Distribution and Fate of Selected Oxygenated Organic Species in the Troposphere and Lower Stratosphere Over the Atlantic

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 105, NO. D3, PAGES 3795-3805, FEBRUARY 16, 2000

Distribution and fate of selected oxygenated organic species in the troposphere and lower stratosphere over the Atlantic

H. Singh,• Y. Chen,• A. Tabazadeh,• Y. Fukui,• I. Bey,2 R. Yantosca,2D. Jacob,2 F. Arnold,3 K. Wohlfrom,3 E. Atlas,4 F. Flocke,4 D. Blake,• N. Blake,• B. Heikes,6 J. Snow,6R. Talbot,7G. Gregory,8G. Sachse,8S. Vay,8 and Y. Kondo9

Abstract. A large numberof oxygenatedorganic chemicals (peroxyacyl nitrates,alkyl nitrates, acetone,formaldehyde, methanol, methylhydroperoxide, acetic acid and formic acid) were measured during the 1997 SubsonicAssessment (SASS) Ozone and Nitrogen Oxide Experiment (SONEX) airbornefield campaignover the Atlantic. In this paper,we presenta first pictureof the distribution of theseoxygenated organic chemicals (Ox-organic) in the troposphereand the lower stratosphere,and assesstheir sourceand sink relationships.In both the troposphereand the lower stratosphere,the total atmosphericabundance of theseoxygenated species (ZOx-organic) nearly equalsthat of total nonmethanehydrocarbons (ZNMHC), which have been traditionallymeasured. A sizablefraction of the reactivenitrogen (10-30%) is presentin its oxygenatedorganic form. The organicreactive nitrogen fraction is dominatedby peroxyacetylnitrate (PAN), with alkyl ni- tratesand peroxypropionyl nitrate (PPN) accounting for <5% of totalNOy. Comparisonof observationswith the predictionsof the Harvard three-dimensionalglobal model suggeststhat in many key areas(e.g., formaldehydeand peroxides)substantial differences between measurements and the- ory are presentand must be resolved. In the caseof CH3OH, there appearsto be a large mismatch betweenatmospheric concentrations and estimatedsources, indicating the presenceof major un- known removalprocesses. Instrument intercomparisons as well as disagreementsbetween obser- vationsand model predictionsare usedto identify neededimprovements in key areas. The atmosphericchemistry and sourcesof this groupof chemicalsis poorly understoodeven thoughtheir fate is intricatelylinked with uppertropospheric NOx and HOx cycles.

1. Introduction per troposphere(UT) and sequesterreactive nitrogen [Singh et al., 1995; Jaeglg et al., 1997; Wennberg et al., 1998; Craw- Oxygenated hydrocarbonspecies have the potential to play ford et al., 1999]. In general, measurementsof organic oxy- an important role in processes of atmospheric ozone forma- genated species in remote troposphere are extremely sparse tion [Loyd, 1979; Singh et al., 1995; Wennberg et al., 1998]. and often limited to a few species. In the stratosphere, these Unlike most hydrocarbons, many oxygenated species can be measurementsare virtually nonexistent [Singh et al., 1997]. photolyzed throughoutthe troposphere. Carbonyls and perox- The Subsonic Assessment (SASS)Ozone and Nitrogen Oxide ides are potential sourcesof HOx free radicals, and molecules Experiment (SONEX)/DC-8 airborne field campaign of Octo- such as acetonecan form peroxyacetylnitrate (PAN) in the up- ber/November1997, with its focus on the NOx-HOx-O3 chemistry, provided a unique opportunity to characterize the distri- 1EarthScience Division, NASA Ames Research Center, Moffett bution of several important oxygenatedorganic species in the Field, Califomia. troposphereand the "lowermost" stratosphere (LS) over the 2Departmentof Earthand Planetary Sciences, Harvard University, Atlantic. An overview of the SONEX mission has been pro- Cambridge, Massachusetts. 3MaxPlanck Institute ftir Kemphysik,Heidelberg, Germany. vided by Singh et al. [1999]. 4AtmosphericChemistry Division, National Center for Atmospheric In this paper, we provide a description of the atmospheric dis- Research, Boulder, Colorado. tribution of selectoxygenated organic speciesas measureddur- 5Departmentof Chemistry,University of Califomia,Irvine. ing the 1997 SONEX mission. These species include organic 6Centerfor AtmosphericChemistry Studies, University of RhodeIs- nitrates (per.oxyacyl nitrates, alkyl nitrates), carbonyls land, Narragansett. 7Institutefor theStudy of Earth,Oceans, and Space, University of (acetone, formaldehyde), alcohols (methanol, ethanol), perox- New Hampshire,Durham. ides (methylhydroperoxide), and acids (acetic acid, formic 8EarthScience Division, NASA Langley Research Center, Hampton, acid). Instrumentation and measurement methods are not de- Virginia. scribedhere but have been previously published [Singh et al., 9SolarTerrestrial Environmental Laboratory, Nagoya University, 1999; Schumann et al, 1999; and references therein]. Toyokawa, Japan. SONEX/DC-8 and the Pollution From Aircraft Emissions in the North Atlantic Flight Corridor (POLINAT-2)/Falcon aircraft Copyright2000 bythe American Geophysical Union. platformsalso offered limited opportunitiesfor instrument in- Papernumber 1999JD900779. tercomparisonsand theseresults are discussed. In addition, the 0148-0227/00/1999JD900779509.00 Harvard three-dimensional(3-D) global model has been used to

