21562202d, 1999, D5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/98JD02309 by University Estadual De Campina, Wiley Online Library on [23/01/2023].

See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 104, NO. D5, PAGES 5829-5843, MARCH 20, 1999

On the origin of troposphericozoneand NOx

over the tropical South Pacific
MartinG. Schultz,
• DanielJ.Jacob,
• Yuhang Wang, •'2Jennifer
A. Logan,• ElliotL.
Atlas,
3DonaldR. Blake,4NicolaJ.Blake,4JohnD. Bradshaw,5Edward V. Browell,6
MartaA. Fenn,6FrankFlocke, 3GeraldL. Gregory,
6BrianG. Heikes,?GlenW.
Sachse,?ScottT. Sandholm,8RichardE. Shetter,
3Hanwant B. Singh,
9 andRobert
W. Talbot•ø

Abstract. The budgetsof ozoneandnitrogenoxides(NOx = NO + NO2) ill the tropicalSouth

Pacifictroposphere areanalyzedby photochemical pointmodelingof aircraftobservations at 0-12
km altitudefrom the PacificExploratoryMission-Tropics A campaignflown in September-
October1996. The modelreproduces the observedNO2/NO concentration ratioto within 30% and
hassimilarsuccess in simulatingobserved concentrations of peroxides
(H202,CH3OOH),lending
confidencein its useto investigateozonechemistry.It is foundthatchemicalproductionof ozone
balancesonlyhalf of chemicallossin thetropospheric columnoverthetropicalSouthPacific.
Thenetlossis 1.8x 1011
molecules
cm-2S'1.Themissing
source
ofozone
ismatched
bywest-
erlytransportof continentalpollutionintotheregion.Independent analysisof theregionalozone
budgetwith a globalthree-dimensional modelcorroborates theresultsfromthepointmodeland
revealsthe importanceof biomassburningemissions in SouthAmericaandAfrica for the ozone
budgetoverthetropicalSouthPacific.In thismodel,biomassburningincreases averageozone
concentrationsby 7-8 ppbvthroughout thetroposphere. The NOx responsiblefor ozoneproduc-
tion withinthe SouthPacifictroposphere below4 km canbe largelyexplainedby decomposition
of peroxyacetylnitrate
(PAN) transported intotheregionwith biomass burningpollutionat higher
altitudes.

1. Introduction focusedon improving our understandingof ozone chemistry

and the resulting ozone budget in differentregions of the
Ozonein the tropical troposphereis a major global source tropical troposphere[Harriss et al., 1988, 1990; Hoell et al.,
of OH radicalsand henceplays an importantrole in maintain- 1996, 1997; Fishman et al., 1996]. The latest of these cam-
ing the oxidizingpowerof the atmosphere[Thompson,1992]. paigns, the Pacific Exploratory Mission-Tropics A (PEM-
Ozoneis produced in the troposphereby photochemicaloxi- Tropics A), surveyed the distributions of ozone, NO•, CO,
dation of CO and hydrocarbonsin the presenceof nitrogen hydrocarbons, and relatedspeciesover the tropical South Pa-
oxide radicals(NO•- NO + NO2). NOx is in generala limiting cific in September-October1996 [Hoell et al., this issue].We
precursorfor ozoneproduction[Chameideset al., 1992]. Loss present here a photochemicalmodel analysis of the PEM-
of ozone is principally by photolysis and by chemicalreac- Tropicsdata aimed at constructinga budgetof ozone and NO•
tions with OH and HO2. Several aircraft campaignsof the for the SouthPacific troposphereand understandingthe links
NASA Global TroposphericExperiment(GTE) programhave to otherregionsof the tropicaltroposphere.
The SouthPacific is the region of the tropical troposphere
•Division
of Engineering
andApplied
Sciences
andDepartment
of mostremotefrom humanactivity.Troposphericozone columns
Earthand PlanetarySciences,Harvard University,Cambridge,Massa- are the lowest in the world, with values as low as one third
chusetts.
2Nowat School of EarthandAtmospheric
Sciences, Insti- thosefound in pollutedregionsof the tropics[Fishmanet al.,
Georgia
tute of Technology,Atlanta. 1990; Kirchhoff et al., 1991; Fishman and Brackett, 1997].
3National CenterforAtmosphericResearch,
Boulder, Colorado. The timing of PEM-Tropicscorresponded to the endof the dry
4Department of Chemistry,University
ofCalifornia,
Irvine. seasonin the southern tropics, when considerablebiomass
5Deceased June16, 1997.
6NASALangley ReasearchCenter,
Hampton,Virginia. burning takes place in South America,Africa, and Oceania.
7Graduate School of Oceanography, of RhodeIsland, The GTE/Transport and Atmospheric Chemistry Near the
University
Narragansett. EquatorialAtlantic (TRACE A)campaignconductedin Sep-
8School
of EarthandAtmospheric
Sciences,
Georgia
Institute
of tember1992 [Fishmanet al., 1996] demonstrated
that
Technology,Atlanta. biomassburning in southern Africa and Brazil causessea-
9NASA AmesResearch Center,
Wallops,
MoffettField,California.
•øInstitute
fortheStudy of Earth,Oceans,
andSpace, Universityof sonal enhancementsin troposphericozone by a factor of 2
New Hampshire,Durham. overthe tropicalSouthAtlantic [Olsonet al., 1996; Thomp-
son et al., 1996]. The TRACE A observations confirmed
Copyright1999 by the AmericanGeophysicalUnion.
previousevidenceof large-scalebiomassburning pollution
overthetropicalSouthAtlanticl•omsatelliteobservations
•f
Paper number98JD02309. ozone and CO [Fishman et al., 1990; Watson et al., 1990].
0148-0227/99/98JD-02309509.00 Fishman et al. [1991] proposedthat long-rangetransport of
5829

