Feb 16, 2000 - quantify the photochemistry of tropospheric O 3 and compare the main source categories. We simulated a 15 ..... sphere where they dissipate, which exerts a drag force onto the stratospheric ...... Thunder- storm convection ...
JOURNAL OF GEOPHYSICAL
RESEARCH, VOL. 105, NO. D3, PAGES 3531-3551, FEBRUARY
16, 2000
What controls tropospheric ozone? Jos Lelieveld
and Frank
J. Dentener
Institute for Marine and AtmosphericResearch,Utrecht University,Utrecht, Netherlands
Abstract. We have applied a global three-dimensionalchemistry-transport model to quantifythe photochemistryof troposphericO3 and comparethe main sourcecategories. We simulateda 15 year period (1979-1993) on the basisof the European Centre for Medium-Range Weather Forecastsmeteorologicalreanalysesand a time-varyingemission data set. We calculatethat stratosphere-troposphere exchange(STE) stronglycontributes to 0 3 in regionswhere the photochemistryis quiescent.Sincesuchregionsplay a minor role in radiativeand chemicalprocesses, we arguethat STE-derived0 3 is muchlessimportant than is suggested by its columnabundance.By distinguishing betweenphotochemicalpathways in the model we calculatethat tropospheric0 3 in the extratropicalNorthern Hemisphere is stronglyaffectedby industrialand fossilfuel-related emissions.In the tropicsand SouthernHemisphere,natural emissionsstill play a major role. Our model resultsindicate a lessimportant role for man-madebiomassburning emissionsthan previousanalyses. Further, the resultsshowthat tropospheric03 trends are stronglyinfluencedby transports of pollution and by meteorologicalvariability. Scenariocalculationsfor the year 2025 suggestthat man-madeemissionsat low northern latitudes,in particularin southernand easternAsia, will becomea very strongtropospheric0 3 sourcein the next decades.This will influence0 3 levelson a hemisphericscaleso that despitepollution regulationsin Europe and North America, surface0 3 will continueto grow. 1.
Introduction
(R5)
HO2 + NO-•
NO2 + OH
About 90% of atmosphericozone is present in the strato- (R6) 2[NO2 + h v(+O2) --• NO + Os] ,k < 420 nm sphere,and only 10% is presentin the troposphere.Despite this relativelysmall fraction, troposphericozone governsoxi- Net dation processesin the Earth's atmospherethrough the forRH + h•; + 302 ----> 203 + H:O + CARB mation of hydroxyl(OH) radicals.OH, which controls the atmosphericlifetime of many gases,is formed by photodissoHydroxyl radicalsfrom the reactions(R1)-(R2) initiate the ciation of 03 in the presenceof water vapor [Levy, 1971]: reactionsequence(R3)-(R6) in which 0 3 is producedthrough (R1) O3+ h•--• O('D) + O: ,k< 330 nm NO-to-NO2 conversionand OH is regenerated.Photodissociation of carbonylcompounds(CARB), in particularformalde(R2) O('D) + H:O -• 2OH, hyde, yields CO and causesadditional ozone formation. To some extent the above radical reaction chain can be where h v is the product of the Planck constantand the frecomparedwith a combustionprocessfor which RH and CO quencyof light at wavelength Traditionally,it wasassumedthat troposphericozoneis con- emissionssupply the "fuel" [Johnstonand Kinnison, 1998]. trolled by stratosphere-troposphere exchange(STE) acrossthe