994
GI
1994
'• :']:::::*.':•::•i::•ii:::;½.•:.,-?.•ii•i::•i';",.':" ;;•.•```•.``.•!:.*•:i:•:!{•;!•i.?;•:!:•i:•::!•:;.`:•!•iiii•i:.i:•iii•i•ii•. .'.,.•:' •:,+.•:i•::i:•.•::;•,,-.•:::½*.:,i:;•;:•., ..... -.;:.•.:*.iii.::! ,.":;::**::•'ii•i•iS**i':':*'*'"'.**'"" ........ .:.:::.:,,:., iii,:!*: ....
•-0.94
-0.03 •
•"• "?' *:':0.88 - 1.78
0.04 -0.87
'•.......... 1.79 - 2.88
-'---• 2.89 - 3.59 •
3.60 - 4.50 Ts -Ta
• 2.00 - 2.20 : -'-'........2.41-2.60
"• ........... 2.61 - 2.80
"::'':.'-'"':'.';• 2.81 - 3.00
'
•
2.21 - 2.40 •3.01 - 3.20
1996
GI
1996
'*:*•i'""•'•*"•"' '••••di "•'•'"'•'" ' ::: ''"*;-'•!i':'"" '•:•..g '"'"•'"' '••*•... '-••;:•!i'":'*'x'"'" .................... '•:''" ' ' "":.•7• "P;;:::•-i•*i;* .....
?•""'""'*"•-.•.'•.•.*a.•xS;."•.i*'".-aa•]•: -':'" '"'-s.':•.'-'....'.•Z;;,.:..-.,-.•ssiia•ii ...... '*•"ii:*'*'•;*c":: ....... :":•;'"' •
1.60 -2.40
•
•
3.07 - 3.70
-•-'•3.71 '",-'-"-'----*.. - 4.00
•
•
•'-'• ]':'":"-""-'":"""•*'""*:'•:":" 2.47 - 2.58
...........•--:-a 4.01 - 4.96
2.41- 3.06
4.97 - 5.64
•2.00
- 2.11
2.24 - 2.35
•
'"'• '•'*'"'
2.12 - 2.23
2.36 2.46
2.59 -2.70
Figure2. (a-d)Mapsof surface minus airtemperature Ts-Ta(øC)andgreenness index(GI)atthenorthern studyarea(NSA) grid in 1994and 1996.
w,, t,, 0,, andq, (Table3) aretypicalfor a deepmixedlayer with vigoroussurfaceheating[Stull, 1988; Bettsand Ball, 1994].The averagegrowthof theboundarylayerduringthe flight was greaterin eachyear at the NSA than at the SSA, just asthe averagevaluesof the mixedlayerheightz• were higherat the NSA (1056 m in 1994and 1082m in 1996)than at the SSA (902 m in 1994 and 976 m in 1996). The structure andgrowthof theboundarylayer is knownto be relatedto H andLE; a changein oneof the variablesmay be explained partly by a changein the other. However, conclusiveinferencescouldnot be drawnfrom the linear regression of z• againstH and LE in our data set. A differencein atmospheric stabilitybetweenflight days couldalso causevariationin the grid averagefluxes. As a majorcomponent of the flux footprintfunction,the stability parameter-z/L definesthe zone within which surfacecover significantlycontributesto the measuredfluxes. Unlike in 1994,the averagedvalueof L wasthe samefor bothsitesin 1996, with a wider rangeof valuesat the SSA than at the NSA (Table2). This, as well as the fact that the footprint analysisof Kaharabataet al. [1997] showedthatthe various components of thetwogridsurfaces weresampledadequately
evenat a more constrained footprintat L = -100 m, makes
stability related variations inthefootprint anunlikelycontributor to the flux variability. 3.2. Maps of Surface Characteristics
The mapsof excessof surfacetemperature over air temperature (AT)andgreenness (GI) areshown in Figures 2 and 3. Sincea calibration problemmighthaveaffected absolute values of AT andGI andsincetheemphasis is on relative spatial variability, different scaling wasusedforthese maps.Thespatial distribution of AT reflects thepatternof absorption andpartitioning of energybetween thedifferent components of the surfacecover, and the distributionof GI
documents theirphysiological status. The NSA AT maps(Figures2a and 2b) exhibitthe contrastin surfacetemperature betweenthenorthernandthe southernhalves of the site. The areas with low AT were
dominated bytallandclosed-canopy conifers, whiletheareas withhighAT wereprimarilycovered by shrubs, shortand opencanopytrees,withan oftendry underlying surfacethat
is typicalof a regenerating burnarea.Thecorrespondence
OGUNJEMIYOET AL. ßAIRBORNEFLUX MAPPING IN 1994AND 1996 a
....... •.........