5830 SCHULTZ ET AL.: ORIGIN OF TROPOSPHERIC OZONE AND NOx
biomassburning pollution would affect ozone on a hemi-
sphericscale.A major objective of PEM-Tropics A was to
determinethe extent of seasonal biomassburning influence
over the South Pacific.
In additionto biomassburning,naturalsourcesof NOx from
soils and lightning contribute to ozone formation in the
tropical troposphere.Using the first aircraftobservationsof
NOx over the SouthPacific, Liu et al. [1980] proposeda self-
sustainingmodel for ozone in the tropical tropospherewhere
net productionof ozonein the uppertroposphere balancesnet
loss in the lower troposphere.Ozone production in their
model was sustained by transport of NOx from the strato-
sphere, which is now known to be negligibly small
[Kasibhatla et al., 1991; Penner et al., 1991; Lamarque et
al., 1996]. More recent aircraft observationsof NOx in the
tropicaltroposphere have demonstrated major primarysources
fi'omlightning [Drummond et al., 1988; Smyth et al., 1996;
Crawford et al., 1997], soils [Jacoband Wofsy,1988, 1990],
and biomass burning [Andreae et al., 1988; Singh et al.,
1996b; Smyth et al., 1996]. A notable featureof these three
NOx sourcesis that they are all continental (lightning over
the oceansis extremelyinfrequent).
A photochemical modelanalysisof the TRACE A observa-
tions [Jacob et al., 1996] indicated a chemicalbalance for
ozoneon the scaleof the troposphericcolumn.On the basis of
this analysis,Jacob et al. [1996] proposeda schematicfor the
ozonebudgetin the tropical tropospherewhereNOx pumped
convectively over the tropical continents (upwelling
branchesof the Walker circulation) promotesozone produc-
tion in the upper troposphere,to be balanced by ozone
depletionin the lower troposphereafterthe air subsidesover
the subtropical anticyclones (downwelling branchesof the
Walker circulation).A global three-dimensionalmodel analy-
sis by Wang et al. [1998c] finds a close balance between
zonal meanproduction and loss of ozone in the tropical tro-
posphere,suggesting that the tropical troposphere is self-
sustainingwith respectto ozone, at least on a zonal scale.
The budgetof ozone in the troposphereis strongly linked
to that of NOx. Considering that the lifetime of NOx with re-
spectto oxidationto HNO3 is only of the orderof l day, one
would expectNOx concentrations to decayrapidly downwind
of the continents. However, observationsover the tropical
oceansindicatefree troposphericNOx concentrationsof typi-
cally 10-50 parts per trillion by volume(pptv) [Carroll and
Thompson,1995; Emmons et al., 1997], which play a major
role in ozone production and are muchhigher than would be
expectedfi'omdirect transport of primaryNOx fromthe conti-
nents. It has been proposed that chemicalrecycling of NOx
from HNO3 and othernon-radicalreservoirs,including in par-
ticular peroxyacetylnitrate(PAN), could play a majorrole in
maintaining NOx concentrations over the oceans [Singh,
1987]. However, previous photochemicalmodel analysesof
observationsin the tropical marine troposphereconcluded
that chemicalrecyclingbasedon known reactionswas insuffi-
cient to maintainNOx concentrations at their observedlevels,
and speculatedon the possibility of fast conversionof HNO3
to NOx in aerosols [Chatfield, 1994; Hauglustaine et al.,
1996; Jacob et al., 1996; Lary et al., 1997].
2. Data Summary
Two aircraftparticipatedin the PEM-TropicsA campaign,a
DC-8 and a P-3B. Both carried instrumentation for measure-
ment of ozone, NO, CO, hydrocarbons, peroxides, and a
21562202d, 1999, D5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/98JD02309 by University Estadual De Campina, Wiley Online Library on [23/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
numberof other species.The DC-8 included measurements of
PAN and HNO3 and had a higher ceiling (12 km altitude, as
comparedto 7 km for the P-3B). The analysishere focuseson
the DC-8 data.
The DC-8 operatedfrom four bases: Tahiti, Easter Island,
Christchurch(New Zealand), and Fiji. Sorties out of these
basesexaminedthe chemicalcompositionof the South Pacific
troposphereover a domainstretchingfi'omthe warm pool of
the equatorialPacific to the subsiding region of the tropical
easternPacific, and fi'omthe intertropical convergencezone
(ITCZ) to high southern latitudes (Figure 1). The ITCZ was
found in transitflights to be locatedat 7ø-9øN.
Tables l a and lb list median concentrationsof ozone, NOx,
PAN, HNO3, peroxides,CO, and selectedhydrocarbonsob-
served at different altitudes in the equatorial and tropical
South Pacific. We divide the data into two latitude bands, 0 ø-
10øSand 10øS-30øS,becauseof the pronounceddifferencesin
ozoneand CO concentrations(Figure 2) [seealso Gregory et
al., this issue].Figure 3 comparesthe observationsfi'omPEM-
TropicsA in the latitude band 0ø-30øS to previous tropical
data collected in the same season over the northwestern Pa-
cific (PacificExploratoryMission-WestPhaseA) (PEM-West
A) [Hoell et al., 1996] and over the South Atlantic (TRACE
A) [Fishmanet al., 1996]. While the verticalprofiles of ozone
and PAN (both photochemicalproducts) fi'omPEM-West A
fall on the low end of the PEM-Tropicsdata,the profilesof CO
and ethane (tracersfor urban/industrial pollution or biomass
burning) are shifted to somewhathigher concentrations.The
datafrom TRACE A exhibit higher concentrationsfor several
observedspecies.The largestdifferences are found for PAN:
above 2 km, the TRACE A PAN concentrationis at least a fac-
tor 4 higherthan in the other two data sets.
The statistical distribution of ozone and CO at 4-8 km dur-
ing PEM-Tropics A is displayed in Figure 2 for the five
geographicalregionsdefinedin Figure 1. Both speciesfollow
similarpatterns.The cleanestair was observedover the west-
ern equatorial Pacific and north of the subtropical
convergencezone (SPCZ). Here, medianmixing ratios were
27-32 ppbv for ozoneand 56-58 ppbv for CO. The variability
is alsovery small:the central50% of the data spana rangeof
lessthan 10 ppbv.The highestmedianvaluesfor both species
were found over the western tropical Pacific (south of the
SPCZ),with valuesof 61 ppbvfor ozoneand 78 ppbv for CO.
Thesedataalsoshow a large variability, indicatingthe influ-
ence of episodic pollution events.During PEM-TropicsA,
severallayers with continentalpollution were encounteredin
westerly air flow which originated frombiomassburning in
Africaor SouthAmerica[Talbotet al., this issue;Fuelberget
al., this issue].Figure 2 showsoccasionalozone and CO mix-
ing ratios in excessof 100 ppbv for the tropical western Pa-
cific south of the SPCZ. Almost 25% of the data from this
regionhad ozonemixing ratiosabove80 ppbv.This pollution
influence decreasedfrom west to east, reflecting dispersion
and photochemicaldecay.
3. Model Description
We use the Harvard photochemicalpoint model to recon-
structthe photochemicalenvironmentalong the DC-8 flight
tracks. The model calculates diurnal steady-stateconcentra-
tions of radicals and chemical intermediatesusing as con-
straints the ensemble of observations from the aircraft. Diurnal
steadystateis definedby repeatability of concentrationsover
 21562202d, 1999, D5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/98JD02309 by University Estadual De Campina, Wiley Online Library on [23/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SCHULTZ ET AL.: ORIGIN OF TROPOSPHERIC OZONE AND NOx 5831

45'

ß
' Ha•ai! ,
,

ß ,
, 30'
ß

ß
'" ,

,
:ITCZ
,
, :
,

.
,

Christmas ,

'western IS central 15'

:equatorial. equatorial

eastern
'tropical
western 'central
,

15'
tropical tro

' o
,
,
,

Christ- .

,
,
o

ehuz'9.
h
. ,

1;5. ; . ' ' .' .

is,
o'" •5o-w
' •:•p• ..'"...
"' : ' " .'";0.•'•:.'."¾

Figure 1. Flight tracksof the DC-8 aircraftduringPEM-TropicsA. Thin solidlinesdelineatethe regionsdiscussed

in the text. Aircraft crossingsof the intertropicalconvergence
zone (ITCZ) and the South Pacific convergencezone
(SPCZ) are indicatedby tilde symbols.

a 24-hourcycle.The chemicalmechanism is basedon the com- non-methanehydrocarbons were interpolatedfor a maximumof

pilationsof DeMore et al. [1997] andAtkinsonet al. [1997] 5 min when necessary;remaininggaps in CH4, HNO3, PAN,
with updates(e.g,. pressure-dependent
quantumyieldsfor ace- and nonmethanehydrocarbonswere substituted by median
tone photolysis [McKeen et al., 1997]). All model results valuesfrom Table lb. For the analysisof the NOx budget (sec-
were submittedto the NASA Langley Distributed Active Ar- tion 6) we usedthe time resolutionof the HNO3 measurements
chive Center(DAAC) archiveand are also available•om the which had the longestintegrationtimes, and we only consid-
GTE web site http://www'gte'larc'nasa'gøv' ered time intervals with observations of both PAN and HNO3.
Actinic fluxes are calculated with a six stream discrete ordi- For water vaporwe used measurements fromthe differential
nate radiative transfer model for the Rayleigh scattering absorption CO measurement (DACOM) instrument(available
atmosphere [Loganet al., 1981]. The overheadozonecolumn in 70% of all cases)and fromthe cryogenicallychilled mirror.
for the radiativetransfercalculationis specifiedfromdaily to- Although both measurements agree on averagewithin 20%,
tal ozonemappingspectrometer (TOMS) observationsof total for mixing ratios below 100 ppmv the cryogenically chilled
ozone columns,subtractingthe ozone columnbelow the air- mirror tendsto be lower by a factorof 2 to 5. This can be ex-
craftas measuredby the differentialabsorptionlaser (DIAL) plained partly by the fact that the chilled mirrorhas a longer
instrumentand interpolating when necessary.In order to ac- responsetime and seemsto undershootwhen the aircraft en-
count for cloud scattering and absorption, we scaled the countersdry air (J. Barrick, personal communication,1997).
actinicflux for eachindividualdata point with the ratio of ob- The DACOM dataappearto be morereliable especiallyin the
servedto clear-sky modeledphotolysis frequenciesin two upper troposphere.However, the relative humidity derived
wavelengthregimes:for wavelengthsup to 310 nm a correc- from the DACOM data often exceeds 100% below 220 K,
tionfactorforj(OiD)was applied,above310 nmwe useda which would indicatea positivebias.
correctionfactor for j(NO2). Median correctionfactors were Acetone can act as the major primarysourcefor hydrogen
1.21forj(NO2)and1.03forj(O1D). oxideradicals(HOx = OH + peroxyspecies)in the upper tro-
For model input the observationswere mergedover 60 s posphere[Singh et al., 1995; daegldet al., 1997; McKeenet
time intervals. Measurementswith a higher sampling fi'e- al., 1997; Arnold et al., 1997]. Since there were no measure-
quencywere averaged.An observationwas consideredvalid ments of acetone,we prescribeda constant acetone mixing
only when all mandatoryspecies(03, NOx, H20, GO, HNO3, ratio of 0.4 ppbv as a typical value for the southerntropics
PAN, peroxides)coveredat least 25% of the interval. Obser- [Wang et al., 1998b].
vationsof water vapor,CO, CH4, HNO3, PAN, peroxides,and Exceptfor NO, all input variablesto the modelcalculation
 21562202d, 1999, D5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/98JD02309 by University Estadual De Campina, Wiley Online Library on [23/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
5832 SCHULTZET AL.: ORIGINOF TROPOSPHERIC
OZONEAND NOx

Tablela. MedianConcentrations
OvertheEquatorial
South
Pacific
(0ø-10øS)
in PEM-Tropics
A

Altitude Band,km

0-2 2-4 4-6 6-8 8-10 10-12

Observedquantities

Ozone,ppbv 12 23 38 32 37 31

NOx,pptv 4 7 15 19 37 28

PAN,pptv <1 2 19 23 32 19

HNO3,pptv 23 59 74 52 59 27

H202,pptv 1410 1270 670 530 200 270

CH3OOH,pptv 1170 840 360 250 80 110

CO,ppbv 55 53 53 57 61 57

Ethane,
pptv 245 265 290 280 300 290

Propane,
pptv 13 16 17 17 18 19

Acetylene,
pptv 16 18 25 28 31 34
Point model results

OH,106molec./cm
3 1.64 1.76 1.11 1.35 0.91 0.66
HO2,106molec./cm
3 139 129 86 77 39 34
CH302,
106molec./cm
3 234 166 69 37 13 11
HNO4,pptv <0.1 0.1 1 5 19 16