Some differences,however,need to be emphasized.Combusextratropical tropopause [Regener, 1957; Junge, 1962; tion itself releases the heat to overcome reaction activation Danielsen,1968;D•tsch, 1971].This first analysiswasbasedon energies,so that spontaneousreactionsmaintain the process. the observed03 gradient with altitude, suggestinga sourceat In the atmospheresolar photonsmust provide additional enthe tropopauseand a sink at the surface.In the 1960s,in situ ergy. At night the sequenceceases.Furthermore, NOx emisphotochemicalozone formation in the tropospheredrew at- sionsplay a key role. If only the fuel substancesRH and CO tention as it was shownthat the breakdownof hydrocarbons are added to the troposphericreaction mixture without NOx, can cause03 episodesin urban environmentsduring summer the HO 2 radicals formed destroy 03, or they recombine into [Haagen-Smitand Fox, 1956;Leighton,1961]. The in situ 0 3 peroxides.Hence, without NO• the reaction chain resembles formationis catalyzedby nitrogenoxides(NOx = NO + NO2), decayingcombustion.If this mechanismwould prevail in the often emittedsimultaneously with hydrocarbons (RH) andcar- troposphere,the removal of RH, CO, and other pollutants bon monoxide(CO), in particularby man-madesources(Table through OH attack would be fully dependenton 0 3 transport from the stratosphere. 1). The main RH oxidationsequencein the atmosphereis In the early 1970sit was suggestedthat photo-oxidationof (R3) RH + OH (+O:)-• RO: + H:O the simplestand most abundant of all hydrocarbons,methane (CH4) , and CO can cause03 formation in large areas of the (R4) RO2 + NO (+02)-• NO2 + HO2 + CARB troposphere.It waspredictedthat net troposphericozone proCopyright2000 by the American GeophysicalUnion. ductionpredominatesin NO•-rich air, in particular,over much of the continental Northern Hemisphere, while destruction Paper number 1999JD901011. 0148-0227/00/1999JD901011 $09.00 prevails in NO•-deficient air [Crutzen, 1973]. In subsequent 3531
3532
LELIEVELD
AND
DENTENER:
WHAT
CONTROLS
TROPOSPHERIC
OZONE?
Table 1. EstimatedAnnual 0 3 PrecursorEmissionsfor the Preindustrial(1860), Recent (1993), and Future (2025) Period CO, TgCyr-1 SourceCategory
1860
NMHC, Tg yr-1
1993
2025
112
142
1860
1993
NOx,TgN yr-1 2025
1860
1993
2025
0.3
24.4
41.1
0.4 0
1.3 0.5
1.3 1.6
0
1.5
2.8
0.9 0.2 1.6 0.9 0 3 5
3.1 1.1 0.8 2.2 2.2 3 5
3.6 1.6 0.9 3.9 4.5 3 5
0.6
0.6
0.6
Energy use fossil fuel combustion
2
fossilfuel production biofuel aircraft
combustion
Industrial processes Biomassburning savannahburning tropical deforestation temperatewildfires agriculturalwasteburning Agricultural soils Natural vegetation/soils Lightning
22
1
37
67
0
26
65
83
83
8
32
32
6
15
18
0
56
102
24 8 90 36
77 48 46 89
95 71 50 156
5 1 7 5
15 8 4 16
17 12 4 19
115
115
115
403
403
403
205
161
165
410
407
407
424
565
190
314
585
730
597
721
NOy fromstratosphere Natural
Anthropogenic Total
98 303
20 430
10.2
2.7 12.9
9.4
9.5
36.3
60.4
45.7
69.9
HereTg = 1012g. Data arefromJ. A. Van Aardenneet al. (A high-resolution datasetof historical anthropogenic tracegasemissions for the period 1890-1990, submittedto GlobalBiogeochemical Cycles,1999). NMHC is nonmethanehydrocarbons.