Ts -Ta 1994
:..'.,
....:..• •.:::•,•?/..•:;•:•. ......
c
GI
27,761
1994
.:.................... ............. ......... ................ ............ i
::":•¾"'"•••:•• '••'"" ..2.4•::::i::i•?" "•' '•:•:.-'"-:i•' ' ?'::'%::':?':" ::•;•i:;:::".•.::• .... :-•:i ??:?"•'"'"'"'"'"'•'••••••• •.•,.,•.....:•.. :.: ..
•::•4
•-• ,*.
,• :•
ß, . .....
•
-2.06 - -0.61
Ts -Ta ......
1996
...,,• • .•:•,'.::•. -.•• •'•-:•:•i•:•"•::
•. :::•. '•..
..
,..... .....
....•.•. -,,,, .•?;;;•:•:.•-% .:;.;'
1.99- 2.26
•
••
2.56 - 2.63
!'"-•:•;J 2.64 - 3.10
•3.71 - 5.15
b
• . •;:•:;.:, • ........... .... •:•,
•
•'e•, ,•;..';•o:• 0.83 - 2.26 •
..:'::'•= •' '
......... •'•:-:.::•
I:• .........:'.':':-;'• 3.11 - 3.38
•
d
:
2..27 - 2.55 •3.39 - 3.66
GI 1996
...• ß :... '.Z;' ': ........ .•.•": ""'L•; •.:.:' ........... ß .............. '"'"':"'•'":'"'"':"' :'::•':::' ........... ...-•..:'" :;•:•:':':""
;:::•;...:: ..... ,'•...'-'
•
•+• '• '•...::;•.:..:..:..:•½:..:.:.. .......
.....
ß :;.i ;:.'. :...
:...
1.99- 3.34
•
3.35- 4.69
4.70 - 6.04
•;•.••.•
6.05 - 7.39
7.40 - 8.74
•
•
1.99-2.20 •
2..21- 2.41
•
2.42 - 2.62
2.63 - 2.83
•>•-•]
::::::::::::::::::::::::::::::::::::::::::::::::::::::::: 2.84 - 3.04 • •3.05 - 3.25
• 8.75 - 10.1
Figure3. (a-d)Mapsof T,-TaandGI atthesouthern study areaSSAgridin 1994and1996.
area between1994and1996patterns is very strong,withcoeffi- ViningandBlad,1992],withlowAT overtheforested bythefactthatoverpartially covered canopies the cientof similarityC, of 0.94 andcorrelation coefficient r of explained measured bytheverticalradiometer isa combination 0.84, in spiteof somelocaldifferences suchastherelatively radiance fromthetop of thecanopyandtheunderlying high values of AT over black spruce (around of emissions 55.9øN/98.45øW)and over old jack pine (top left-hand surface[Sunand Mahrt, 1995], whichwas coolerat the corner)in 1996. To what degreesuchdifferencesmay be forested area than at the disturbed areas. There is some for theexpected inverserelationship between AT associated withflightorientation (predominantly N-S in 1994 evidence thesharpdistinction between theforested and and E-W in 1996), leadingto potentialdifferencesin the andGI, though areasis absent fromtheGI maps. sampling of anisotropic surfacepatches, is notyet clear.A theregenerating of GI patterns between thetwo dependence of the spatialextentof mappedtemperature Thereishigherpersistence attheSSAsite( Figures 3cand3d;C,andr, 0.94and anomalies on theorientationof flightlinesis suggested by the years thanattheNSAsite(Figures 2cand2d); comparison of temperature mapsfromN-SandE-W linesin 0.90,respectively) C,andr, 0.84and0.64,respectively), partlyassociated with the 1994data[Ogunjemiyo et al., 1997]. in thenorthern partof theburnarea. At the SSA (Figures3a and 3b) the NW-SE diagonal therapidregeneration separates thesiteintoa warmareaabovetheline anda cool areabelowit. The high AT aroundthe middleof the siteto 3.3. Flux Maps the NE edge corresponds to the loggedand regenerating 3.3.1. Spatialdistributionof fluxes.The NSA andSSA areas,whilelow AT valuesareassociated withpredominantly mapsof C, H andLE are shownin Figures4a-4f and5a-5f, wet conifersinterspersedwith fen. The similarity and respectively.In the NSA grid, both 1996 and 1994maps low correlationcoefficients between1994and1996mapsare0.92 showhighvaluesof H overtheforestandcomparatively formerburnareas.Estimated Cs and 0.90, respectively.The viewing angleproblemswith valuesovertheregenerating respectto observedAT havebeenaddressed before[e.g., andr betweenthe 1994andthe 1996mapsof H are 0.84 and
27,762
OGUNJEMIYO ET AL. ßAIRBORNE FLUX MAPPING IN 1994AND 1996 a
c
e
1994
l
0.45
H
• ......