CH20,pptv 340 220 130 100 50 40

Datacollected
inSeptember-October
1996over
theequatorial
regions
shown
inFigure
1.Model
values
are24-hour
averages.
This
tablecontains
1430values
forspecies
with60ssampling
timeorless,380hydrocarbon
samples,
and620model
runs.

are assumedconstantover the diurnal cycle. Becauseof the

andnumberdensityof aerosols(0.007-6 gin)measuredaboard
rapid cycling betweenNO and NO2, we specifyinsteadtotal
the P-3 aircraftin two altituderegimes.
NOxt (= NO + NO2+ NO3+ 2-N205+ HNO4+ HONO)ascon- Becauseof the requirementof concurrentmeasurements
of
stantandcalculate theconcentrationsof the NOxt componentsNO, peroxides, andCO, modelresultsareavailableonly for
as time-dependent
variablesoverthe diurnalcycle [Jacobet about40% of the PEM-TropicsA flight period.We checked
al., 1996].Theconcentration
of NOxt is adjusted iterativelyto therepresentativeness
of themodeldatasetin variousregions
match the observed NO concentration at the local time of the oftheSouthPacificby comparingplotsofthefrequency dis-
observation. Observations of NO2 are not used to constrain
tributionof the key speciesNO, 03, H20, H202, and CH3OOH
themodelandserveinsteadfor modelevaluation
(section4). from boththe completeobservational andthe modeldataset in
The chemicalmechanismcontains
threeheterogeneous reac- altitudebandsof 2 km.Most species comparewell in all re-
tions:
gionsexceptabove10 kmwheresampling
of somespecies
is
(R1) N205 + H20 •> 2 HNO3, ¾= 0.1
relativelysparse:in that altituderangethe modeldatahave
significantlow biasesfor NO (factor0.48) overthe central
tropicalPacific,andfor ozone(factor0.55) overthe western
(R2) NO3+ H20 •> HNO3, T = 0.01 tropical Pacific south of the SPCZ. There are no model data
above10 km in the easterntropicalPacific.
(R3) HO2
+H20
-4HO
2 H20 HO2 ß ) A different
aspectof representativeness
is the questionof
H20 2 + 0 2 q-H20, ¾= 0.1 whetherthe PEM-Tropics observations represent
average
conditions
forthe southern
Pacifictroposphere
in September.
The reactionprobabilities
¾ follow the recommendations
of Whilethemeteorologicaldataindicatethattemperatures
and
Jacob[1998]withtheupperlimitfor (R2).The aerosolsurface flowpatterns
in 1996werecloseto the climatological
mean
areawasestimated
fromthe condensation
nuclei(0.004-1grn) [Fuelberget al., thisissue],the ozonesonde
datafromChrist-
measurementsusing mean correlations between surface area church
andTahitiexhibitsome
record
highvaluesduringthe
 21562202d, 1999, D5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/98JD02309 by University Estadual De Campina, Wiley Online Library on [23/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SCHULTZ ET AL.: ORIGIN OF TROPOSPHERICOZONE AND NOx 5833

Table lb. MedianConcentrations

OvertheTropicalSouthPacific(10øS-30øS)
in PEM-Tropics
A

Altitude Band,km

0-2 2-4 4-6 6-8 8-10 10-12

Observedquantities

Ozone,
ppbv 28 50 59 43 44 45

NOx,pptv 3 22 26 18 58 98
PAN,pptv 1 19 89 43 48 42

HNO3,pptv 18 145 155 60 65 47

H202,pptv 930 1040 620 560 240 130
CH3OOH
, pptv 880 480 250 310 90 < 35
CO,ppbv 58 68 70 59 63 58
Ethane,
pptv 310 380 400 340 350 360
Propane,
pptv 22 22 24 22 26 27
Acetylene,
pptv 52 67 68 51 52 53
Point model results

OH,106molec./cm
3 1.47 1.27 1.17 0.94 1.04 0.96
HO2,106molec./cm
3 121 115 78 61 39 22
CH302,
106molec./cm
3 217 90 49 33 13 5
HNO4,pptv < 0.1 0.7 3 5 28 29
CH20, pptv 280 200 125 80 65 50
Datacollected
in September-October
1996overthetropical
regionsshownin Figure1. Modelvaluesare24-houraverages.
This
tablecontains
3780valuesforspecies
with60 s sampling
timeorless,1182hydrocarbon
samples,
and1387modelruns.

time of the expedition (J. Logan, personal communication, andby increasing

the airflow.For the photodissociation
of
1997). NO2 the previouslyused Xenon lamp was replacedby a
pulsedlaser.All threeimprovements
wereexpected
to reduce
wall lossesand interferenceeffectson the NO2 measurement(S.
4. Cycling of NO and NOz Sandholm,personalcommunication, 1997).
Comparisons
betweenmodeledandobserved
NO2/NO con- Figure4 displays
the ratio of the modeledversusobserved
centrationratios have been used previously to evaluateNOx NO2/NO ratio in PEM-Tropics A as a function of altitude.
photochemistry
in modelsandto assess the qualityof the NO2 Only datawith NO2 photolysisfrequencies j(NO2) greater
[e.g., Charneldeset al., 1990; Davis et al., than7 x 10'3 s'• wereusedin orderto preventa possiblebias
measurements
1993; Schultz, 1995; Crawford et al., 1996]. Many previous fi'omincorrectNOx estimatesin the point model. The model-
studies found that model-calculatedNO2/NO ratios are sys- calculatedNO2/NO ratios are generallytoo high by about
tematicallylowerthanobserved ratiosby a factorof about 1.5. 10%. This is well within the stated uncertainties of the meas-
This discrepancyshoweda tendencyto increasefor cleaner urements
of NO2 andj(NO2) (20-30%and 10%,respectively).
air, that is, closerto the limit of detection(LOD) for the NOx Below2 km a systematic
negativebiasof about20% is ob-
measurements [Schultz, 1995; Crawford et al., 1996]. In served which is still within the stated uncertainties,
PEM-WestA, Crawfordet al. [1996] found a medianratio of especially
sinceNOx concentrations
werevery low. However,
experiment to theoryof 3.3. Theiranalysisled to the conclu- halogenchemistryin the marineboundarylayercouldalso
sion that the measurement of NO2 which was made with an cause this bias. We were able to reconcile the modeled
older version of the two photon laser induced fluorescence NO2/NO ratio with the observations
below 2 km by introduc-
(TP-LIF) instrument [Bradshawet al., 1985, Sandholmet al., inga chlorine atomconcentration of 105molecules cm '3 as
1990]wasproneto interferences with otherphotolabileNOy suggested fi'omobservationsof hydrocarbonratios over the
compounds.For PEM-TropicsA the TP-LIF techniquewas tropicalwesternPacific[Singhet al., 1996a]. Under these
power conditions the C10 + NO reaction is a significant pathway
improvedby installingtwo new laserswith enhanced
 21562202d, 1999, D5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/98JD02309 by University Estadual De Campina, Wiley Online Library on [23/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
5834 SCHULTZ ET AL.' ORIGIN OF TROPOSPHERIC OZONE AND NOx

150 underestimates the sourceof CH3OOH by 40-130%, while the

H202 budget remainsbalanced within 10-40% on average.
Deep convectionelevatesCH3OOH in the upper troposphere
125 above chemicalsteady state, while having little impact on
H202 due to scavenging by the convective precipitation
[Prather and dacob, 1997;daegldet al., 1997]. During PEM-
• 100 TropicsA, Cohanet al. [this issue]found that CH3OOH con-
centrationsin deep convective outflow at 8-12 km altitude
.,-.•
were elevatedby a factorof 6 relative to background,while
• 75 H202 concentrations were elevatedonly by 50%. Using CH3I
as a chemicalclock, they estimateda timescaleof 2-3 days for
the decay of the convective enhancementof peroxides in the
.,-.•

