JOURNAL
OF GEOPHYSICAL
RESEARCH,
VOL. 100, NO. Dll,
PAGES 23,125-23,131, NOVEMBER
20, 1995
Ozone fluxes over a patchy cultivated surface
L. Mahrt, • D. H. Lenschow, Jielun Sun,• and J. C. WeiP
National Center for Atmospheric Research,Boulder, Colorado
J. I. MacPherson
Flight ResearchLaboratory,National ResearchCouncil, Ottawa, Ontario, Canada
R. L. Desjardins
Land ResourcesResearchCentre, Agriculture Canada, Ottawa
Abstract. This studyexaminesthe spatialvariability of ozone fluxesover fiat
heterogeneousterrain consistingof a patchworkof irrigated and nonirrigatedsurfaces.
Fluxesof ozone and other quantitiesare computedfrom eight sequentialflight legsof the
Canadian
Twin
Otter
research aircraft
over the same track at 33 m above the surface for
each of 2 days.The fluxesare compositedover the eight runs to reduce the random flux
error. The fluxesof heat, moisture,and carbon dioxide are closelyrelated to spatial
variationsof surfacevegetation.However, the ozone flux is affectedby additional factors
includingreaction with NO releasedfrom point sources.This effect is illustrated here with
two examplesof irrigation pumpingstationsdriven by dieselengines.We concludethat
the ozonedepositionto the surfacecannotbe estimatedfrom the measuredozone flux
without
1.
correction
for the NO
sources.
is due to spatialvariationsof surfaceconditions,while most of
Introduction
the variance
The daytime boundary layer is typically characterizedby
downward
turbulent
flux of ozone. This downward
ozone flux
of the ozone flux is due to transient
behavior.
There are situations, however, where the ozone flux is
closelyrelated to the vegetativecover and highly correlated
with fluxesof heat, moisture,and carbondioxide.For example,
MacPherson[1992] found an approximaterelation between
ozone flux and the greennessindex when comparingaircraft
data collectedat homogeneoussitesover a 1-month observational period. Similarly,Lenschowet al. [1981] found a good
qualitative relationshipbetween ozone fluxes and vegetation
over a region of mixed rangelandand irrigated cropland. In
their casethe grasswas senescentand there were no apparent
significantsourcesof NO.
As an additional complication,the flux measuredabovethe
surface may be considerablydifferent from the surface flux.
These differencesmay result from a large entrainment rate,
horizontal advection,rapid time changes,or, in the case of
chemicallyreactive species,production or lossof the species
betweenthe surfaceand the measurementheight. Ozone can
react with NO on timescalesthat are short enoughto significantlymodify the vertical divergenceof the ozone flux.
Lenschow[1995] estimatesthe turbulent diffusiontimescale
to be of the order of a couple of minutesin the upper part of
a convectivesurfacelayer (lowest 30 m), and a few tens of
minutesup to an hour in the overlyingmixedlayer (1 or 2 km
depth). This meansthat the reactionof ozonewith NO, which
has a time constantof roughlya coupleof minutes,can signif•Now at Collegeof Oceanicand AtmosphericSciences,
Oregon
icantly alter the ozone flux measured,even within the surface
State University, Corvallis.
2Nowat Programin Atmospheric
andOceanicSciences,
University layer [Lenschow,1982; Fitzjarraldand Lenschow,1983; Lenof Colorado, Boulder.
schowand Delany, 1987;Kramm et al., 1991].
3Nowat Cooperative
Institutefor Researchin Environmental
SciA major sourceof NO is combustion.Therefore its emission
ences,University of Colorado, Boulder.
is inherently inhomogeneousand often episodic.When first
Copyright 1995 by the American GeophysicalUnion.
released,it reactsquicklywith ozone and, in large concentrations, can effectively remove the ozone. This is a transient
Paper number95JD02599.
0148-0227/95/95JD-02599505.00
effect. Eventually, as the air is diluted and the NO reaches
is partly associatedwith entrainment of ozone from the overlyinglayer [Neu et al., 1994] and ultimatelyleadsto deposition
of ozone at the Earth's surface.Surfacedepositionof ozone
onto vegetatedand bare surfacesdependson factorsdifferent
from thoseaffectingsurfacefluxesof heat, moisture,and carbon dioxide [Massmanet al., 1994]. A large fraction of the
ozone depositionis directly onto soil and leaf surfacesand is
not part of the stomatalexchange[Leuninget al., 1979; Wesely,
1983].