years, two lines of thinking evolvedabout the origin of troposphericozone:one emphasizedthe role of in situ photochemistry [Chameidesand Walker, 1976; Crutzen,1974;Fishmanet al., 1979], and the other emphasizedthat of ozone transport from the stratosphere[Chatfieldand Harrison, 1976;Fabian and Pruchniewicz,1977;Levy et al., 1985]. Recently, three-dimensional (3-D) global chemistrytransportmodelshavebeen developedto explicitlyaccountfor both photochemicaland meteorologicalprocesses[Crutzen and Zimmermann,1991;Mallet and Brasseur,1995;Roelofsand Lelieveld, 1995; Beretsenand Isaksen, 1997; Levy et al., 1997; Tie and Hess, 1997; Houwelinget al., 1998; Wang et al., 1998; Hauglustaineet al., 1998; Crutzenet al., 1999]. The model resultsindicate that the mean global troposphericO3 columnis 350 ___ 80 Tg, the transportfrom the stratosphereaccountsfor
quentlyused as a diagnostictool to quantify 0 3 sourcecategories and observed trends. We also present scenario calculations,of historical(1860) and possiblefuture (2025) tropospheric03 distributions,the latter basedon IntergovernmentalPanelon ClimateChange(IPCC) recommended moderate growth emissionestimates. 2.
Model
and Emissions
The global transport-chemistrymodel used has a spatial resolutionof 5ølongitudeand 3.75ølatitude.The verticalspacing, in 19 levelsup to 10 hPa, is defined accordingto terrain followingsigmacoordinatesnear the surface,pressurecoordinatesin the stratosphere,and a hybrid of the two in between. Tracer transport,cloudproperties,precipitation,temperature, 550 _+300Tg yr-•, thenetcontribution byin situphotochem- and other physical parameters have been derived from istryis 150 _+300 Tg yr-•, andthe dry deposition completes 6-hourlymean meteorologicalfieldsfrom the ECMWF reanalthe budgetthroughthe removalof 700 _+300 Tg yr-1. The ysesover the period 1979-1993 [Gibsonet al., 1997]. Tracer wide ranges of model results illustrate the uncertaintiesin- advectionis simulatedwith the slopesschemeof Russeland volved. These sourceand sink terms are suggestiveof an im- Lerner [1981]. Convectivetracer transportsare calculatedwith portant role of STE. However, net in situphotochemistryis the an updated mass flux scheme that accounts for shallow, residualof much larger O3 formation and lossterms,being of midlevel,and deep convection[Tiedtke,1989]. Turbulent verorder3000-3500Tg yr-•. Globally,thesetermsarein approx- tical transportis calculatedby stability-dependentvertical difimate balance. Locally, however, this may be very different, fusion [Louis, 1979]. Dry depositionof gasesand aerosolsis which determinesthe highly variable O3 distribution. parameterizedaccordingto Ganzeveldet al. [1998], and wet The aim of this article is to studythe contributionby natural depositionis parameterizedaccordingto Guelle et al. [1998]. and man-made trace gas emissionsto the troposphericO3 The descriptionof wet depositionaccountsfor both in-cloud distributionthroughphotochemistryand to determinethe im- and subcloudremoval of gasesand aerosols. portanceof STE. We presentresultsfrom a globalchemistryA detailed comparisonbetween simulated and measured transport model that uses European Centre for Medium- 222Rn indicates thatthe synoptic-scale modeltransport propRange Weather Forecasts (ECMWF) meteorological ertiesare representedaccurately[Dentenetet al., 1999].Stratoreanalysisdata to drive tracer transportsand removal.Recent spheric O3 is prescribedabove the 10 hPa level, and it is model improvementsinclude the representationof STE, non- relaxed toward zonal mean O3 between 10 and 50 hPa on the by the Total Ozone MappingSpectrommethane hydrocarbon(NMHC) chemistry,emissions,and basisof measurements depositionprocesses. We used 15 years of actualmeteorolog- eter (TOMS) [McPeters et al., 1996]and ozonesondes[Fortuin ical data and a recentlydevelopedhistoricalemissiondata set and Kelder, 1998], whereasthe 3-D ozone variabilityis mainto study troposphericO3 over the 1979-1993 period and to tained by simulated transports. Above 10 hPa a column validate the model by a direct comparisonwith in situ obser- amount is prescribedon the basis of measurementsby the vations,for example,from ozone sondes.The model is subse- HalogenOccultationExperiment(HALOE) instrumenton the
LELIEVELD
AND
DENTENER:
WHAT
CONTROLS
TROPOSPHERIC
OZONE?