.,-.• 50 uppertroposphere;this timescaleis still relatively short com-

pared to the 10 day turnover time for the upper troposphere
[Prather and dacob, 1997].
Thereforeit doesnot appearthat convectivetransportalone
can explainthe imbalancein the CH3OOH budgetin Figure 5.
Another contributing factor could be an underestimateof the
o
rate constant for the CH302+ HO2 reaction which is uncertain
by abouta factor of 3 at temperaturesaround220 K [DeMore
et al., 1997].
Previous studies of the PEM-West and TRACE A data
[Davis et al., 1996a;dacobet al., 1996; Crawford, 1997] cal-
culated peroxide concentrationsfrom chemicalsteady state
and comparedthese to the observations.dacob et al. [1996]
Figure 2. Statisticaldistributionsof ozone and CO mixing ra- report a significant underestimateof H202 model concentra-
tios at 4-8 km altitude in the five PEM-Tropics A regions tions (median ratios of model/observedof 0.99, 0.87, and 0.57
definedin Figure 1. Horizontal lines are medianvalues, boxes
in the altitude bins 0-4, 4-8, and 8-12 km), and an even worse
spanthe central 50% of the probability distribution, and the
vertical lines cover the central 90%. The individual data are 1 imbalancefor CH3OOH (ratios are 0.59, 0.79, and 0.28). For
min. averages. PEM-West B the modelof Crawford [1997] typically overes-
timatesH202 by about30% in the equatorialregion below 10
km, exceptfor the surfacelayer wheredepositionplays a role.
converting NO to NO2. Sensitivity calculations with BrO On the other hand, CH3OOH is generally predicted within
levels of 107molecules cm'3[Vogtet al., 1996]or Ix(= HOI + about20% up to 8 km, and it is underpredictedby a factorof
HI + IO + I + minorcomponents) concentrations of3.8 x 107 2.4 and 5 in the altitude bins of 8-10 and 10-12 km, respec-
molecules cm-3[Davisetal., 1996b]showed no significant ef- tively. For PEM-West A, Davis et al. [1996a] give only one
fect on the NO2/NO ratio. medianvalue for the ratio of observedto modeled peroxide
concentrations above 1 km (0.92 for H202 and 0.62 for
5. Peroxides CH3OOH). With the exceptionofdacob et al. [1996] all these
studies indicate that a photochemicalsteady-statemodel can
The peroxidesH202 and CH3OOH are producedby recombina-
successfullysimulate H202 in the free troposphere,whereas
tion of peroxy radicals and have lifetimes of 1-2 days with
CH3OOH is systematically underestimatedby the models
respectto photolysis and reaction with OH for the PEM- above 8 km.
Tropics A conditions. One would thereforeexpectH202 and
CH3OOH to be in generalnear chemicalsteadystate,and the
ability of the model to reproducethis steady state provides a
6. Sources and Sinks of NOx
test for the peroxy radical concentrationsand hencethe ozone
productionrates calculatedby the model.In our calculations The lifetime of NOx with respectto photochemicalloss, as
we use the observed peroxide concentrationsas model con- calculated by our photochemicalmodel for PEM-Tropics A
straints and examine the resulting imbalance between conditions, is 1-5 days for mostof the tropical South Pacific
photochemicalloss and photochemicalproduction as diag- troposphere(Table 2). The PEM-Tropicsstudyregion is suffi-
nostic of departurefrom steady state. This approachhas the cientlyfar from mostknownprimarysourcesof NOx (fossilfuel
advantageof not imposingthe assumptionof steady state for combustion,biomassburning, soil emissions,lightning) that
peroxideson the calculationof HOx concentrations. we would expectNOx to be generally near chemicalsteady
Figure 5 showsthe balanceof simulated24-hour-averaged state,maintainedby recyclingof HNO3 and PAN.
chemicallossand productionratesfor H202 and CH3OOH as a Figure6 displaysmeanmodeledNOx budgetsfor the equa-
functionof altitude.Between2 and 8 km the balanceis gener- torial and tropical regions,constrainedwith observationsof
ally within 10% for H202 and 15% for CH3OOH. Below 2 km NO, PAN, and HNO3. The chemicalmechanismof the model
the agreementis still within 10% for CH3OOH, but resultsfor takes into account the cycling between NO, NO2, HNO3,
H202 indicate a missing sink of magnitude comparableto PAN, and HO2NO2. Methyl nitrate is also included as NOx
chemicalloss.This missingsink can be attributed to wet and sourcebecauseobservationsof severalpptv even above 6 km
dry deposition, which is not included in the point model. suggestedit might be importantunder conditions of very low
Above 8 km the scatterincreasessubstantially,and the model HNO3 and PAN (F. Flocke, personal communication,1997).
 21562202d, 1999, D5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/98JD02309 by University Estadual De Campina, Wiley Online Library on [23/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SCHULTZET AL.' ORIGINOF TROPOSPHERIC
OZONEAND NOx 5835

12

10

20 40 60 80 100 0 50 100 150 200 250 300 350

08 [ppbv] PAN[pptv]

12 ß d'
10

: //

•, ''

ß
ß

0 i i i I , , , I , , , I , , ,

40 8o zoo o zoo 400 600 800

CO[ppbv] Ethane[pptv]
Figure3. Medianvertical
profiles
ofmixingratios
for(a)ozone, (b)PAN,(c)CO,and(d)ethaneobservedoverthe
equatorial
andtropicalSouthPacific
(0ø-30øS,
170øE-120øW) during
PEM-TropicsA, overthetropical
north-
westernPacific(0ø-ISøN,120øE-170øE)
duringPEM-WestA, andoverthetropicalSouthAtlantic(0ø-25øS,
30øW-10øE)during TRACEA. Alsoshown (dashed lines)aremonthly
mean September
valuesfroma global3-D
modelsimulation
[Wanget al., 1998a]forthePEM-Tropics
A region.ErrorbarsforthePEM-Tropics
A observa-
tionsdenote
thecentral
50%rangeof thedata,andthegreyshadingrepresents
thecentral90% rangeof monthly
mean3-D modelconcentrations
for theensemble
of gridboxesovertheequatorial
andtropicalSouthPacific.

We correctedthe rates for productionand loss of PAN at
everytimestepasdescribed by [Jacobet al., 1996].
Thermaldecomposition of PAN is the dominantproduction
termfor NOx below 4 km, closely balancingthe loss of NOx
(Figure6). We conclude
thatPAN provides
themainsourceof
NOx at thesealtitudes.Long-rangetransportof biomassburn-
ing pollutionwasa majorsourceof PAN overthe tropical 7. Ozone Formation
SouthPacificduring PEM-TropicsA, as shown in Figure 7
by thecorrelation
betweenPAN andC2H2at 4-8 km.
Above 4 km we find that the chemical source of NOx is in-
sufficientto balancethe loss of NOx fromoxidation, implying
a largemissingsourceforNOx. Sucha chemicalimbalance in
the uppertroposphere has been foundin previousmissions
[e.g.,Davis et al., 1996a;Jacobet al., 1996].Recentstudies
by Jaegldet al. [1998],Pratherand Jacob[1997], and Wang
et al. [1998b]suggest thatthis imbalanceis mostlycausedby
convectiveinjection and primary productionof NOx from
lightning.However,Wanget al. [1998b] find evidenceof a
missingchemicalsourceof NOx in tropical regionswhere
biomassburning is of importance.Further analysisof the
PEM-TropicsA data(M. Schultzet al., Reconciliationof the
chemicalbudget of NOx in the upper troposphereover the
tropicalSouthPacific,submittedto J. Geophys.Res., 1998)
suggests that largeamountsof ammonia emittedby biomass
burningmayslow downthe lossof NOx by dehydrating the
aerosol,while alsoprovidingan additionalchemicalsourceof
NOx, thuscorrectingmostof the NOx budgetimbalanceappar-
ent from Figure 6.
and Loss Rates
Figure8 displays
medianproduction
andlossratesof odd-
oxygen(Ox= 03 + O + NO2 + HNO4 + 2oNO3+ 3oN205+
HNO3 + PAN + PPN) asa functionof altitudefor the equato-
rial andtropicalSouthPacific.The chemical
budgetanalysis
mustbe appliedto Ox,ratherthanto 03, becauseof the fast
chemicalcyclingwithinthe Oxfamily.Considering that ozone
makesupmorethan90% of Ox forthe PEM-TropicsA condi-
tions,we will use both termsinterchangeably.
During PEM-
TropicsA thelower4 km of the troposphereoverthe tropical
SouthPacificexhibitnetdestruction
ratesof about2 ppbvd'•
forozone,whilethereis about1 ppbvd'• net production
above 8 km. The crossover at 6 km is similar to results from
PEM-WestA andB [Daviset al., 1996a;Crawfor& 1997].
Thegeographical
distributionof ozoneproduction(Plate
1) showshighestvaluesover the westerntropicalPacific
 21562202d, 1999, D5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/98JD02309 by University Estadual De Campina, Wiley Online Library on [23/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
5836 SCHULTZ ET AL.: ORIGIN OF TROPOSPHERIC OZONE AND NOx

0.35-0.7 Dobson units (DU) per day. This result does not
10- l•km
changesignificantly when we extrapolatethe data to the tro-
popauseheightof about 16 km (Table 3).
Above 4 km the ratesof ozoneproductionand lossover the
South Pacific are similar to those computedfor the northern
8-10km- tropical Pacific during PEM-West A (Figure 8). Comparedto
the troposphere over the tropical South Atlantic during
TRACE A, productionand loss ratesover the tropical Pacific
are a factor of 2-5 lower.
6-8km -
Our analysis of the photochemicalbudget for hydrogen
peroxides(section5) revealedsomeuncertaintyregardingthe
HOx budgetin the upper troposphere.In orderto investigate
4-6km - the implicationson the calculatedbudget of ozone, we per-
formeda seriesof sensitivity simulations,using as referencea
simulation constrained by median observed values in the
2-4km
tropicallatitude band for the altitudebins of 8-10 km and 10-
12 km (Table lb). In these calculations,water vapor was var-
ied by a factor of 2, the rate coefficientfor the HO2 + CH302
reactionwas varied by a factor of 3, and the acetoneconcentra-