Ozone depositionalsodependson the soilwater contentand
atmospherichumidity.Sinceozoneis not very solublein water,
depositionof ozoneto wet soilsis lessthan for dry soils[Chamberlain, 1986] and becomesquite small over saturated soils
[Weselyet al., 1981].McLaughlinand Taylor [1981] found that
ozone depositionto plants can increaseby a factor of 2 or 3
when relative humidityincreasesfrom 35% to 75%. As a result
of thesefactors,the spatialvariationof ozonefluxesmay show
a more complexrelationshipto spatialvariationsof vegetation
cover and soil conditionsthan is shownby heat and moisture
fluxes[e.g.,Massmanet al., 1994;Sun and Mahrt, 1994]. For
example,Mahrt et al. [1994] find that for the data set analyzed
in this study,mostof the varianceof the heat and moistureflux
23,125
23,126
MAHRT
ET AL.: OZONE FLUXES OVER A PATCHY CULTIVATED
photochemicalequilibriumwith NO2, ozone, and variousreactivehydrocarbons
(and partiallyoxidizedhydrocarbons),
the
ozone concentrationmay reach even higher valuesthan were
presentbefore the introductionof NO. This is due to ozone
productionon longer timescalesby other chemicalprocesses
(e.g., reactionswith hydrocarbons)that are enhancedby NO
•
[Seinfeld,1986].
This paper investigatesthe relationshipof measuredozone
fluxesto surfacevariability in contrastwith the relationshipof
heat and moisture fluxes to the same surface variability. A
major goal of this studyis to evaluatethe abilityof the aircraft
to estimatespatiallyaveragedsurfaceozonefluxesin the presence of chemical
reactions
between
the aircraft
level and the
surface.Surfaceconditionswill be posed in terms of the normalizeddifferenceof vegetationindex (NDVI) asdescribedby
Tucker[1979]. The next sectiondescribeshow the fluxesand
NDVI are computedfrom aircraft data.
2.
Data Description and Flux Computations
SURFACE
I i , i i j [ i i i.
36
04
35 E
o•
34
o
o 1
o
_
-0.1 .... I .... l, ,'..r"/A'l',,,
,1•,, i ,,,, I,,,,
15
20
0 bare5 mixed10
25mixed 30
soil
irrigatedcrops
35
distance(km)
Figure 1. Normalizeddifferenceof vegetationindex(NDVI)
(solidline) and air temperatureat 33 m (dashedline) filtered
with a movingaveragewindowwith width of 375 m and compositedoverall of the flightlegsfor flight19 (definedas {&(x, t)}
in section2). Northeast (southwest)is directedto the right
(left).
We analyze data from flights of the Twin Otter aircraft
operated by the Canadian Flight ResearchLaboratoryof the
pared with the nonirrigated areas, similar to the variation
Research Council. The instrumentation
is described
by MacPherson[1992] and MacPhersonet al. [1993]. The air- shownby Dotart et al. [1992]. The air temperatureat 33 m for
craft measurementswere obtained at 33 m aboveflat ground, flight 19 varies by 2ø-3øCbetweenirrigated and nonirrigated
which is near the top of the surface layer. The underlying areas and by about 1.5øCfor flight 13.
To estimate fluxes from the aircraft data, we first define a
surface consistsof a patchwork of relatively cool, moist irrigated fields and hot, dry nonirrigatedfields. Ozone measure- movingaverage[&(x, t)], where x is the distancealong the
mentswere carried out usingthe DeutscheForschungsanstalt aircraftleg and &(x, t) representsone of the velocitycompoffir Luft- und Raumfahrt (DLR) fast responseozoneanalyzer nents, temperature, moisture or chemical species.The total
[Schmidtet al., 1991]. This analyzer is based on a surface flow is then decomposedas
chemiluminescentreaction of ozone with a selected organic
4,(x, t) = [4,(x, t)] + 4,' (x, t)
dye absorbedon an aluminumdiskcoatedwith a dry silicagel.
These measurementswere compared with the fast response where O'(x, t) is the turbulentpart of the signalcomputedas
ozonedetectordescribedby Pearsonand Stedman[1980].Both deviationsfrom [ O(x, t) ]. We definethe operator[ O(x, t) ] to
instrumentsagreed well with each other in terms of overall be an equallyweightedmovingaveragewith window width of
spatial structure.We arbitrarily report resultsonly from the either375m (100pointsat thesample
rateof 16s-•), 1 km,or
DLR sensor.The instrument root-mean-square(rms) noise 5 km and then computethe turbulentflux [w' (x, t)O' (x, t)].
level was estimatedto be lessthan 0.1 ppb and is thoughtnot This window is sequentiallytranslated one point at a time
to be a factor in the following analysis.
(approximately3.5 m). A window width of 375 m omits a
The data were collectedduring the California Ozone Dep- significantfraction of the turbulent flux but provides good
osition Experiment (CODE) in the San Joaquin Valley of spatial resolution over the heterogeneoussurface. Window
California about 100 km southof Fresno.The sameflight track widths of 1 km and 5 km capture most of the turbulent flux
was repeated eight times during each of 2 clear days;July 23, [Sunand Mahrt, 1994] but have poorer spatialresolution.