3533
stratosphericwesterlies.By thermal wind balance the wave forcing inducespoleward motion and a downwardmassflux toward the midlatitudeand high-latitudetroposphere[Holton et al., 1995]. The stratosphericmassbalanceis maintainedby upward motion acrossthe tropical tropopause. The stratosphericwave propagationis most efficientin the westerlyflow duringwinter [Charneyand Drazin, 1961]. Further, orographic wave forcing is strongestin the Northern Hemisphere.Consequently,STE is relativelystrongestat midlatitudesand high latitudesin the Northern Hemisphericwinter. Furthermore,seasonalchangesin the massof the lowermost stratosphere also cause STE. During spring the tropopausealtitude increases,which entrainsstratosphericair into the troposphere[Reiter,1975;Appenzeller et al., 1996].The combinedeffect of theseprocesses is that downward03 transport reachesa maximumin late winter and early spring.Since this almostcoincideswith a springozonemaximumobservedat severalbackgroundmonitoringstations,it is tempting to assumean importantrole for STE, evenat the surface.In section 4, however,we will showthat in many locations,especiallyin information from the Emission Database for Global Atmothe midlatitude Northern Hemisphere, the spring 03 maxisphericResearch(EDGAR) [Olivieret al., 1996]. Anthropo- mum is due to in situ photochemistryrather than to STE. genic emissioncalculationshave been performed for 10 year In our modelwe accountfor STE by constrainingzonal and intervals since 1860 (J. A. Van Aardenne et al., A high- monthlymean stratosphericozoneon the basisof a combinaresolutiondata set of historicalanthropogenictrace gasemis- tion of ozonesondeand satellitemeasurements[Russellet al., sionsfor the period 1890-1990, submittedto Global Biogeo- 1993;McPeterset al., 1996; Fortuin and Kelder, 1998]. In the chemicalCycles,1999,hereinafterreferred to asVan Aardenne three upper model layers (midlevel at 10, 30, and 50 hPa), whereas et al., submittedmanuscript,1999). Emissionfactors account column03 is relaxedtoward zonal mean observations, for demographical, economical,agricultural,and technological the 3-D ozone distribution is simulated by ECMWF-based developmentsduringthe past century.Emissionestimatesfor transports.Synopticwavesdominatethe extratropicalcircula1860, 1993, and 2025 are presentedin Table 1. Note that the tion in the troposphereand lowermoststratosphere,beingwell 1860, 1993, and 2025 emission scenariosare simulated on the resolvedin the ECMWF-derived air masstransports,and STE basisof the meteorologicaldata set for 1993. Emissiontrends is simulatedaccurately.Plate 1 showsa comparisonof modeled of 03 precursorsbetween 1979 and 1993 are scaledwith re- 03 columnswith TOMS observations.Note that althoughthe gional trends of fossil fuel-related CO2 emissions[Carbon zonal average03 densityabove50 hPa is prescribed,the longitudinalozonevariabilityup to 10 hPa is simulatedexplicitly DioxideInformationAnalysisCenter(CDIAC), 1997]. Biomassburning emissionestimateshave been improved by the model. This variability resultsfrom planetary wave acand evaluatedby Marufu et al. [2000]. Marufu et al. [2000] tivity, which determinesthe mean tropopauseheight. The efpresent a comparisonof model resultswith measurementsat fectsof planetary-and synoptic-scale wave disturbanceson O3 low latitudes,mostly over Africa. Natural emissionsare dis- in the lower stratosphereand upper troposphereare thus extributed according to Lelieveld et al. [1998], and those of plicitly accountedfor in the