O-2km
t 0.2 0.5 1.0 2.0
I I I

5.0
tion was varied from 400 to 50 pptv. From the combinationof
effects the maximum deviations for Po3 and Lo3 were ñ35%. The
deviationof (P - L)o3 was smaller(-22 to +9%).
Let us now investigatewhether the advectionof continen-
tal air canbalancethe large observedozone loss in the South
Figure 4. Modeled versus observedNO2/NO ratios in PEM- Pacifictroposphere.Layers of continentalair sampledduring
TropicsA (0ø-30øS)as functionof altitude.For explanationof PEM-Tropics A containedelevated CO due to pervasivebio-
the boxesand whiskers, see caption of Figure 2. Only data massburning on the south tropical continents. The mean
with NO2photolysis
frequencies
greater
than7 x 10'3 s': are enhancementratio AO3/ACO in these layers was 1.8 mol/mol
displayed. (d. Logan, personalcommunication,1997), reflectingthe fact
that NOx fi'omcontinental sources(including biomassburn-
ing, lightning, and soil emissions) allowed considerable
where biomassburning influenceis strongest(Figure 2). As
ozone production. Let us take AO3/ACO = 1.8 mol/mol as a
shownin Table 3, chemicalproductionof ozonebalancesonly
about half of the chemical loss for the 0-12 km column over the general characteristic of continental air advected over the
South Pacific; then
South Pacific; the column-integratednet loss amounts to
about1-2 x 10ll molecules
cm'2 s': whichis equivalent
to VFo3= VFco(AO3/ACO) (1)

I I I I I I II I I I I I I I

10-12 b

8-10

6-8-

4-6-

I
2-4-

0-2
t , o.1 0.2 0.5 1.0 2.0 5.0 10.œ
i iiii•:•11•1
0.2
i iiiiiii
0.5 1.0 2.0 5.0 10.0

L(H•0•)/P(H•0•) L(CHaO
0H)/P (CHa0
0H)

Figure 5. Ratioof 24-houraveragepointmodelcalculatedchemicallossto chemicalproductionratesfor (a) H202

and(b) CH3OOHin PEM-Tropics A (0ø-30øS)
as a functionof altitude.Seecaptionof Figure2 forexplanation
of
boxes and whiskers.
 21562202d, 1999, D5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/98JD02309 by University Estadual De Campina, Wiley Online Library on [23/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SCHULTZ ET AL.' ORIGIN OF TROPOSPHERIC OZONE AND NOx 5837

Table2. Median24-HourAverageLifetimesfor OzoneandNOx in PEM-TropicsA

Altitude Band, km

0-2 2-4 4-6 6-8 8-10 10-12

Ozonelifetime, days

0ø-10øS 5 7 31 31 110 147

10øS-30øS 11 27 48 66 110 201

NOx lifetime,days
0 ø- 10øS 0.9 1.0 1.4 1.8 5.0 7.4

10øS-30øS 0.7 1.0 1.1 2.1 4.2 5.4

where VFo3and VFco are the flux divergencesof 03 and CO modelanalysis,

(L - P)o3= 0.9 x l0 ll molecules
cm'2 s-1,we
over the South Pacific,that is, the net influx into the region. concludethat advectionof continentalair canreadily account
The net influx of CO is exclusivelyfrom continentalpollution. for the net influx of tropospheric
ozoneoverthe tropicalSouth
We write a massbalance equation for CO in the tropospheric Pacific. The AO3/ACO ratio used in our calculation is cer-
column over the South Pacific, tainly too high becausethe aircraftundersampled
continental
VFco+ (P- L)co= 0 (2) air advected in the lower tropospherefrom South America,
which had a low AO3/ACO ratio due to fast chemical loss of
where (P - L)co is the net production of CO over the South ozone. A mean AO3/ACO ratio of 0.6 mol/mol for continental
Pacific as determinedfromour photochemicalmodel analysis air advected over the South Pacific would close the ozone
(Table 3). In equation (2) we have ignored the CO tendency budgetand is in the rangeof expectedvalues.
term,which is negligibly small.Equation (2)yields VFco = Our analysis above indicatesthat ozone production over
1.6x l0ll molecules cm-2s'l (Table3, resultsfor0-16km);re- the SouthPacific drivenby influx of PAN and direct influx of
placinginto equation(1), we obtainVFo3= 2.9 x 10TM ozonefromtropical continentalregionsboth provided major
molecules
cm-2s-1.Comparingto thenetcolumn
lossofozone sourcesfor ozone in the South Pacific troposphereduring
over the South Pacific computed from our photochemical PEM-TropicsA. It appearslikely that biomassburningmade a

10-12km - 10-12km -

o_oos 10øS-30øS

8-10km - 8-10km -

E 6-Skm - E 6-Skm-

.,•
• 4-6km - • 4-6km -

2-4km - 2-4km -

0-2km - 0-2km -

I I I I

0 10 20 30 0 10 20 30 40 50

N0x productionand N0xproductionand

loss rate [pptv/day] loss rate [pptv/day]

"-'"•'-•
PAN+T "::':•
''-" PAN+hv •//• HNOa+hv•x• HN0•+0H•.._•CH•0N0•_+hv
I N0,+CH,C0,
• N0,+0H •• N•O?aerosol*•"• NOa+aerosol
Figure 6. Mean chemicalproduction
andlossratesof NO.•calculated
overthe equatorialand tropical SouthPacific
duringPEM-TropicsA: (a) 0ø-10øSand (b) 10øS-30øS.Valuesare24-hour averages.The modelwas constrained
with concurrent observations of NO, HNO•, and PAN. Rates of interconversion between NOx and PAN are cor-
rectedfor PAN-CH•COO2 cyclingas describedin the text.
 21562202d, 1999, D5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/98JD02309 by University Estadual De Campina, Wiley Online Library on [23/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
5838 SCHULTZ ET AL.: ORIGIN OF TROPOSPHERIC OZONE AND NOx

400

3OO

100
15 20 25

PO
s [105molec./cmSs]
0
0 100 200 300 12 •.... I .... I .... I .... I ' 'b'
CzHz[pptv] lO \
,

Figure 7. Correlation between PAN and acetylene in the

'\
tropical South Pacific (10øS-30øS)during PEM-Tropics A at
4-8 km altitude.

dominant contribution to both of these sources. Ozone and

PAN concentrationsmeasuredin the free troposphere over
Brazil andAfrica during TRACE A were considerablyhigher
than in PEM-Tropics A (Figure 3) and were due unambigu-
ously to biomassburning [Singh et al., 1996b; Thompsonet o 5 10 15 20 25

al., 1996]. Zonal transport at both tropical and subtropical LO

a [105molec./cmas]
latitudes[Fuelberg et al., this issue] would supply this bio-
i ' ' '
mass burning ozone and PAN to the South Pacific
12
troposphere.As mentionedpreviously,PAN in PEM-Tropics
A was correlatedpositively with tracersof biomassburning
10
(Figure 7). Although lightning could also have contributed
to long-rangetransport of NOx over the Pacific, year-round
ozonesondemeasurements at South Pacific sites (S. Oltmans
et al., personal communication,1997) argue against a major
contribution of lightning to ozone in PEM-Tropics A. The
ozonesonde data show a pronounced seasonal maximumof
ozone in austral spring, during the biomassburning season, ß PEM-TropicsA
consistentwith ozonesonde data in the TRACE A region [Ol- [] PEM-West A
son et al., 1996]. In contrast, lightning peaks in austral A Trace A
summer[Turinartand Edgar, 1982].
-4 -2 0 2 4

8. Global 3-D Model Simulation (P-L)Oa [ppbv/day]

In orderto provide an additional perspectiveon the large-
scaleregionalbudgetof ozoneover the tropical South Pacific,
we analyzed results from the Harvard/Goddard Institute for
SpaceStudies(GISS) global 3-D model[Wang et al., 1998a].
The model uses meteorologicalfields from the GISS Global
Climate Model (GCM) and has 4ø x 5ø horizontal resolution
with nine vertical layers, seven of which are in the tropo-
sphere.A suite of 15 chemical tracers including 03, NOx,
PAN, HNO3, CO, H202, and five primary hydrocarbonsis
simulated,with parameterizations for tracer production and
loss derived from point model simulations with a detailed
mechanismover a wide range of conditions. Fossil fuel com-
bustion, woodburning, and biomassburning emissions are
from J. Logan (manuscript in preparation, 1998). Biogenic
Figure 8. Median profilesof (a) the chemicalproductionrate,
(b) the chemicallossrate, and (c) the net productionrate(P-L)
of ozoneover the equatorialand tropicalSouthPacificduring
PEM-TropicsA. Values for the northwesterntropicalNorth
Pacific duringPEM-West A [Davis et al., 1996a] andthe
tropicalSouthAtlantic duringTRACE A [Jacobet al., 1996]
are shownfor comparison.All valuesare 24-houraverages.
See captionin Figure 3 for regionboundaries.
emissions are parameterized
similar to Guentheret al. [1995],
and soil emissionsare formulatedaccordingto Yiengerand
Levy [1995] and Jacob and Bakwin [1991]. The model has
beenextensivelyevaluatedwith observations[Wang et al.,
1998b].
 21562202d, 1999, D5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/98JD02309 by University Estadual De Campina, Wiley Online Library on [23/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SCHULTZ ET AL.: ORIGIN OF TROPOSPHERIC OZONE AND NOx 5839

15øS

30øS

4.5- 13 km

. 15øN

.... 15%
.