'•'•• 108 - 123 r'"'"'" •••'•:• 124 - 139 ' ........... ,---':'-• 140 - 155 , , •>155
LE •2Ol
174 - 200
226
'•...... 227 - 253 r'-'""-l> 253
f
1996
Figure 5. (a-f) Maps of C, H, and LE at the SSA in 1994 and 1996.
betweenthe ratiosandV r showsan exponentwo factorsby comparing thecoefficiem of variationof the fit relationship incoming radiation Vr against thecoefficients of variation of tial decaywith increasingVr. The influenceof radiationon greaterin 1996thanin the fluxesV/(for LE), Vh (for H), and Vc (for C). The thevariabilityof thefluxesappeared coefficients for the fluxes showed noticeable differences
1994, which is also corroboratedby statisticaltests (see
between gridflightsandlittleor nodifference in theaverage below). From the valuesof theseratios(mostof whichare was more valuesbetweenthe two years. At the NSA site, average _>5)it is apparentthat surfaceinhomogeneity thanradiation in inducing variabilityin thesurface values of V/,Vh,andVcare0.31,0.27,and0.30for 1994and significant 0.30, 0.27, and 0.32 for 1996. In the same order, the fluxes. 3.3.3. Direct mesoscalefluxes. Vertical tramport of coefficiemsare 0.26, 0.27, 0.36 for 1994and0.37, 0.28, and 0.36 for 1996at the SSA site.The reducedvalueof V/at the scalarsat scalesgreaterthan turbulenteddiesmay be an component of thetotalsurfaceflux. Tramportby SSAin 1994is mostlikely associated with extensive precipi- important motionis generallysmallat low observation levels tation acrossthe site at the start of IFC-2, which would have mesoscale reduceddifferencein evaporation dueto variationsin surface but canbe importantat the aircraftlevelof 30 m [Sunand types(seediscussion of Bowenratio).Thismeansthatexcept Mahrt, 1994].Furthermore,thepartitionbetweencoolforests areasat thegrid sitesandtheassociated for LE at theSSAsite,thebiophysical differencesat thesites andwarmdisturbed variations in surface radiationtemperature (10ø-15øC)andair betweenthe two years were insufficiemto significantly
(2.0ø-3.0øC) aresimilarto thoseresponsible for modifythesourceandsinkconfigurations of thescalars.By temperature motionsreportedby $egalet al. [1989],Mahrt contrast,incomingradiationwassomewhat moreuniformly themesoscale distributedin 1996 than in 1994, with averagevaluesof Vr et al. [1994], andDoran et al., [1995]. We examined theinfluenceof surface-induced mesoscale 0.05 in 1994 and 0.10 in 1996 at the NSA and 0.04 in 1994 thegridby estimating mesoscale fluxesfrom and 0.12 in 1996 at the SSA. The valuesof the ratiosV//V r, fluxesacross averages andcomposited overrepeated runsof a Vh/Vr, andVc/Vragainst Vr areplottedin Figure6. Thebest segment
27,764
OGUNJEMIYOET AL. ßAIRBORNEFLUX MAPPINGIN 1994AND 1996
SSA 94 H SSA 96 H NSA 94 H NSA 96 H
---
20 ",I
¸
I
•
15
lO
¸ 0.05
0.1
0.15
0.2
0.25
0.3
Vr
35
SSA 94CSSA 96CNSA 94CNS•6 C
30
25
20
> 15 10
0.05
0.1
0.15
0.2
0.25
0.3
Vr
SSA 94LE SSA 96LE NSA 94LE NSA•6 LE
20 '• 15
10
0.05
0.1
0.15
0.2
0.25
0.3
Vr
Figure6. Plotsof theratioof V/Vr against Vr.