model, consistentwith the wave NMHC are distributedaccordingto Guenteret al. [1995].The forcingof the Brewer-Dobsoncirculation. anthropogenicemissionshavebeen scaledup to the year 2025 An important indication about the model performance is on the basis of the IS92a scenario of the IPCC. This scenario providedby a comparisonof the resultswith 03 sondemeaaccountsfor World Bank and United Nations globalpopula- surementsobtainedat midlatitudesand high latitudesbetween tionforecasts (2025:8.4x 109people),economic growth(2.3- 36ø and 74øN, i.e., at Tateno, near Tokyo (Japan), Boulder 2.9%yr-•), fossilfuel-relatedemissions (including an energy (central United States), Hohenpeissenberg (Germany), useefficiency increase of 0.8-1% yr-•), estimated deforesta- Edmonton (southwesternCanada), and Resolute (northern tion rates, and agriculturaldevelopmentsby the Food and Canada). Plate 2 showsgenerallygood agreementbetween Agriculture Organization. model resultsand measurements,for example,of the seasonal 03 cyclesat both 200 (-12.5 km) and 700 hPa (-3 km). Since 200 hPa is locatedwithin the lowermoststratosphereat these 3. Stratosphere-Troposphere Exchange latitudes,this indicatesthat stratosphericozone,as a boundary Stratosphericozone is formed from 02 photolysisby short- conditionfor our troposphericozone simulations,is well capwaveultravioletradiation(• < 240 nm), mostlyin the tropical tured by the model (Plate 2). Nevertheless,the model doesnot stratosphere.It is transportedpolewardin the Brewer-Dobson reproducesomeof the highest03 peaks at 200 hPa observed circulation,which establishesthe ozone layer [Brewer,1949; when the sondestraverse localized 03 maximumsassociated Dobson, 1956]. The Brewer-Dobsoncirculation is primarily with stratosphericintrusionsinto the troposphere.These indrivenby wave disturbancesthat originatein the troposphere trusions,or tropopausefolds,havea typicaldimensionof a few [Charneyand Drazin, 1961;Hayneset al., 1991]. Thesewaves tens of kilometers. Because of the horizontal model resolution are excitedby air flow over mountains,synopticweather sys- of 3.75ø x 5ø, the folds are representedas synoptic-scale featems, and deep convection.They propagate into the strato- tures, and cross-tropopause 0 3 transportis averagedover a spherewhere they dissipate,whichexertsa dragforce onto the larger area [Kentarchos et al., 1999].
Upper AtmosphereResearchSatellite(UARS) [Russellet al., 1993]. Methane is prescribedat the surfaceon the basisof observationsand interpolation with a global model: global mean surfaceCH 4 for 1860 is 805 ppbv, for 1993 it is 1745 ppbv, and for 2025 it is 2245 ppbv [Etheridgeet al., 1998; Lelieveldet al., 1998]. The chemicalschemeaccountsfor 47 speciesthat describeCH4-CO-NMHC-NOx-SOx chemistryof which 32 are transported(includingmarked tracers). The model accountsfor 24 photodissociation, 67 thermal reactions [Houwelinget al., 1998], and heterogeneousprocessesaccording to Dentenetand Crutzen[1993].The chemistrycalculations are performedwith a time resolutionof 2400 s. Photodissociation frequenciesare calculatedwith the schemeby Landgraf and Crutzen[1998],includingthe effectsof (multiple) scattering by cloudsand aerosolsand of changing03 in the stratosphere[Kroland van Weele,1997].The chemicalequationsare solvedwith a Eulerian backwarditerative solver[Hertelet al., 1993]. Trace gas emissionestimateshave been recently updated and recalculatedon a 1ø x 1ø grid accordingto historical
3534
LELIEVELD
AND
DENTENER:
WHAT
CONTROLS
TROPOSPHERIC
OZONE?
A
!