To estimatethe contributionsof the surfaceheterogeneity,
1991 (flight 13), and July 30, 1991 (flight 19). Winds at the
--1
we reduce the influence of transient mesoscalemotions by
33-m flight level were from the northwest at about 4 m s
during flight 13 and light with variable wind direction during compositing[&(x, t)] and [w' &'] overall eightaircraftlegsto
flight 19. The eight legsfor flight 13 were flown betweenap- obtain {[&(x, t)]} and {[w'&']}, where the operator { }
proximately1300 and 1500 local solar time, while the eight indicatesan averageover all of the flight legs at a given locaflight legsfor flight 19 were flownbetweenapproximately1045 tionx. Therefore the averageof the spatialdistributionover all
and 1245 local solar time. Flight 13 was less influencedby of the runs,{[&(x, t)]}, is an estimateof the stationarypart
diurnalvariationbecauseof the strongerwind and later time of of the spatial variation. However, it still containssignificant
day.For flight 19,Mahrt et al. [1994]find that about 10% of the transientfluctuationswhichare not completelyremovedby the
flux variance is due to diurnal variation.
compositing.The estimateof the stationarypart of the turbuThe irrigated and nonirrigated areas under the flight track lent fluxesis computedby compositing[w'&'] over the eight
are delineated as depicted in Figure 1 usingthe NDVI as a flight legsto obtain { [w'O'] } for eachlocation.Since[w' •']
measureof the variation of surfacevegetation[Tucker,1979]. is not smoothedbefore compositing,it alsoincludessignificant
It is computedhere from aircraft-measuredreflectanceat two small-scalevariationsthat were only partially removedby the
wavelengths,one centered at 0.73 •m and the other at 0.66 compositing.Flux errors due to limited sample size for the
•m. The surfaceheterogeneityin CODE is well definedby the compositedspatialdistributionare discussed
by Sunand Mahrt
NDVI (Figure 1) whichis highlynegativelycorrelatedwith the [1994].The flux samplingerrorsfor an individualrun at a given
surface radiation temperature. The surfaceradiation temper- point locationare large, and we currentlyhave no techniqueto
ature is about 20øC cooler over the irrigated croplandscom- quantitativelyestimatethem.
National
23,128
MAHRT
Table
1.
ET AL.: OZONE
Correlation
FLUXES
OVER A PATCHY
SURFACE
Coefficients
q
TA
NDVI
rs
1-km
TA
CULTIVATED
[w' 0']
[w'q']
-0.901
0.764
-0.119
-0.794
-0.949
0.698
0.522
-0.797
[w' CO•]
Window
-0.929
NDVI
0.752
-0.848
Ts
[w' 0']
[w'q']
[w' CO•]
[w' O_•]
-0.756
-0.637
0.729
-0.675
0.083
T•
-0.935
0.812
0.753
-0.755
0.776
-0.116
-0.889
-0.839
0.787
-0.850
-0.125
5-km
NDVI
0.917
0.922
-0.891
0.778
-0.103
-0.054
0.266
Window
-0.941
Ts
[w' 0']
-0.899
-0.749
0.839
0.754
[w'q']
[w' CO•I
[w' O_•]
0.753
-0.645
-0.437
-0.894
-0.837
-0.814
0.827
0.613
0.928
-0.875
0.618
0.381
0.805
-0.780
-0.659
-0.573
0.823
Here, q is specifichumidity,T• is air temperature,NDVI is normalizeddifferenceof vegetationindex,
Ts is surfacetemperature,[w' 0'] is heatflux,[w'q'] is moistureflux,[w' CO•] is carbondioxideflux,and
[w' O_•]•sozoneflux.
the locationof largestdownwardozoneflux for eachof the two
ozonefluxduringflight13,weestimate
thefluxto be 1.2ppb
flights.The aircraftintersects
the enhanceddownwardozone m s-• overa 500-msegment.
If thisfluxrepresents
a localized
fluxon onlytwo of the runswherethe downward
fluxis con- plume of large downwardozoneflux resultingfrom a point
source of NO, then the aircraft data would underestimate the
siderablylarger than in the compositedprofile.
For the run with the mostdefinableeventof large downward maximum downwardflux and the width of the plume if the
Flight 13, run 2
Flight19, run 3
a
b
78
76
72
"--'
O
74
72
7O
68
66
13
'
'
'
I
.
'
'
15
'
'
I
17
'
'
'
'
I
19
'
'
'
'
21
I
'
'
'
'
I
23
'
'
'
'
I
25
'
'
'
'
I
27
'
'
'
1
'
iiii
13
iiii
llll
15
17
''''1''''!''''1
il
19
....
,
21
23
I''''1''''1
25
....