that flightline. Theseline estimates,symbolizedhereas Fm•,are thermalproperties,especiallyat the NSA grid, suggest is not largeenough influencedby transienteventssuchaschangesin wind speed the spatialscaleof surfaceheterogeneity andcloudcoverandshowedsignificantvariationacrossflight to enhance mesoscale motion. 3.3.4 Stabilityof flux map estimates. The flux maps lines. Highestvaluesof Fm•were associated with moisture transfer and could be _• 20% of the turbulent fluxes. The presented aboveare from griddeddatacomposited overfour mesoscale flux averaged over the grid Fmg,is primarily or more flightsundersimilar weatherconditions,with each relatedto surfaceheterogeneity,althoughit may still include grid line sampledtwice per flight. We have examinedthe someinfluenceof nonstationarity. As expected,theobserved reliabilityof themaps,in thefaceof boundarylayervariabilvalues of Fmg weremuchsmaller thanthose of Fm• ' andforall ity, by analyzingthemagnitudeof variationin flux estimates thegridflights andfluxtypes,Fmg didnotexceed 10%of the obtainedfrom six or morerepeatedpassesof theTwin Otter turbulentflux andwasfar lesswhencomposited overan IFC. under uniform radiation conditions, over the OBS site at the The low observedvalues of the mesoscalefluxes at the site, NSA. Flux averagesover adjoining,nonoverlapping 2km producedfive data pointsfor eachof the 10 km evenin thepresenceof a well- developedgradientin surface segments
OGUNJEMIYO
ET AL. : AIRBORNE FLUX MAPPING IN 1994 AND 1996
C
LE
•
178
156
• 4.2 x
underdevelopment) mayhelpto elucidate theparameters that definesuchrelationships remainsto be seen. Acknowledgments. The financialandtechnical supportfor
thesestudies fromtheCanadian NaturalScience andEngineeringCouncil,Agricultureand Agri-FoodCanada,the
NationalResearchCounciland from the Atmospheric Service of Canada isgratefully acknowledged. 105Jdeg-1(forwater,quartz,andorganic matter),thiswould Environment account for approximately 20 W m-2,suggesting thatsuch
References effectsmightpartiallyaccountfor theobserveddiscrepancies. However, consideringthatthereis little evidencethat any of Avissar, R., andR. A. Pielke,Theimpartofplantstomatal control onmesoscale atmosphericcirculation. Agtqc. For. Meteorol. , 54, thefactorsmentionedaboveweremoresignificantin oneyear 353-372, 1991. thanthe other,we assumethaterrorsin the energybalance Baldocchi, D. D., C. A. Vogel.,andB. Hill, Seasonal variation will have little effect on our analysisof relative spatial of carbondioxideexchange rates aboveandbelowa boreal distributionsbetweenthe two years. jackpineforest,Agric.For. Meteorol.,83, 147-170,1997.
Barr,A.G., andA. K. Betts,Preliminary summary of BOREAS 4.
Conclusion
upper-air soundings. Tech Rep., 12, Atmos.Environ.Serv, Saskatoon,1994.