250
300
350
400
450
DU
Plate 1. (a) Model-simulatedcolumn O3 comparedto measurementsby (b) the Total Ozone Mapping Spectrometer(TOMS), both aYeragedfor April 1993,showingthat planetarywa¾½-dri¾½n variabilityin total 03 is well reproducedby the model. The black area indicatesan absenceo½measurements.DU is Dobson
Units(1 DU = 2.7 x 10•6 molecules cm-•).
LELIEVELD
AND
DENTENER
Tateno 36 N 140 E 700 h Pa 80i
100
....
WHAT
CONTROLS
TROPOSPHERIC
OZONE?
model NOx 700 hPa
average seasonalcycle 700 hPa "''
'
I
'
'
'
i
'
'
'
i
' v ,
I
'
'
' 1 '
3535
0.15
"''
I ,• ,11
'
'
'
l'
]
I
....
I
'
'
'
'
I
0.10
60 I
.
'!
4o
o
0,05 2
20
0.00
75
80
85
95
o
2
4
Tateno '
'
"''1
36 N 140 E 200 hPa ....
I
6
8
lO
12
I
II,lJllli
75
80
85
month
year
'
'
'
I
'
'
model NOx 200 hPa
average seasonal cycle 200 hPa i
300
,
95
year
'"' •' ' I , . , i , '," , 1 . , . '1 , , . i , , .
0,20
25O
6OO
0.15
"øø /
I
i
i O.lO
o
lOO
2OO
0.05 5o
0 75
....
I .... 8O
r
I .... 85
i
I ,I,, , . 90
0
95
,,, 0
I,,., 2
I,,, 4
I,, 6
, I , , , I 8 10
month
year
0,00 12
..... 75
, .... 80
i 85
....
, .... 90
95
year
Plate 2. (left) Comparisonof model simulations(red) and measurements(black) at (a) Tateno (36øN, 140øE),(b) Boulder(40øN,105øW),(c) Hohenpeissenberg (47øN, 11øE),(d) Edmonton(53øN,113øW),and (e) Resolute(74øN,94øW)on the basisof ozonesoundingsbetween1979 and 1993.The model O3 hasbeen sampledat about the sametimes and locationsas the sondemeasurements.(top) The 700 hPa level and (bottom)the 200 hPa level.Linear trendsoverthis 15year periodare indicatedby the straightlines.(middle) The 15 year averageseasonalcyclesof 03 Ozone of stratosphericorigin (O3s) is indicatedin green. (right) The model-calculatedNO•cmixing ratios and linear trends. The measurementdata were obtained from the World Ozone Data Center of the World MeteorologicalOrganization(http://www.tor.ec.ca).
Our
results
indicate
that
the contribution
of STE
to the
more gradual downwardtransport through subsidenceat the
globaltropospheric ozonebudgetis565Tg yr-•, in reasonable anticyclonic side of these weather systems [Reed, 1955; agreementwith the UARS-derived STE flux of 450-590 Tg 0 3
Danielsen,1968;Mahlman, 1969].Our modelcanonly simulate this gradual processsince details, such as tropopausefolds, occur on a subgrid scale. The favorable comparisonof the 1999].The total model-calculatedphotochemical03 produc- model resultswith measurements(Plates 2 and 3) indicates tionin thetroposphere (3314Tg yr-]) is in approximate bal- that this limitation is not significant,especiallyif we focuson ancewith03 destruction (3174Tg yr-') (Table2). timescalesof about a week or longerrather than on individual
yr-' [Gettelman etal., 1997]andin themiddleof the rangeof recentmodeling studies (400-850Tg 03 yr-') [LelieveM etal.,
events.