I
27
....
0
0
cO
O
illl
64
62
0.51'
-1
_
o
.o.5
-•
-1
.2
-3
-1.5
-4
-2
-5
Jill
-6
-2.5
0.5 1....
13
I
15
'
'
'
'
I
17
....
I
19
'
'
'
'
I
21
....
I
23
'
'
'
'
I
25
'
'
'
'
I
1
0.31
27
'
'
'
till
13
Ill'
'
II
15
,,,I
II
ill&
17
Ill'
,,,,i,,,,i,,,,i,,,,
19
I1',
21
Ill,
23
I,,,
25
till
27
'111
0.25
0.4
•
cr
0.3
'•
0.2
11.2
-•: 0.15
0.1
0.1
0.05
0
0
• • , , I ....
-0.1
1
13
I ....
15
I , , , , I , , , , I , , , , I , , , , I , , , ,
17
19
Km
21
23
25
,,,,I,,,,I,,,,I,,,,I,,,,i
-0.05
27
1
13
....
15
17
19
21
i,,,,I,,,,
23
25
Km
Figure 3. Spatialvariationof ozoneconcentration
for the flightrun with largestdownwardozonefluxfor (a)
flight 13 and (b) flight 19. The ozoneconcentration
is computedfrom a movingaveragetranslatedone point
at a time (approximately
3.5 m) with a windowwidthof 375 m for ozone(partsper billion),specifichumidity
(gramsper kilogram),and the NDVI.
27
MAHRT
ET AL.' OZONE
FLUXES
OVER A PATCHY
aircraft did not fly through the center of the plume. On the
CULTIVATED
SURFACE
23,129
03 + NO--• NO2 q- 02
(1)
02 q- O •-> 03
(2)
NO2 q- hv--> NO + O
(3)
other hand, the aircraft could overestimate the width of the
plume if it flew obliquelythroughthe plume. Smoothingassociatedwith the 375-mwindowartificiallyreducesthe amplitude
of the downwardozonefluxandincreases
the computedwidth
of the plume. Decreasingthe windowwidth increasesthe amplitude of the flux peak but leadsto noisierspatialdistribution
of the flux. The instantaneous
flux showsseveralsubplumesof
strong flux occurring over a horizontal distance of 500 m.
Fortunately,the total transport is not very sensitiveto the.
choiceof window width, so that the estimationof the ground
sourceis relativelyrobust.With the aforementioneddifficulties
where h v is a photon of sufficientenergyto photolyzeNO 2.
Reaction (2) is sufficientlyfast that (2) and (3) can be combined, sothat neglectingother slowerreactionsinvolvingthese
species,the concentrationbudget of 03 can be expressedas
d[O3]/dt = k3[NO2]- k,[O•][NO]
(4)
where k• and k 3 are the reactioncoefficientsof (1) and (3),
in mindandassuming
a fluxvalueof 1.2ppbm s-I is repre- respectively.Similar equationshold for NO and NO 2. If we
sentativeof a 500-m square,the computedflux correspondsto
a totalintegrated
arealozonelossof about4.8 x 10-4 kgs-I,
or 1.8kgh -•.
The spa.tial
positionof the largedownward
ozonefluxfor
neglect other competing reactions, rapid light intensity
changes,and concentrationfluctuations,then the right-hand
side of (4) is approximatelyzero, so that
[O31[NOl/[NO21= kl/k2.
(5)
flight 13 doesnot coincidewith the positionof the large ozone
deficit.This is probablybecausethe mean ozoneconcentration This is the so-calledphotostationarystate relation [Seinfeld,
resultsmainly from the time integral of the flux divergence 1986].
rather than from the instantaneous flux.
We now considerthe injectionof NO from a surfacepoint
The enhanceddownwardflux of ozone for flight 19 was also sourceinto the boundarylayer. As releasedNO mixesinto the
observedmainly on one run (Figure 3b). Even though the boundary layer, it reduces the ambient O3 concentration
spatial variation of the fluxes for an individual run is quite throughreactions(1)-(3). We assumethat the mixing is slow
noisy,the region of maximum downwardozone flux is quite enough that chemical equilibrium is reached as the mixing
clear.The downward
ozonefluxreaches
5 ppbm s-• overa takesplace. Sincethe reactiontime constantis of the order of
400-m-widearea. This integratedflux correspondsto a surface 100 s, this is a reasonableassumptionexcept within a few
sinkof about1.3 x 10-3 kg s-•. The 400-m-wideregionof meters of the source[Lenschow,1995]. We describethis prolarge downwardflux of ozone coincideswith a region of ozone cessby the following relations:
deficitof about-5 ppbv.Frominstantaneous
16s-• data,this ([O.,] + a[O,])([NO] + a[NO])/([NO:] + a[NO2])
ozone deficit is found to result mostly from a concentrated
plumeof -40 ppbvozonedeficitover a 50-m width. The 400-m
= [O,][NO]/[NOd = k•,/k,
(6)
thicknessof the plume width estimatedaboveis influencedby
8[NO] + a[NO2]----aS[NO]
(7)
the 375-m window usedto computethe flux.