0-4.5 km
,45øS

150øW 120øW 90øW

150øE 180 150"W 120"W 90"W

0 2 4 6 > ppbv/day
0 2 4 6 ppbv/doy Plate 2. Mean ozone productionrates in Septembercalculated
by the Harvard/GISSglobal 3-D model.
Plate 1. Geographical distribution of chemical production
rates of ozone fromthe PEM-Tropics A data at 0-4.5 km and
4.5-12
km.Values
are24-hour
averages
andaredisplayed
ona around
6km.Thesimulated
vertical
profile
ofCOagrees
with
2øx2.5øgrid. theobservations
within5 ppbv.Asin thecase
ofozone,
the
variability of the observationsis significantly larger than the
variability in the 3-D simulation below 4 km. The vertical
Figure 3 comparesmodel results (dashedlines) with the profile of PAN is also predicted within one standard devia-
PEM-Tropics A observations(solid circles) for median con- tion of the observations; the model concentrations are about
centrationsof ozone, CO, and PAN. Ozone concentrations in 20% higher.
the 3-D modelare lower thanthe observations below 3 km by The ratesof photochemicalozoneformationand loss calcu-
7 ppbv (30%), and they show a smallerenhancement near 6 km lated with the global modelagreewith those fromthe point
(the 25 percentileof the observationsis closeto the 95 per- model within 20% (Figure 8). Ozone loss below 2 km and
centile of the model). However, the general tendency is production above 6 km are somewhathigher in the point
reproduced: median concentrations of ozone are largest model.The geographicaldistribution of ozone production

Table3. Median
24-Hour
Average
Column
Budgets
ef OxandCOinPEM-Tropics
A

P03 L03 (P-L)03 Pco Lco (P-L)co

0-12 km column

0%10øS 1.36 3.20 -1.84 1.59 2.66 -1.07

10øS-30øS 1.82 3.05 - 1.23 1.24 2.68 - 1.44

0-I 6 km column

0%10øS 1.72 3.24 -1.52 1.61 2.88 -1.27

10øS-30øS 2.18 3.09 -0.91 1.26 2.90 - 1.64

Column
rates
ofproduction
(P)and
loss
(L)areinunits
ofl0ll molecules
cm '2s'l. At12-16
km,rates
arecomputed
bylinearily
extrapolating
medianconcentrationprofilesabovethe 12km ceilingof theaircraft.Theextrapolated
profilesforNOx,ozone,andCO
agreewithin30% with observations
fromtheStratospheric
Tracersof Atmospheric Transport
(STRAT)campaign in August1996(P.
Wennberg,personal
communication, 1997).
 21562202d, 1999, D5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/98JD02309 by University Estadual De Campina, Wiley Online Library on [23/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
5840 SCHULTZ ET AL.: ORIGIN OF TROPOSPHERIC OZONE AND NOx

rates(Plate 2) is similar in both models.Both show enhanced
ozoneproductionabove 4 km in the westernPacific around
15øS due to advection of continental air. The global model
also predicts enhanced ozone production over the eastern
equatorialPacific (Plate 2), indicating outflow of pollution
fi'omSouth America.Data fi'omthe P-3B aircraftflights along
the westerncoastof SouthAmerica indicate that easterlycon-
tinental outflow can have biomass burning as a major
component(N. Blake, personalcommunication, 1997).
Table 4 lists the individual terms that contribute to the 3-D
modelbudgetfor ozoneover the tropicalSouthPacific in Sep-
tember.The two altitude regimes (0-4.5 km and 4.5-13 km)
reflectthe transition fi'omsoutheasterlytrade winds to pre-
dominantwesterly flow [e.g.,Fuelberg et al., this issue]. In
the lower part of the tropospherethe ozone budget is domi-
nated by advection from the east together with vertical
downward transport, and by photochemicalloss inside the
region. In situ formationof ozone balancesonly one third of
the photochemical loss.Above 4.5 km, ozone is advectedpre-
dominantlyfrom the west;this term is about twice as large as
the advection fromthe east in the lower part. The net down-
wardverticalflux of 0.53 x 10• moleculescm'2 s'• amountsto
about 17% of the lateral outflux above 4.5 km to the east and
south. For the entire 0-13 km column over the South Pacific,
we find in the 3-D model that 55% of the ozone originates
from within the region and 45% is transportedin the flux di-
vergenceterm. This result is consistentwith our point model
analysis, where column-integratedchemicalproduction bal-
anced 50% of the chemical loss (Table 3). Although not
apparentfrom Table 4, a traceranalysisin the 3-D model indi-
catesthat 20% of troposphericozone over the tropical South
Pacific (0ø-30øS) originates from cross-tropopausetransport
in the extratropical Southern Hemisphere [Wang et al.,
1998c].
In order to assessthe impact of biomass burning on the
ozone budget over the tropical South Pacific, we conducteda
3-D model simulation without biomass burning emissions.
Ozone mixing ratios fi'omboth simulations are comparedin
Figure9. Septembermeanozone mixing ratios over the tropi-
cal South Pacific (0ø-28øS; 165øE- 100øW) are on average
higherby 7 and 8 ppbv (---20%)for altitudesbelow and above
4.5 km, respectively.Without biomassburning,influx from the
east at lower altitudesdecreasesby almost a factor of 2, while
the advection fi'omthe west above 4.5 km decreasesby 30%
(Table 4). Photochemicalproduction above 4.5 km now ex-
ceedsloss by almost a factorof 2, but the absolute difference
between these two termsremainsabout the same.In general,
the chemicalterms decreaseproportionally to the advection
terms, supportingour finding that biomassburning pollution
affectsadvection of ozone as well as in situ production over
the South Pacific.
At first sight, thesesimulationssuggestthat biomassburn-
ing in South Americaimpactsozone below 4.5 km, whereas
sourceregions to the west (Africa and South America) con.
tribute to ozone advection and formation above. However, as
discussed in section 6, the abundance of NOx in the lower
troposphereis controlled by thermal decompositionof PAN
which is advectedabove4 km as part of the biomassburning
pollution.Becauseof the warm temperatures and the availabil-
ity of UV radiation, the lifetimes of NOx and PAN are too
short to allow significant transport of these species in the
trade winds at lower altitudes. This implies that the chemical

Table4. September
MeanOzoneBudgetOvertheTropicalSouthPacificfromtheHarvard/GISSGlobal3-D Model.

Process BaseSimulation No BiomassBurning

0-4.5 km 4.5-13 km 0-4.5 km 4.5-13 km

Mean03 mixingratio,ppbv 31 44 24 36

Budget
terms,
1011
molecules
cm'2s'1
Advection from west -0.33 +3.84 -0.26 +2.70

from east + 1.81 -2.51 + 1.03 - 1.85

from south -0.14 -0.92 -0.07 -0.65

from north -0.19 +0.04 -0.08 +0.04

Total(fluxdivergence) + 1.14 +0.46 +0.62 +0.24

Vertical
transport +0.53 -0.49a +0.50 -0.46a
Chemical
production +0.98 +0.92 +0.52 +0.74
Chemical loss -2.61 -0.71 - 1.63 -0.48

Dry deposition -0.27 0 -0.20 0

Residual
b -0.23 +0.18 -0.19 +0.04

The budgetwascomputed for the region0ø-28øS,165øE-100øW.

Fluxesarenormalized
to horizontalsurfacearea.Positive
transporttermsindicatenet inflow.
a Includesdirectinputfromthestratosphere.
bDiffusive
transport
andtendency
term.
 21562202d, 1999, D5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/98JD02309 by University Estadual De Campina, Wiley Online Library on [23/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SCHULTZ ET AL.: ORIGIN OF TROPOSPHERIC OZONE AND NOx 5841

a b

, 50•S

150'"E 180 150"W 120"W 90"W 150øE 180 150"W 120"W 90"W

Figure 9. Mean ozonemixing ratiosin ppbv over the tropical South Pacific for the month of Septembercalculated
by the Harvard/GISS global 3-D model; (a) standardsimulationand (b) simulationwithout biomassburning.

ozone formationin the lower troposphereis controlled by The ozone budget in the lower 4 km of the South Pacific
downward transport of precursorsthat have been advected troposphere is largely affectedby PAN that is advected in
fi'omthe west. This is supported by the differencein subsi- biomassburning pollution layers at higher altitudes.Below 4
dence fluxes of PAN between the two simulations: without km the thermal decompositionof PAN can accountfor most of
biomassburning emissionsthis flux is a factor of 3 lower. the NOx production, and it thereforecontrols chemicalozone
Adding the vertical transport of ozone to the chemicalforma- formation.Subsidence of ozonefrom above and photochemical
tion of ozone in the lower troposphere,we find that the sum is production contribute as much to the ozone budget in the
comparablein magnitudewith the horizontal advection of lower troposphereas advectionof biomassburning pollution
ozone fromthe east (Table 4). Hence biomassburning pollu- from SouthAmericain easterlytradewinds.The comparisonof
tion advected from the west can account for all the biomass a global 3-D simulationwith and without biomassburning
burning ozone enhancementabove 4.5 km and for half of the emissionsshowedthat biomassburning causedan increasein
increaseat lower levels. Biomassburning pollution advected ozone concentrationsover the tropical South Pacific of 7-8
fromthe eastaffectsozoneonly in the lower troposphereover ppbv at all altitudes.
the eastern South Pacific.