Spatialandtemporalvariationsin the surfacefluxesof heat Barr,A.G., andA. K. Betts,Radiosonde boundary layerbudgets abovea borealforest, J. Geophys. Res.,102, 29,205-29,213, (H), moisture(LE), andCO2(C) were examinedby compar1997. ing the 1994 and 1996 airbornedata obtainedat the two 16 Betts,A. K., andJ. H. Ball, Budgetanalysis of FIFE sondes, J. km x 16 km BOREAS grid sites.A sensitivitytest showed Geophys.Res., 99, 3655-3666, 1994. thatthedominantfeaturesof thecomposite flux mapsusedin Desjardins, R. L., P. H, Schuepp, J. I. MacPherson, andD. J. Buckley,Spatialandtemporalvariations of thefluxesof carbon our studyare relativelyinsensitive to boundarylayervariabildioxide andsensible heatovertheFIFEsite,J. Geophys. Res., ity. A comparisonof the coefficientof variationof the fluxes 97, 18,467-18,475, 1992. with those of available energy showedthat variability in Desjardins R.L., et al., Scalingupfluxmeasurements of theboreal surfacefluxesat the sitewas associated more with heterogeforestusingaircraft-tower combinations, J. Geophys. Res,102, 29,125-29,135, 1997. neity of the surfacecover than with variationsin available R. L. radiantenergy.Analysisof the mapsindicatedsomenotice- Dobosy,R. J., T. L. Crawford,J. I. MacPherson, Desjardins, R.D. Kelly, S. P. Oncley,andD. H. Lenschow, able changesin the biophysicalpropertiesof the surface Intercomparison amongfourflux aircraftat BOREASin 1994, covers over the 2 year period, but the changeswere not J. Geophys.Res, 102, 29,101-29,111, 1997. enoughto significantly modifythesourceandsinkconfigura- Doran,J. C., W.J.Shaw,andJ.M. Hube,Boundary layercharactionsof the scalarsthatwere examined.The mapsof surface teristics overareasof inhomogeneous surface fluxes,J. Appl. Meteorol., 34, 559-571, 1995. characteristics showedgreatersimilarityin spatialpatterns Hollinger, D.Y., F. M. Kelliher,J. N. Byers,J. E. Hunt,T. M. between1994 and 1996thanthe flux maps.SpatialdistribuMcSeveny, andP. L. Weir,Carbon dioxide exchange between an tions of C showed the highestvariation between the two undisturbed old-growth temperate forestandthe atmosphere, years, especiallyin the NSA, where a southwardshift in Ecology,75 134-150, 1994. greenness matcheda similarshiftin CO2 absorption. Kaharabata S. K., P. H. Schuepp, S. O. Ogunjemiyo, S. Shen,M. Y. Leclerc, R. L. Desjardins, andJ. I. MacPherson, Footprint Correlationbetween greennessindex, GI and C was considerations in BOREAS, J. Geophys. Res.,102, 29, 113significantin bothyears(r valuestypicallybetween0.6 and 29,125, 1997. 0.8) andhigherat the SSA thanNSA. The significantvalues Kelly,R. D., E. A. Smith,andJ. I. MacPherson, A comparison of of H in spiteof a low surfacetemperatureexcessover the surfacesensibleand latent heat fluxes from aircraft and surface forest, and the resultingdecouplingbetweenAT and H (r measurements in FIFE 1987, J. Geophys. Res., 97, 18,44518,453, 1992. values< 0.2, negativeat the NSA andpositiveat the SSA), turbulence withinand was alsoobservedin bothyears.The high Bowenratio over Lee, X. andT. A. Black, Atmospheric abovea douglas-fir stand,II, Eddyfluxesof sensible heatand the forestattributedto physiologicalcontrolof the stomatal watervapour. BoundaryLayer Meteorol., 64, 369-389,1993. openingof plants should favor the developmentof deep Lenschow, D. H., andB. B. Stankov,Lengthscales in theconvectiveatmospheric boundary layer,J. Atmos.Sci.,43, 1198-1209, boundarylayers over the forests. However, we could not 1986. establish,onthebasisof ourobservations, a clearrelationship MacPherson,J.I., Wind and flux calculationson the NAE Twin betweenboundarylayer