The smallregionof large downwardozoneflux can contrib8[03] + a[$O2] = 0
(8)
ute significantlyto the area-averagedozoneflux. For example,
considerthe run for flight 19 wherelarge downwardozoneflux The quantities8[ ] are the perturbationsin [03], [NO], and
of 5 ppbm s-• occursovera 400-msegment.
This flux event [NO2] introducedby the point sourcereleaseof NO denoted
increasesthe spatiallyaveragedflux over the 16-km irrigated
regionby about 32% comparedwith the value obtainedwithout the flux event. The influenceof this plume is considerably
reduced in a time-spaceaverageusing all of the flight runs.
However, this time average appears to capture additional
lessereventsof large ozoneflux probablyrelated to vehiclesor
other pumpsin the area. Similar conclusions
are reachedfor
flight 13.Theseobservations
suggestthat the spatiallyaveraged
ozoneflux measurementsignificantlyoverestimatesthe ozone
depositionto the natural vegetatedsurfacebecauseof the
ozone lossresultingfrom NO releases.
by aS[NO]. Equation(7) is the conservation
equationfor odd
nitrogen,and (8) is the resultof the stoichiometricreactionof
O 3 with NO [Seinfeld,1986].
Solvingthese equationsfor aS[NO] as a function of 8[03]
and the backgroundconcentrations,we obtain
,a[NO]:
- a[O•][,4 + •]
(9)
where
A = {(k3/k,) + [NO]}/{[O31 + 81031}
To estimate the magnitude of A in (9), we note that the
observedmean concentrationO 3 is about 70 ppbv (Figure 3),
amountof ozoneis releasedas a point source.The immediate a typical deficit 8[03] on the scale of the 375-m window is
impactof a releaseof concentrated
NO into a volumeof air in about -5 ppbv, and a typical clear daytimevalue of k•/k2 is
whichNO, NO2, and 03 are in photochemicalequilibriumis to about 10 ppbv [Seinfeld,1986]. Then the value of aS[NO] is
react with ozone and form NO 2. Within a few minutes the 5.84 ppbv for [NO] = 1 ppbv and 6.54 ppbv for [NO] = 10
of [NO] that are likely presenthere,
three speciesreach a new equilibrium,in which the ozone is ppbv.For concentrations
reducedat the expenseof increasedNO 2 and NO. The air with A is smallcomparedwith 1 and aS[NO] • -8[03]. The validity
decreasedozone mixes upward, resulting in the observed of this approximationdependson scale.For example,the apdownwardozone flux. The timescalefor this mixing is of the proximationgivenabovebreaksdownlocallywhenwe consider
order of a few tensto a few hundredseconds,dependingon the the largestozone deficit on the scaleof a few tens of meters,
which was closerto -50 ppbv.
altitude [Lenschow,1995].
The discussionabove suggeststhat a likely source of the
The photochemicalsequesteringof 03 in sunlightby NO
observedO 3 deficitsis a point release of NO. The maximum
emissioncan be estimatedfrom the rate equations
We assume that the source of NO needed to consume this
23,130
MAHRT
ET AL.' OZONE FLUXES OVER A PATCHY CULTIVATED
SURFACE
downwardozone fluxesfor both flightswere located over sites
with dieselenginesdrivingpumpingstations.The dieselpumps
were locatedwith the downwardlookingvideo cameraand the
useof detailedmaps.The networkof canalsand patchworkof
fieldsallowedunambiguousgroundidentificationof the pumping systems,althoughthe connectionbetween the aircraft observedplume and the pumpingstationscannotbe categorically
model. Gifford's model describesthe rms o-, and mean C
concentrationfields due to the wanderingof an instantaneous
plume by the large eddiesin a turbulentflow. Along the plume
centerline, the model predicts that the fluctuation intensity
o¾/C is given by
proven.
wheref = O'r/O'
m. Sinceo-m = o¾/(1+ f2)1/2, onecanwrite
o'r = a* o¾,wherea* = f/(1 + f2)1/2. According
to Taylor's
In examiningthe regulationsfor allowable emissionsfrom
nonhighwayheavy-dutydiesel enginesfor California, the values are in the range of 6.9-10.7 g of equivalentNO 2 per brake
hp hour (D. Stedman,personalcommunication,1994), where
1 hp = 746 W. The engineemits predominantlyNO, but the
units of measurement are equivalent grams of NO 2. If we
assumean enginerunningat the allowablemaximumstandard
of 10.7g (brakehph)-i, theestimated
ozonesinkof 1.8kgh-l
would result from an engine running at 160 hp.