3795 3796 SINGH ET AL.' DISTRIBUTION OF SELECTED OXYGENATED ORGANIC SPECIES compare these observationswith model predictions. We also pospheric observations have also been compared with the provide an assessmentof the sourcesand sinks of these chemi- Harvard3-D globalmodel (4 ølatitude x 5 ø longitude grid size), cals. an earlier versionof which has been describedby Wang et al. [1998a, b]. The presentversion has improved descriptionsof chemistry and emissions, and higher vertical resolution (20 2. Discussion of Results layers versus9 layers) making it more appropriate for upper It is pertinent to note that during SONEX attention was fo- troposphericsimulations. The model transports 24 tracers, cusedon the UT regionof the atmospherethat has been previ- including oddoxygen (Ox = O3 + NO2 + 2NO3), NOx, HNO3, ously minimally studied. SONEX objectives also were such N205, HNO4, three peroxyacylnitrates, a lumpedalkyl nitrate, that much of the data were collected between 40ø-65øN lati- CO, alkanes (ethane, propane, higher alkanes), alkenes, iso- tudes with only one flight going as far south as 20øN. The prene, carbonyls(acetone and higher alkyl ketones, formalde- stratospherewas frequentlysampled although O_•concentra- hyde, and higher alkyl aldehydes,methacrolein, methylvinyl tions never exceeded450 ppb, and N20 was never below 300 ketone), and peroxides(H202, CH3OOH). A present shortcom- ppb. This region of the stratospherehas been commonly de- ing is that the model contains 1993 meteorology. Compari- son of the 1993 simulation with SONEX observations should fined as the "lowermoststratosphere" (LS). With this perspec- still be useful in terms of mean concentrations and concentra- tive in mind, a greater emphasis is placed on the UT/LS re- gions. Only SONEX data collectedon the DC-8 are usedin this tion gradients. Fuelberg et al. [this issue] conclude that, study. POLlNAT Falcon data are occasionally consideredfor withinthe normalyear-to-year variability, 1997 wasa clima- purposes of comparisons. tologically averageyear. In the following section, we shall TheSONEX NOy data were examined against several tracers discussthe distribution,sources, and sinks of selectedorganic that may be usedto distinguish the stratospherefrom the tro- speciesin the troposphereand the LS. posphere(03, N20 , and CO). It was determinedthat a simple 2.1. Tropospheric Distributions filter (O3>100ppb and Z>6 km) was adequateto define periods of stratosphericinfluences. This filter has been applied to all Figure1 showsthe vertical distribution of 03 andCO (a use- tropospheric/stratosphericobservations discussedhere. Tro- ful tracer) as observedduring SONEX over the North Atlantic