Plate 3 showsthe seasonalcycleof 03 at severalmonitoring stations aswell as the model-derivedstratosphericcomponent The contributionof STE to 03 levels in the lower tropoof surface 03 (solid green lines). This 03 fraction from STE sphere,includingits influenceon air quality, hasbeen studied for manyyears [Tuck et al., 1985;Logan, 1985]. In rare occa- hasbeencalculatedby addingan 03 tracerto the model(O3s) sions,STE eventscan deeply penetrate the troposphereand that has the samevalue as 03 at the 100 hPa level and above. reachthe surface[Derwentet al., 1978;Daviesand Schuepbach, After transferacrossthe 100 hPa level, O3s is destroyedby the as "regular"03 [Fol1994].Especially,elevatedsitescan sometimesexperiencethe samechemicaland depositionprocesses direct influenceof STE [Reiter,1991;Elbernet al., 1997; Tsut- lowsand Austin, 1992;Roelofsand Lelieveld,1997]. Chemical sumi et al., 1998]. In general,however,stratospheric03 intru- loss of 03 and O3s is for the most part determined by the sions associatedwith extratropicalcyclonesare followed by reactions(R1) and (R2) andwith OH andHO 2. Reactionsthat
4.
STE Effect on Surface 03
3536
LELIEVELD
AND
DENTENER:
Boulder 40 N 105 W 700 hPa 8O :•-'
200
WHAT
CONTROLS
TROPOSPHERIC
model NOx 700 hPa
averageseasonalcycle700 hPa • 1 ' '
' 1 "•
:':!
' •
rl:'
'
OZONE?
0.40
' I
i
,
• "'""1'
'
'
ß :'
1 •
ß -
'
i
•
ß ß '
,
0.30
6O
160
0.20
8
O.lO
2O
0
i
i
,
75
,
i
....
80
i
....
85
i
.
.
.
90
o. oo
.
S5
0
2
4
6
Boulder 40 N 105 W 200 hPa ....
i
....
i
10
ß
75
12
.
.
i
i
....
80
month
year
800
8
....
i'
. ' ', '•
'½ '
'
'
'
t
• •"
I
• '
'
I
'
....
i
,
,
,
.
90
9,5
year
model NOx 200 hPa
averageseasonalcycle200 hPa
40O
I
85
'
'
I
'
'
'
•1 "'"•
0.30
'"'
....
• ....
• ' '",
'
• ""
,
,'
0.25
0,20
4OO
•
0.15
z
0.10 2OO
0
0
75
8O
85 year
90
95
0
• I = I , , , 2 4
0.00
6 month
8
10
12
7.=
Plate 2. (continued)
do not destroy"odd oxygen"species(e.g., 03, O, and nitrogen oxides)are not countedas ozoneloss.The differencebetween 03 and O3s below 100 hPa equalsthe amountof ozonephotochemicallyproducedwithin the troposphere(dashedgreen linesin Plate 3). Interestingly,the columnamountof O3sin the troposphereis hardly affectedby temporal changesin troposphericozone,for example,by the anthropogenic03 increase, asdeterminedfrom the scenariostudiespresentedin section7. This implies that the O3s lifetime is rather constantin time, beinglargelydeterminedby transportprocesses. Thereforethe "O3s tracer technique"providesa robustmethodfor estimating the contributionof STE to troposphericozone. Monitoring stations in the extratropical Northern Hemisphere,for example,Jungfraujoch(Switzerland),Mace Head (Ireland), and Mauna Loa (Hawaii), showa clear seasonal03 cyclewith a maximumduringMarch-May (Plate 3). Note that the Mauna Loa stationis located at 3.4 km altitude and Jungfraujochis locatedat 3.6 kin, i.e., in the free troposphere.As mentioned above, the near coincidenceof the spring STE maximumat northernmidlatitudesand highlatitudesmaygive rise to misinterpretation.Our model analysis,however,demonstratesthat at midlatitudesin the Northern Hemispherethe STE maximum occursin February-March, at least a month ahead of the seasonalO3 peak. Plate 3 alsoshowsthat during midlatitude spring, photochemicalO3 formation stronglyincreases.Some pollutants,suchas peroxyacetylnitrate (PAN),
accumulatein the winter midlatitudeand high-latitudetropospherebecausetheir photochemical removalis slow.After the winter, enhancedbreakdownof thesegasescontributesto 0 3 formation[PenkettandBrice,1986].In effect,springconditions (May-June) are optimal for net photochemical0 3 buildup, whichmore than compensates for the decreasinginfluenceof STE.