Acknowledgments.We would like to thank C. Spivakovskyfor her

9. Summary
The modelanalysisof the PEM-TropicsA datafrom the DC-
8 revealedthe importantcontribution of biomassburning to
the ozone budget over the tropical South Pacific. PEM-
Tropics A took place during the burning season,and it will
be interesting to comparethese results to the data of PEM-
TropicsB which is scheduledfor the rain seasonin February-
March 1999. Both, the point model and the Harvard/GISS
global 3-D model, calculate a photochemicalcolumnnet loss
forozoneof about1.5x 10]] molecules
cm-2s-] duringPEM-
Tropics A. Photochemicalproductionof ozonebalancesabout
half of the chemicalloss;the ozonebudgetcan be closedwith
advection of continental air. The turnover fi'omnet photo-
chemicalloss to net photochemicalproduction takes place
around 6 km, as was previously found for the tropical South
Atlantic in TRACE A and for the tropical North Pacific in
PEM-West A and B.
continuous effortto developandmaintainthe pointmodel,L. Jaeg16for
helpful discussions and the updateof the chemical mechanismof the
pointmodel,and G. Gardnerfor preparingthe mergeddata sets.We
alsothankJ. Crawfordfor providingdatafor PEM-West A, and J. Bar-
rick for helpful discussionson the measurementsof water vapor. The
work of M. Schultzwas fundedby the DeutscheForschungsgemein-
schaft(DFG). PEM-Tropics was funded as part of the NASA GTE
program.
References
Andreae, M.O., et al., Biomass-burning
emissions and associated
haze
layersover Amazonia,J. Geophys.Res.,93, 1509-1527,1988.
Arnold, F., V. BrQger,B. Droste-Fanke,F. Grimm, A. Krieger, J.
Schneider,andT. Stilp,Acetonein the uppertroposphereand lower
stratosphere: Impact on trace gasesand aerosols,Geophys.Res.
Lett., 24, 3017-3020, 1997.
Atkinson,R., D.L. Baulch, R.A. Cox, R.F. Hampson,J.A. Kerr, M.J.
Rossi,andJ. Troe,Evaluatedkineticand photochemicaldata for at-
 21562202d, 1999, D5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/98JD02309 by University Estadual De Campina, Wiley Online Library on [23/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
5842 SCHULTZ ET AL.: ORIGIN OF TROPOSPHERIC OZONE AND NOx

mosphericchemistry- Supplement
VI, J. Phys. Chem.Ref. Data,
26(6), 1329-1499, 1997.
Bradshaw,J.D., M.O. Rodgers,S.T. Sandholm, S. KeSheng,and D.D.
Davis,A two-photonlaser-induced fluorescence
field instrument
for
ground-based and airbornemeasurements of atmospheric NO, J.
Geophys.Res.,90, 12,861-12,873,1985.
Carroll,M.A., andA.M. Thompson,NOx in the non-urbantroposphere,
in Progressand Problemsin Atmospheric Chemistry,editedby J.
Barker,World Sci., River Edge,N.J., 1995.
Chameides,W.L., et al., Observedandmodel-calculatedNO2/NO ratios
in tropospheric air sampledduringthe NASA GTE/CITE 2 field
study,J. Geophys.Res.,95, 10,235-10,247,1990.
Chameides,W.L., et al., Ozone precursorrelationshipsin the ambient
atmosphere, J. Geophys.Res.,97, 6037-6055,1992.
Chatfield,R.B., AnomalousHNO3/NOxratioof remotetroposphericair:
Conversionof nitric acid to formic acid and NOx?, Geophys.Res.
Lett., 21, 2705-2708, 1994.
Cohan, D.S., M.G. Schultz,D.J. Jacob,B.G. Heikes, and D.R. Blake,
Convectiveinjectionand photochemicaldecay of peroxidesin the
Harriss,R. C., et al., The Amazon BoundaryLayer Experiment(ABLE
2A): Dry season1985,d. Geophys.Res.,93, 1351-1360,1988.
Harriss,R.C., et al., The Amazon BoundaryLayer Experiment:Wet
season1987,d. Geophys.Res.,95, 16,721-16,736,1990.
Hauglustaine,D. A., B. A. Ridley, S. Solomon,P. G. Hess, and S.
Madronich, HNO3/NOx ratio in the remote troposphereduring
MLOPEX 2: Evidence for nitric acid reduction on carbonaceous
aerosols,Geophys.Res.Lett., 23, 2609-2612, 1996.
Hoell, J.M., D.D. Davis, S.C. Liu, R.E. Newell, M. Shipham,H. Aki-
moto, R.J. McNeal, R.J. Bendura, and J. W. Drewry, Pacific
Exploratory Mission-WestA (PEM-West A): September-October
1991,•. Geophys.Res., 101, 1641-1653,1996.
Hoell, J.M., D.D. Davis, S.C. Liu, R.E. Newell, H. Akimoto, R.J.
McNeal, and R.J. Bendura,The Pacific ExploratoryMission-West
PhaseB: February-March 1994, d. Geophys.Res., 102, 28,223-
28,240, 1997.
Hoell, J.M., et al., Pacific ExploratoryMissionin the tropicalPacific:
PEM-TropicsA, August-September 1996,d. Geophys.Res., this is-
sue.

tropicaluppertroposphere: Methyl iodideas a tracerof marinecon- Jacob,D.J., Heterogeneous chemistryand troposphericozone,Atmos.

vection,J. Geophys.Res.,this issue. Environ.,in press,1998.
Crawford, J., An analysisof the photochemicalenvironmentover the Jacob,D.J., andP.S.Bakwin,Cyclingof NOxin tropicalforestcanopies
western, North Pacific based on airborne field observations,Ph.D. anditsimplications for the globalsourceof biogenicNOx to the at-
thesis,GeorgiaInst. of Technol.,Atlanta, 1997. mosphere,in MicrobiologicalProductionand Consumptionof
Crawford,J., et al., Photostationary
stateanalysisof the NO2-NO system Greenhouse Gases,editedby W.B. Whitman,pp. 237-253,Am. Soc.
based on airborne observations from the western and central North of Microbiol.,Washington, D.C., 1991.
Pacific,J. Geophys.Res.,101, 2053-2072, 1996. Jacob,D.J., andS.C.Wofsy,Photochemistry of biogenicemissionsover
Crawford,J.H., et al., Implicationsof large scaleshiftsin tropospheric theAmazonforest,•. Geophys.Res.,93, 1477-1486,1988.
NOx levels in the remote tropical Pacific, J. Geophys.Res., 102, Jacob,D.J., and S.C. Wofsy, Budgetsof reactive nitrogen,hydrocar-
28,468-28,477, 1997. bons,and ozoneover the Amazonforestduringthe wet season,•.
Davis,D.D., et al., A photostationarystateanalysisof the NO2-NO sys- Geophys.Res.,95, 16,737-16,744,1990.
tem based on airborne observationsfrom the subtropical/tropical Jacob,D.J., et al., Originof ozoneandNOx in the tropicaltroposphere:
North and SouthAtlantic,J. Geophys.Res.,98, 23,501-23,523, 1993. A photochemicalanalysisof aircraft observationsover the South
Davis,D.D., et al., Assessment of ozone photochemistry in the western Atlanticbasin,d. Geophys.Res.,101, 24,235-24,250,1996.
North Pacificas inferredfrom PEM-WestA observations duringthe Jaeg16,L., et al., ObservedOH and HO2 in the uppertroposphere sug-
fall 1991,J. Geophys.Res.,101, 2111-2134, 1996a. gest a major source from convectiveinjection of peroxides,
Davis, D.D., J. Crawford,S. Liu, S. McKeen,A. Bandy, D. Thornton,F. Geophys.Res.Lett., 24, 3181-3184, 1997.
Rowland, and D. Blake, Potentialimpactof iodine on tropospheric Jaeg16,L., D.J. Jacob,Y. Wang, A.J. Weinheimer, B.A. Ridley, T.L.
levels of ozone and other critical oxidants,d. Geophys.Res., 101, Campos,G.W. Sachse,and D. Hagen,Originof NOx in the upper
2135-2147, 1996b. troposphere over the centralUnitedStates,Geophys.Res.Lett.,25,
DeMore, W.B., S.P. Sander,D.M. Golden,R.F. Hampson,M.J. Kurylo, 1709-1712, 1998.
C.J. Howard, A.R. Ravishankara, C.E. Kolb, and M.J. Molina, Kasibhatla,P.S., H. Levy, W.J. Moxim, and W.L. Chameides,The rela-
Chemicalkineticsand photochemicaldata for use in stratospheric tive importanceof stratosphericphotochemicalproductionon
modeling,JPL Publ.,97-4, NASA Jet Propol.Lab., Pasadena,Calif., tropospheric NOy levels:A model study,d. Geophys.Res., 96,
1997. 18,631-18,646, 1991.
Drummond,J.W., D.H. Ehhalt, and A. Volz, Measurementsof nitric Kirchhoff,V.W.J.H.,R.A. Barnes,andA.L. Torres,Ozoneclimatology
oxide between 0-12 km altitude and 67øN to 60øS latitude obtained at NatalBrazil,from in situozonesonde data,•. Geophys.Res.,96,
duringSTRATOZ III, d. Geophys.Res.,92, 15,831-15,849,1988. 10,899-10,909, 1991.
Emmons, L.K., et al., Climatologies
of NOx andNOy: A comparison of Lamarque, J.-F., G.P. Brasscur, P.G. Hess, and J.-F. Muller, Three-
data and models,Atmos.Environ., 31, 1851-1903, 1997. dimensional studyof the relativecontributions
of the differentnitro-
Fan, S.-M., D.J. Jacob,D.L. Mauzerall, J.D. Bradshaw, S.T. Sandholm, gen sourcesin the troposphere,d. Geophys.Res., 101, 22,955-
D.R. Blake, H.B. Singh,R.W. Talbot, G.L. Gregory, and G.W. Sa- 22,968, 1996.
chse,Origin of troposphericNOx over subarcticeasternCanadain Lary, D. J., A.M. Lee, R. Toumi,M. J. Newchurch,M. Pirre, and J. B.
summer,J. Geophys.Res.,99, 16,867-16,877,1994. Renard,Carbonaerosolsand atmospheric photochemistry. d. Geo-
Fishman,J., andV.G. Brackett,The climatologicaldistribution of tropo- phys.Res., 102, 3671-3682, 1997.
sphericozone derived from satellitemeasurements usingversion 7 Liu, S.C.,D. Kley, M. McFarland,J.D. Mahlman,and H. Levy II, On
Total Ozone MappingSpectrometer and StratosphericAerosol and the originof tropospheric ozone,d. Geophys.Res.,85, 7546-7552,
Gas Experimentdata sets,J. Geophys.Res., 102, 19,275-19,278, 1980.
1997. Logan,J.A.,M.J. Prather,S.C.Wofsy,andM.B. McElroy,Tropospheric
Fishman,J., C.E. Watson,J.C. Larsen,and J.A. Logan, Distributionof chemistry:A globalperspective, •. Geophys.Res., 86, 7210-7254,
tropospheric ozonedeterminedfrom satellitedata,d. Geophys.Res., 1981.
95, 3599-3617, 1990. McKeen, S.A., T. Gierczak, J.B. Burkholder,P.O. Wennberg,T.F.
Fishman,J., K. Fakhruzzaman,B. Cros, and D. Nganga, Identification Hanisco,E.R. Keim, R.-S. Gao, S.C. Liu, A.R. Ravishankara,and
of widespreadpollutionin the SouthernHemispherededucedfrom D.W Fahey,The photochemistry of acetonein the upper tropo-
satelliteanalysis,Science,252, 1693-1696,1991. sphere: A sourceof odd-hydrogen radicals,Geophys.Res.Lett.,24,
Fishman, J., J.M. Hoell, R.D. Bendura, R.J. McNeal, and V.W.J.H. 3177-3180, 1997.
Kirchhoff, NASA GTE TRACE A Experiment(September-October Olson,J., J. Fishman,V.W.H.J. Kirchhoff,D. Nganga,and B. Cros,
1992):Overview,d. Geophys.Res.,101, 23,865-23,879,1996. Analysisof the distributionof ozone over the southernAtlantic re-
Fuelberg,H.E., R.E. Newell, S.P. Longmore, Y. Zhu, D.J. Westberg, gion,•. Geophys.Res.,101, 24,083-24,094,1996.
E.V. Browell, D.R. Blake, G.R. Gregory,and G.W. Sachse,A mete- Penner,J.E., C.S. Atherton,J. Dignon, S.J. Ghan, J.J. Walton, and S.
orologicaloverview of the PEM-Tropicsperiod,J. Geophys.Res., Hameed,Troposphericnitrogen:A three-dimensional studyof
this issue. sources,distributions,
anddeposition,d. Geophys.Res.,96, 959-990,
Gregory, G.L., et al., Chemicalcharacteristicsof Pacific tropospheric 1991.
air in the regionof the ITCZ andSPCZ,J. Geophys.Res.,thisissue. Prather,M.J., and D.J. Jacob,A persistentimbalancein HOx and NOx
Guenther,A., et al., A globalmodelof natural volatile organic com- photochemistry of the uppertroposphere drivenby deeptropical
poundemissions, d. Geophys.Res.,100, 8873-8892,1995. convection, Geophys.Res.Lett.,24, 3189-3192,1997.
 21562202d, 1999, D5, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/98JD02309 by University Estadual De Campina, Wiley Online Library on [23/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
SCHULTZ ET AL.: ORIGIN OF TROPOSPHERIC OZONE AND NOx 5843