12 12

10 10

E 8 E 8

50 100 150 200 50 100 150 200 CO, ppb 03, ppb 150 150

lOO lOO

50 50

10 20 30 40 50 60 70 10 20 30 40 50 60 70 LATITUDE,ON LATITUDE,ON Figure 1. Troposphericvertical and latitudinal distribution of 03 andCO over the Atlanticduring SONEX. Periodsof stratosphericinfluences (O3>100 ppb; Z>6 kin) are excluded. The vertical structureis for 45ø-65øN with averagescalculated over 2-km bins. Latitudinaldistribution is for the uppertroposphere based on data collectedbetween 8-12 km with averagescalculated over 5ø bins. Resultsfrom the Harvard3-D global model are averagedover 5ø-70øW longitude (SONEX regime) and are shownin similarlines (no datapoints). SINGH ET AL.' DISTRIBUTION OF SELECTED OXYGENATED ORGANIC SPECIES 3797

12 I I I 2.1.1. Organic nitrates. An onboard gas chroma- tograph (GC)operated by the NASA Ames group measured 10 peroxyacetyl nitrate (PAN), peroxypropionyl nitrate (PPN),

E methyl nitrate (MN), and ethyl/i-propyl nitrates in real time. The stable alkyl nitrates (methyl, ethyl, i-propyl, and 2-butyl) _ %% NM H C were also measuredfrom the whole air samplescollected by the University of California, Irvine group. Figure 3 shows a scat- Ox-org _• terplot of thesemeasurements obtained by the two groups during the SONEX mission. Recognizingthat the two techniques of sample collection are widely different, sampling times are not exactly identical, and mixing ratios are quite low, agreement between the onboard GC and canister sampled measure- I I I ments is consideredquite good. Statistical measuresof ensem- 0 1000 2000 3000 4000 ble averages indicated that these two data sets could be com- Concentration, ppt bined for all intents and purposes. PAN measurementswere 2OOO not duplicatedon the DC-8, and it was also not measuredon the 8-12 km AIt. POLINAT-2/Falcon, so no comparisonis possible. Figure 4 shows the troposphericdistribution of reactive or- 1500 ganicnitrogen species along with thatof NOy,a measureof the total available reactive nitrogen. Among this group of organic nitrates, PAN was the most abundant at --200 ppt in the lOOO lower troposphereand --50 ppt in the UT (Figures 4a and 4b). TypicallyPAN accounted for 10-30% of the availableNOy. The higher concentrationsin the lower troposphereare indica- 5OO tive of continental pollution influences over the Atlantic. Figure4 alsoshows a comparisonof modeledPAN and NOy with SONEX observations with reasonable agreement. The

10 20 30 40 50 60 70 LATITUDE, øN 10 I I I I Figure 2. Tropospheric vertical and latitudinal distribution y = 0.73 + 0.78x R2=0.50 of total oxygenated chemicals (:ZOx-organic) and total nonmethane hydrocarbons(:ENMHC) over the Atlantic as measured during SONEX. Layout is the sameas in Figure 1.

(45ø-65øN). The mean vertical distribution is basedon 2 km altitude bins and mean, median, and percentiles are shown. The uppertropospheric latitudinal distribution is describedby i . . averagingdata over the 8-12 km altitudebin. The actualdata (1-min average)are alsoshown to give someidea of the distri- butionof sampling.It is evidentthat southof 40øN the sam- 0 2 4 6 8 10 pling is muchsparser than at higherlatitudes. Higher CO mix- Methyl Nitrate UCI, ppt ing ratiosin the lower troposphereare indicativeof pervasive continental influences over the Atlantic. The Harvard 3-D 25 I I I I model capturesthe observeddistribution of CO quitewell. It y = 0.04 + 0.79x R2=0.84 overestimatesOs and slightlyunderestimates CO in the UT. A • 20 possiblereason for this may be the higher than normal tro- popauselevels observedduring SONEX [Fuelberg et al., this . • 15 issue].