We thereforeinvestigatedthe possibilityof a largestationary
diesel engine operatingin this area. We then found that for
flight 13, a 70- to 100-hpAllison dieselengine,usedto drive a
tile drain pump, was located underneaththe airplane flight
track at the approximate location of the strong downward
ozoneflux. The pump automaticallyturnsoff and on depending on water levelsandwasapparentlyon for at leastone of the
eight flights.
For flight 19, the largestdownwardozoneflux (Figure 2) is
about 1 km farther east near a canal and correspondsto an
ozone sink 2-3 times larger than that during flight 13. At the
locationof this ozone sink for flight 19, two portableirrigation
pumpswere located,whichwere drivenby 150-hpdieselengines.The large upwardmoistureflux over this area (Figure
3b) may be due to currentirrigationof the underlyingfields.
On the basis of these calculations, we conclude that the most
likely source of such a concentratedozone loss was large,
stationarydiesel engines.
5.
Plume Dispersion
In this section we make a rough estimate of the distance
rr•/C2 =
o-,:.(o-,:.
+ 20-},)]= 1/[f2(f 2 + 1)]
(lo)
[1921]short-rangeresult,absolutedispersionin the cross-wind
(y) directionbecomesoyy = rr•,x/U, where o-•,is the rms
lateral turbulencevelocityand U is the mean wind speed.The
relativedispersion
rrryin the cross-wind
directionis then
o-,y= a* rr7x/U
(11)
The a*(f) requiredin (11) can be estimatedusingFackrell
and Robin's[1982]wind tunnel measurementsof rr,.,/Cdue to
a surfacetracer releasein a neutral boundarylayer. Their data
showedthat alongthe plume centerline(y = z -- 0), rr,./C is
approximately0.6 and is independentofx over the range 1 -<
x/H -< 6, whereH is the boundarylayer height.The invariance
of rr•./C withx can be explainedby a constantratiof = O'r/O',,
in (10). With rr•./C = 0.6, f is approximatelyunity and a* is
approximately0.7.
For flight 19, where the wind is approximatelyperpendicular
to the flight track, the observedinstantaneousplume width Ay
inferred from the instantaneousspatialdistributionof the aircraft measuredozoneflux is 50 m. Assuminga Gaussianspatial
distributionand definingthe width Ay to be the regionwhere
the concentrationexceeds10% of the peak value, we obtain
theobserved
relativedispersion
fromthe approximation
(r,.y=
Ay/4.3 whichyieldsO'ry= 11.6 m. Usingmeasured
valuesof
o-•,= 0.51 ms-l U= 2ms -l anda* =0.7,(11) yieldsx=
65m.
Intermittencyof the pump operation,slightvariationsin the
flight track or the effect of variable wind direction on the
plume might explainthe observationthat the plume at aircraft
level was either
less defined
or nonexistent
on seven of the
eight runsfor flight 19. Generally,the light windswere normal
to the flight track, in which casethe aircraft may have flown
flux measuredby the aircraft is from a "quasi-instantaneous"
over the plume on most of the runs.The best defined plume
plume rather than an Eulerian-averagedplume whichwould was observed when the wind vector had shifted around to
be obtainedby a set of fixed-pointmeasurements.The instanabout 30ø from the flight track, in which case the aircraft
taneousplume is a problemin relativedispersionrepresented
intersectedthe plume downwindfrom the NO source.
by the rms spread rrr about the moving plume centroid,
The small sample size of flux data within the plume is a
whereasthe Eulerian-averageplume is a problem in absolute
seriousproblemfor the presentstudy,and the choicesof some
dispersionrepresentedby the rms spreado¾relativeto a fixed
of the numerical values given above are uncertain. Future
(absolute)coordinatesystem.Both rms valuesrr,.and o¾are
aircraft measurements need to intentionally interrogate a
evaluated at a fixed point in the downstreamdirection and
plume havinga known NO sourcewith many repeatedflight
2 - o-m
2 + o-r2, whereo-,, istherms
related by the expressiono¾
legsparallel with the wind direction.
meander of the instantaneousplume at a fixed point in the
from the NO
downstream
source to the aircraft-measured
ozone flux. The
direction.
Dispersion from a surface source is complicatedby the
height dependenceof the turbulencelength (l c) and time (r)
scales,and the velocityvariances.For a surfacesource,o-u can
be predicted using a z-dependent diffusivityK(z) = lc r
providedthat o¾ _>l•, [van Ulden,1978;l/enkatram,1988].
In a similar way, K theory could be usedfor predictingo-r,
but observationsof o-r as a function of x would be required to
determine a proportionality constant.Instead, we choosea
simpler approach and estimate o-r from a prediction of o¾,
using Taylor's [1921] statisticalplume model and the ratio
ar/au obtained from Gifford's [1959] meandering plume
6.