Sandholm,S.T., J.D. Bradshaw,K.S. Dorris, M.O. Rodgers,and D.D. Wang, Y., D.J. Jacob,andJ.A. Logan,Globalsimulationof tropospheric
Davis, An airborne compatiblephotofragmentation two-photonla- O3-NO.•-hydrocarbon chemistry,1, Model formulation,Or.Geophys.
ser-induced fluorescence instrument for measuring background Res., 103, 10,713-10,726 1998a.
troposphericlevelsof NO, NOx, and NO2, Or. Geophys.Res., 95, Wang, Y., J.A. Logan,andD.J. Jacob,Global simulationof tropospheric
10,155-10,161, 1990. O3-NOx-hydrocarbonchemistry,2, Model evaluation and global
Schultz,M., Die Bedeutungvon Stickoxidenfar die Ozonbilanzin Re- ozonebudget,Or.Geophys.Res.,103, 10,727-10,756,1998b.
inluftgebieten,Ph.D. thesis, Univ. of Wuppertal, Wuppertal, Wang,Y., D.J. Jacob,andJ.A. Logan,Globalsimulationof tropospheric
Germany, 1995. O3-NOx-hydrocarbonchemistry,3, Origin of troposphericozone
Singh,H.B., Reactivenitrogenin the troposphere, Environ.Sci. Tech- and effects of nonmethanehydrocarbons,Or.Geophy,•.Res., 103,
nol., 21,320-327, 1987. 10,757-10,768, 1998c.
Singh,H.B., M. Kanakidou, P.J.Crutzen,andD.J. Jacob,High concen- Watson,C.E., J. Fishman,and H.G. Reichle,The significanceof bio-
trationsand photochemical fate of oxygenatedhydrocarbons in the massburning as a source of carbon monoxideand ozone in the
globaltroposphere, Nature,378, 50-54, 1995. SouthernHemispheretropics:A satelliteanalysis,Or.Geophys.Res.,
Singh,H.B., et al., Low ozonein themarineboundary layerof the tropi- 95, 16,443-16,450, 1990.
cal Pacific Ocean: Photochemical loss, chlorine atoms, and Yienger,J. J., and H. Levy II, Empiricalmodelof globalsoil-biogenic
entrainment, Or.Geophys. Res.,101, 1907-1917,1996a. NOx emissions,or.Geophys.Res.,100, 11,447-11,464,1995.
Singh,H.B., et al.,Impactof biomass burningemissions on the composi-
tion of theSouthAtlantictroposphere: Reactivenitrogenand ozone,
E. Atlas, F. Flocke, and R.E. Shetter,National Center for Atmos-
Or.Geophys. Res.,101, 24,203-24,220,1996b.
phericResearch,Boulder,CO 80307.
Smyth,S.B.,et al., Factorsinfluencing the upperfree troposphericdis-
D.R. Blake and N.J. Blake, Departmentof Chemistry,Universityof
tributionof reactive nitrogenover the SouthAtlantic during the
California, Irvine, CA 92717.
TRACE A experiment, Or.Geophys. Res.,101,24,165-24,186,1996.
E.V. Browell, M.A. Fenn, and G.L. Gregory, NASA Langley Re-
Talbot,R.W., et. al., Influence of biomasscombustion emissionson the
distributionof acidic trace gasesover the southernPacific basin searchCenter,Hampton,VA 23681.
B.G. Heikes and G.W. Sachse,GraduateSchoolof Oceanography,
duringaustralspringtime, Or.Geophys. Res.,thisissue.
Thompson, A.M., The oxidizingcapacityof the Earth's atmosphere: Universityof RhodeIsland,Narragansett, RI 02882.
D.J. Jacob,J.A. Logan, and M.G. Schultz,Divisionof Engineering
Probablepastandfuturechanges, Science,256, 1157-1165,1992.
andAppliedSciencesandDepartmentof Earth and PlanetarySciences,
Thompson, A.M., K.E. Pickering,D.P. McNamara, M.R. Schoeberl,
R.D. Hudson,J.H. Kim, E.V. Browell, V.W.J.H. Kirchhoff, and D. Harvard University,29 Oxford St., Cambridge,MA 02138. (e-mail:
Nganga,Where didtropospheric ozoneover southern Africa and mgs@io.harvard.edu)
S.T. SandholmandY. Wang, Schoolof Earth and AtmosphericSci-
the tropicalAtlanticcome from in October 19927 Insightsfrom
TOMS, GTE TRACE A, and SAFARI 1992,Or.Geophys.Res., 101, ences,GeorgiaInstituteof Technology,Atlanta,GA 30332.
H.B. Singh,NASA AmesResearch Center,MoffettField,CA 94035.
24,251-24,278, 1996.
R.W. Talbot, Institutefor the Study of Earth, Oceans, and Space,
Turman,B.N., and B.C. Edgar, Global lightningdistributions
at dawn
Universityof New Hampshire,Durham,NH 03824.
anddusk,Or.Geophys.Res.,87, 1191-1206,1982.
Vogt,R., P.J.Crutzen,andR. Sander,
A mechanism
for halogenrelease
from sea-saltaerosolin the remote marine boundarylayer, Nature, (ReceivedOctober13, 1997;revisedJune9, 1998;accepted
July9,
383, 327-330, 1996. 1998.)