Figure2 providesa perspectiveon the abundanceand distri- z bution of total oxygenatedchemicals (:SOx-organic) in rela- •, lO ß . .$' . ß .' o tion to that of nonmethanehydrocarbons (:ENMHC)measured ß. • ß ,.''t.,I=.,••_-...:.:. duringSONEX. TheseNMHC includeda large numberof al- .- + 5 :.= ;_.::: ß kanes, alkynes, alkenes, and aromatics. It is evident from ß.',.'•!' '" '.. ø ' . -- Figure2 that in the remotetroposphere, oxygenated organics are collectivelyas abundantas NMHCs. Becausemany of them 0 .. -. ,. , , are easily photolyzedat troposphericwavelengths, their influ- 5 10 15 20 25 enceon uppertropospheric chemistry can be far greaterthan Ethyl+i-Propyl Nitrate UCI, ppt that of NMHC. As will be discussedin the following sections, Figure 3. comparisonof methyl nitrate and ethyl/i-propyl the sources,sinks, and chemistry of these oxygenated mole- nitrate as measuredaboard the DC-8 in real time by NASA culescan differ greatly from eachother. In many cases, sig- Ames (y axis) and from the analysis of whole air samplesby nificant biogenic sourcesappear to be present. UCI (x axis) during SONEX. 3798 SINGH ET AL.: DISTRIBUTION OF SELECTED OXYGENATED ORGANIC SPECIES

12 I I I I I

1000 10 PAN

E 8 : X ...• •.; NOy

em m 6 mmmm 100 _ 4 • x • ""--.j. 2 IIIiimI :. 0 I I 10 0 200 400 600 800 1000 10 20 30 40 50 60 70 Concentration, ppt LATITUDE, ON

t- o

0 I 0 10 20 30 40 50 10 30 40 50 60 Concentration, ppt LATITUDE, ON Figure 4. Troposphericvertical and latitudinaldistribution of select reactivenitrogen speciesmeasured during SONEX. Layoutis the sameas Figurein 1. PAN, peroxyacetylnitrate; PPN, peroxypropionylnitrate; MN, methylnitrate; TAN, totalalkyl nitrates' Nay is themeasured sum of totalreactive nitrogen (ZN).

distribution of PPN (Figures 4c and 4d)is similar to that of HOx radicals[Crawford et al., 1999]. Acetonewas measuredon PAN, but its mixing ratios are significantly lower. This is the DC-8 with an onboard GC operatedby the NASA Ames largelydue to the lesserabundance of PPN precursorsas well as group and on the Falcon by the Max-Planck-Institute Heidel- its somewhatshorter lifetime. Unlike PAN and PPN, methyl berg group who useda techniqueinvolving Chemical Ioniza- nitrate (MN)was nearly uniformly distributed in the tro- tion Mass Spectrometry (CIMS) [Arnold et al., 1997; posphere(Figures 4c and4d). It was clear that a large oceanic Wohlfrom et al. 1999]. Both instrumentsquoted substantial source,as observedin the tropical pacific (H. Singh, unpub- error bars of_+25% (GC) and _+50% (CIMS). The CIMS instru- lished data, 1996), was not present in this region. Collec- ment was not directly calibrated, and the reaction of product tively, total alkyl nitrates(TAN) were presentat relatively low ions with water vapor was recognizedas a potential sourceof concentrationswith a distribution that was suggestiveof sur- error. The GC methodcollected a sampleover a 150-s period, face, probably anthropogenic, sources(Figures 4c and 4d). while the CIMS instrumenthad a muchfaster response time (10 Total alkyl nitrates on average accountedfor less than 3% of s). Figure 5 shows an example of the acetone distribution reactivenitrogen (Nay). All of thesespecies (pernitrates and measuredby the two aircraftwhen they were in the vicinity of nitrates)are thermally and photochemically dissociatableand each other during the two missions. The altitudes of the two can act both as a sourceor a sink of the available free NOx aircraftare also shown,and the CIMS data have been averaged [Singh, 1987]. over the sampling time of the DC-8 instrument. During mis- 2.1.2. Carbonyls. Acetone and formaldehyde,two of sion 5 (Figure 5a), the two instrumentsproduced comparable the most important carbonyls in the atmosphere,were meas- data and saw acetoneconcentration decrease in the stratosphere uredduring SONEX. In recent years, acetonehas been shown (at 11 km). However, the decline observedby the Falcon was to be a globally abundantspecies that has the potential to less than that observedby the DC-8. Similarly,' during mis- bothsequester reactive nitrogen as PAN andprovide an upper sion 7, the DC-8 measurement is some 30% lower than the troposphericsource of HOx radicals[Singh et al., 1994, 1995; CIMS measurement.These resultsare within the quotederror Arnold et al., 1997]. Similarly, formaldehydeis a well known bars of theseinstruments. Figure 6 showsthe vertical and lati- productof methaneoxidation and a major secondarysource of tudinal distribution of acetonein the tropospherebased only SINGH ET AL.' DISTRIBUTION OF SELE•D OXYGENATED ORGANIC SPECIES 3799