Conclusions
The above analysisof repeated aircraft runs at 33 m over a
flat surfaceof well-defined irrigated fields has illustrated the
complexityof the spatial distributionof the measuredozone
fluxes.While the spatial distributionsof heat, moisture,and
carbon dioxide fluxes are closelyrelated to variations of the
surfacevegetation,the ozone flux exhibitssmall scalevariations only weakly related to the NDVI.
This variability can be explainedby chemicalreactionsof
ozone.The two largestdistinctmaximaof the downwardozone
MAHRT
ET AL.: OZONE
FLUXES
OVER A PATCHY
CULTIVATED
SURFACE
23,131
flux were found to be above pumping stationsusing diesel Mahrt, L., R. Desjardins,and J. I. MacPherson,Observationsof fluxes
over heterogeneoussurfaces,BoundaryLayerMeteorol.,67, 345-367,
engines.This studyshowsthat the spatiallyaveragedozoneflux
1994.
can significantlyoverestimatethe depositionof the ozone flux Mann, J., and D. H. Lenschow, Errors in airborne flux measurement,
to the natural vegetatedsurfacein the presenceof ozone loss
J. Geophys.Res.,99, 14,519-14,526, 1994.
due to NO
release.
Massman,W. J., J. Pederson,A. Delany, G. den Hartog, D. Grantz,
and R. PearsonJr., A comparisonof independentdeterminationsof
the canopyconductancefor carbondioxide,water vapor, and ozone
exchangeat selectedsites in the San Joaquin Valley of California,
Acknowledgments. We expressour appreciationto Richard Pearpaper presentedat Conferenceon Hydroclimatology:Land-Surface/
son Jr. for helpful discussions,Marvin Wesely and an anonymous
AtmosphereInteractionson Global and Regional Scales.Am. Mereviewerfor helpful commentson the manuscript,and Donald Stedteorol. Soc., Anaheim, Calif., 1992.
man for insightson possiblepoint sources.We alsogratefullyacknowledge Dennis Tristao for informationon the two pumpingsystemsand Massman,W. J., J. Pederson,A. Delany, D. Grantz, G. den Hartog,
H. H. Neumann, S. P. Oncley, R. PearsonJr., and R. H. Shaw, An
Jim Pedersonfor supplementaldata.This material is baseduponwork
evaluationof the regionalacid depositionmodel surfacemodulefor
supportedby grantATM-9310576 from the PhysicalMeteorologyProozone uptake at three sitesin the San JoaquinValley of California,
gram of the National ScienceFoundation and grant DAAH04-93-GJ. Geophys.Res.,99, 8281-8294, 1994.
0019 from the Army ResearchOffice. The field program and analysis
was sponsoredby California Air ResourcesBoard. Funding for the McLaughlin, S. B., and G. E. Taylor, Relative humidity: Important
modifier of pollutant uptake by plants,Science,211, 167-169, 1981.
Twin Otter researchaircraft was providedby the San JoaquinValley
Air PollutionStudyAgency,in collaborationwith the PacificGas and Neu, U., T. Kunzle, and H. Wanner, On the relationship between
ozone storage in the residual layer and daily variation in nearElectric Company,the Electric Power ResearchInstitute and the Calsurfaceozone concentration--A casestudy,BoundaryLayer Meteoifornia Air ResourcesBoard, and by the National ResearchCouncil of
Canada.
References
Chamberlain,A. C., Deposition of gasesand particleson vegetation
and soils, in Air Pollutantsand Their E#kctson the TerrestrialEco,•ystem,edited by A. H. Legge and S. V. Krupa, pp. 189-209, John
Wiley, New York, 1986.
Doran, J. C., et al., The Boardman Regional Flux Experiment, Bull.
Am. Meteorol. Soc., 73, 1785-1795, 1992.
Fackrell, J. E., and A. G. Robins, Concentration fluctuations and fluxes
in plumesfrom point sourcesin a turbulentboundarylayer,J. Fluid
Mech., 117, 1-26, 1982.
rol., 69, 221-247, 1994.
Pearson, R., Jr., and D. H. Stedman, Instrumentation for the fast
responseozone measurementsfrom aircraft,/ttmos. Tech., 12, 5155, 1980. (Available from Natl. Cent. for Atmos. Res., Boulder,
Colo.)
Schmidt, R. W. H., A.M. Jochum, and N. Enisirasser, Airborne ozone
flux measurementsusing a new fast responsedetector, paper presented at SeventhSymposiumon MeteorologicalObservationsand
Instrumentation, Am. Meteorol. Soc., New Orleans, La., 1991.
Seinfeld,J. H.,/ttmosphericChemistryand Physicsof /tir Pollution,738
pp., J. Wiley, New York, 1986.