heightsand flux characteristics. Otter,Rep.LTR-FR-109,Natl. Res.Counc.,Ottawa,1990. Our studyunderlinesthe importanceof the role of the MacPherson, J.I., NRCTwinOtterOperations inBOREAS, Rep. surfacemosaicin forest-atmosphere exchange.It ishopedthat LTR-FR-129,Natl. Res. Counc.,Ottawa, 1996. andR. L. Desjardins, Observations the observations reportedhere will providecluesand refer- Mahrt,L., J. I. MacPherson, of fluxesoverheterogeneous surfaces, Boundary Layer encepointsfor the developmentand validationof surfaceMeteorol., 67, 345-367, 1994. vegetation-atmosphere modelsoverthe BOREAS landscape, Mahrt,L. Flux Sampling errorfromaircraftandtowers.J. Atmos. includingthosebasedon remotesensingobservations. They Oceanic.Technol.15, 416-429, 1998. should convey to the potential user our estimatesof the Mitic, C. M., P. H. Schuepp,R. L. Desjardins, and J. I. MacPherson,Spatialdistribution and c oo c cu r r e n c e o f confidencelimits applicableto the givenmapswhenusedas surface-atmosphere energyandgasexchange processes overthe "test patterns"for high-resolutionmodeling, and caution againstoversimplification in the assumption of relationships codegrid site,Atmos.Environ.,29(21), 3169-3180,1995. Ogunjemiyo, O. S, P. H. Schuepp, J. I. MacPherson, andR.L. between radiometric surface features and energy or gas Desjardins.Analysis of flux mapsversus.surface characterisexchange,whichappearto be site-andtime-specific.To what ticsfromTwinOttergridflightsin BOREAS1994,J. Geophys. Res., 102, 29,135-29,147, 1997. degreetheuseof morerefinedsurfacedescriptions (currently
OGUNJEMIYO ET AL. : AIRBORNE FLUXMAPPING IN 1994AND1996
27,769
Perry, E. M., and M. S. Moran, An evaluationof atmospheric Verma,S. B., J. Kim, R. J. Clement, N.J. Shurpail, andD. P. correctionsof radiometricsurfacetemperaturefor a semiarid Billesbach, Tracegasandenergy fluxes:Micrometeorological rangeland watershed,WaterResour.Res., 30, 1261-1269,1994. perspectives, in SoilsandGlobalClimateChange, EditedbyJ. Schuepp,P.H., J. I. MacPherson,andR. L. Desjardins, AdjustKimbleand B.A Stewart,A.F. Lewis, New York, 1995. mentof footprintcorrectionfor airborne flux mapping over Vining, R.C., andB. L. Blad, Estimationof sensibleheatflux from FIFE site,J. Geophys.Res., 97, 18,455-18,466,1992. remotelysensed canopytemperatures, J. Geophys Res., 97, 18,951-18,954, 1992. Segal, M., R. Avissar, M. C. McCumber, and R. A. Pielke, Evaluationof vegetationeffectson the generationandmodification of mesoscale circulations,J. Atrnos.Sci. 45(16), 2268-2294, 1989.
Sellers, P., et al., The Boreal Ecosystem-Atmosphere Study (BOREAS): An overviewand early resultsfrom the 1994 field year, Bull. Am. Meteorol. Soc., 76 (9), 1549-1577, 1995. Stull,R. B., An Introduction to Boundary LayerMeteorology, 666 pp., KluwerAcad., Norwell, Mass., 1988. Sun, J., and L. Mahrt, Spatialdistributionof surfacefluxesestimatedfrom remotelysensed variables,J. Appl. Meteorol., 33, 1341-1353, 1994.
Sun, J., and L. Mahrt, Relationshipof surfaceheat flux to microscaletemperaturevariations: Applicationto BOREAS, BoundaryLayer Meteorol., 76, 291-301, 1995.
R.L Desjardins, Research Branch, Agriculture andAgri-Food
Canada,Ottawa,Ontario,Canada
I. J. MacPherson, Institute for Aerospace Research, National
Research Councilof Canada,Ottawa,Ontario,Canada
S. O. Ogunjemiyo andP. H. Schuepp, Department of Natural Resource Sciences, McGillUniversity (Macdonald Campus), 21,111
Lakeshore Road,Ste-Anne-De-Bellevue, Quebec, H9X 3V9Canada (segun@nrs. mcgill. ca; pschuepp@nrs. mcgill. ca) (Received August 2, 1998;revised December 7, 1998; accepted January21, 1999.)