As notedin previousanalyses[Akimotoet al., 1996;Oltmans et al., 1996;Penkettet al., 1998],pollutant03 and its chemical precursorscan be transportedover long distancesfrom the United States and east Asia, across the North Atlantic and
PacificOceans,respectively.These transportsare quite efficientduringwinter when photochemicaldestructionof pollutantsis relativelyslow.In summerthe photochemicallifetime of O3 in the lowertroposphereis only --•2weeks,and transportis lessefficient,sothat muchphotochemically producedO3 islost beforeit reaches,for example,Mace Head and Mauna Loa. In addition, STE has a minimum in summer. Both these effects
contributeto the relatively low surfaceozone valuesat these stationsduring summer.In Tateno at 700 hPa (Plate 2) this summertimeO3 minimumis particularlypronouncedbecause
of a sharpSTE decline,whichgivesrise to a double-peaked seasonalcycle. At Cape Grim, Tasmania, model-calculatedsurface03 is somewhatunderestimated duringtheApril-Juneperiod(Plate
3). This maybe due to an underestimate of long-rangetrans-
LELIEVELD
120
AND
DENTENER:
Hohenpeissenberg47 N 11 E 700 hPa '"'
ß '
•
ß '"
'
•
....
WHAT
CONTROLS
TROPOSPHERIC
OZONE?
model NOx 700 hPa
average seasonal cycle 700 hPa
• .....
ß'
'
I '
'
'
I
'
r .
I
'' •
"'i'
'•"•
'
i
3537
0.40
k'.
....
• .
.
r '•' i
ß ,
c lOO 0.30
• 40
ß•' 0.20
O
I , 0
....
75
20
ß
80
z
•
85
90
0.00 0
2
4
6 month
year
Hohenpeissenberg47 N 11 E 200 hPa
1200
....
• ....
"•
', '
,,
,
0.10
0
95
I
8
10
12
!
•
,
!
75
"''
'
I
•
'
"'l
'
'
'
I
'
90
95
year
average seasonal cycle 200 hPa
• ....
i
80
'
'
I
'
•
'
I
•
model NOx 200 hPa 0.20 "'
• •
•
'
•
,
.
,
I
i
i
....
'
!
!
i
!
'
,'
ß
i
,
.
,
,
i
!
s
ß ,
i
,
.
.
1000 400 0.15
800 300
0.10 200 400 0.05 100 200
0 75
0.._•_.•. , I , , , I , , , I , , , I , , , I , , ,
8O
85
0
2
4
6
8
month
10
0.00
12
ß
75
80
85
90
95
year
Plate 2. (continued)
ported pollution, for example,from biomassburning in Africa and South America.
Our model indicates that the contribution
by in situ formed 03 at this locationis fairly constantthroughout the year,so that the seasonalcycleis largelydeterminedby STE, which has a maximum in September.Nevertheless,the contributionby in situ photochemistrystronglyexceedsthat of STE duringmostof the year. For Cape Point, the southerntip of South Africa, we somewhatoverestimatesurface03 in the biomassburning(dry) season,i.e., in spring(Plate 3). This is causedby artificial dispersionof biomassburning emissions within the largemodelgrid cells(5ø x 3.75ø).Moreover,O3 is overestimatedin some(1984, 1986,and 1989)but not all years (1983 and 1991-1993),whichpointsto an interannualvariability in biomassburning that is not included in our emission database.It is, nevertheless,clear that at this backgroundlocation,STE contributessignificantlylessto surface03 than in situ photochemistry. Zonally averaged,in the extratropicalSouthernHemisphere the mean STE contributionto surfaceozone can reach up to 30-50%. In the extratropicalNorthern Hemisphere this is