Sun, J. and L. Mahrt, Spatial distributionof surfacefluxesestimated
from remotely sensedvariables,J. /tppl. Meteorol.,33, 1341-1353,
1994.
Fitzjarrald,D. R., and D. H. Lenschow,Mean concentrationand flux Taylor, G. I., Diffusion by continuousmovements,Proc.London Math.
profiles for chemicallyreactive speciesin the atmosphericsurface
Soc., Ser. 2, 20, 196-212, 1921.
layer,/ttmos. Environ., 17, 2505-2512, 1983.
Tucker, C. J., Red and photographicinfrared linear combinationsfor
Gifford, F. A., Statisticalpropertiesof a fluctuatingplume dispersion
monitoringvegetation,RemoteSens.Environ., 8, 127-150, 1979.
model,,•ldv. Geophys.,6, 117-138, 1959.
van Ulden, A. P., Simple estimatesfor vertical diffusionfrom sources
Kramm, G., H. Muller, D. Fowler, K. D. Hofken, F. X. Meixner, and
near the ground,/ttmos. Environ., 12, 2125-2129, 1978.
E. Schaller,A modifiedprofile methodfor determiningthe vertical Venkatram, A., Dispersionin the stableboundarylayer, in Lectureson
fluxesof NO, NO2, ozone, and HNO3 in the atmosphericsurface
Air PollutionModeling,editedby A. Venkatram and J. C. Wyngaard,
layer,J. /ttmos. Chem., 13, 265-288, 1991.
pp. 229-265, Am. Meteorol. Soc., Boston, Mass., 1988.
Lenschow,D. H., Reactivetrace speciesin the boundarylayer from a Wesely, M. R., Turbulent transport of ozone to surfacescommon in
micro-meteorologicalperspective,J. Meteorol. Soc. Jpn., 60, 472the easternhalf of the United States,in Trace/ttmo•7•hericConstit480, 1982.
uents: Properties, Transformations, and Fates, edited by S. E.
Lenschow,D. H., Micrometeorologicaltechniquesfor measuringbioSchwartz,pp. 345-370, John Wiley, New York, 1983.
sphere-atmospheretrace gas exchange,in Biogenic Trace Gases: Wesely, M. R., D. R. Cook, and R. W. Williams, Field measurementof
MeasuringEmissionsFrom Soil and Water, edited by P. A. Matson
small ozonefluxesto snow,wet bare soil, and lake water, Boundary
and R. C. Harriss,chap. 5, pp. 126-163, BlackwellSci., Cambridge,
Layer Meteorol., 20, 459-471, 1981.
Mass., 1995.
Lenschow,D. H., and A. C. Delany, An analytic formulation for NO
R. L. Desjardins, Land ResourcesResearch Centre, Agriculture
and NO 2 flux profiles in the atmosphericsurfacelayer, J. /ttmos. Canada, Ottawa, Ontario, Canada K1A 006.
Chem., 5, 301-309, 1987.
D. H. Lenschow,National Center for AtmosphericResearch,P.O.
Lenschow,D. H., R. PearsonJr., and B. B. Stankov, Estimating the
Box 3000, Boulder, CO 80307.
ozonebudgetin the boundarylayer by useof aircraft measurements
J. I. MacPherson,Flight Research Laboratory, National Research
of ozone eddy flux and mean concentration,J. Geophys.Res., 86,
Council, Ottawa, Ontario, Canada K1A 006.
7291-7297, 1981.
Leuning, R., H. H. Neumann, and G. W. Thurtell, Ozone uptake by
corn (Zea maysL.): A generalapproach,/tgric.Meteorol.,20, 115135, 1979.
MacPherson,J. I., NRC Twin Otter operationsin the 1991 California
Ozone DepositionExperiment,Rep.LTR-FR-118, Flight Res. Lab.,
Natl. Res. Counc., Ottawa, 1992.
MacPherson, J. I., R. W. H. Schmidt, A.M.
Jochum, R. Pearson Jr.,
H. H. Neumann, and G. den Hartog, Ozone flux measurementon
the NRC Twin Otter during the 1991 California Ozone Deposition
Experiment,paper presentedat Eighth Symposiumon Meteorological Observations, Am. Meteorol. Soc., Anaheim, Calif., 1993.
L. Mahrt, College of Oceanic and Atmospheric Sciences,Oregon
State University, Corvallis, OR 97331-2209.
J. Sun,Programin Atmosphericand OceanicSciences,Universityof
Colorado, Boulder, CO 80309.
J. C. Well, Cooperative Institute for Research in Environmental
Sciences,University of Colorado, Boulder, CO 80309.
(ReceivedJanuary4, 1995; revisedAugust4, 1995;
acceptedAugust 23, 1995.)