12 15oo I I I ment accuracyneeds to improve in the future, it is evident that Mission 5 substantial concentrations of acetone are present in the tro- (Oct. 18) 10 posphere. The role of acetone as a source of HOx during SONEX has been discussedby Jaeglget al. [this issue]. 000 8 Global sourcesof acetone are not well known although a first emissionsinventory with a 40-60 Tg yr-1 source was pre- sentedby Singh et al. [1994]. Here we shall updatethis emis- 500 sionsdata based on new information. In recent years, a number of previously suggestedbiogenic pathways have been confirmed, and new ones have been identified. Pastures have been ß ß found to emit significant quantities of acetone [Kirstine et al., o o 1998], and elevated levels of acetone in rural forested areas are 4.5 10 4 5 104 5.5 104 6 104 6.5 104 routinely observed [Goldan et al., 1995; Reimer et aI., 1998; TIME (UTC), seconds Fall, 1999; Lamanna and Goldstein, 1999]. Singh et aI. [1994] deducedthat the 15 (+10)% acetone yield from c•- 15oo pinene/OH reaction, as reported by C. Gu and co-workers Mission 7 (Oct. 23) 1 0 (unpublishedreport, 1984), wouldconstitute an =6 Tg yr-] sourceof acetone. More recently,(z- and [3-pineneshave been

. 000 \T 3> confirmed to have a 8-15% acetone yield upon OH reaction, and acetoneyields from many other monoterpenesfrom OH/O• c reactions have been determined [ReisseII et aI., 1999]. Simi- c• m . larly, OH oxidation of 2-methyl-3-buten-2-ol, another terres- 500 , . . I 4 trial biogenic product, results in a 50-60% yield of acetone [Ferronato et aI., 1998; AIvardo et aI., 1999]. Recent best es- 2 timates of the global emissions of monoterpenes and 2- methyl-3-buten-2-olare 130-150 Tg yr-• and3-6 Tg yr-], re- I I I •x 0 o spectively [Guenther et aI., 1995; 1999; Harley et al., 1998]. 4 104 4.25 104 4.5 104 4.75 104 5 104 On the basis of these emission rates and the available kinetic TIME (UTC), seconds information, we calculate that a secondarybiogenic source of Figure 5. A comparisonof acetoneas measuredon the DC-8 =11 Tg yr'• of acetoneis present.This is likely to be a lower (solid circles) and the Falcon (solid squares)using two inde- limit as it only considers the first stages of chemical reac- pendenttechniques. The altitude profile DC-8 (solid line) and tions, and additional secondaryformation may occur. It is also Falcon-20 (dashed line) is shown. likely that the day/night difference observed in the mixing ratio of acetonein forestedareas [Goldan et aI., 1995] is in part due to this secondary source. Biomass burning sourceof ace- on the GC data collectedon the DC-8. A slight decline in the tone has also beendetermined (3-10 Tg yr'•) from direct at- uppertropospheric abundance of acetonetoward the subtropics mosphericmeasurements and laboratory experiments[Singh et was observed. This is expecteddue to the greater loss of ace- aI., 1994; HoIzinger et al., 1999]. Acetone sourcesl¾om dead tone due to photolysisand has been previously observedover plant matter via abiological processeshave been estimated to the Pacific [Singh et aI., 1995]. Although acetone measure- be 6-8 Tg yr-• [Warnekeet aI., 1999]. Humanbreath contains

Table 1. An Estimate of the Global Source Inventory of Acetone

SourceType Global Annual Comments* Emissions,Tg yr'l

Primary anthropogenic 2 (1-3) --95% NH Primary biogenic 15 (10-20) not well quantified Secondaryanthropogenic and biogenic Propaneoxidation 15 (10-20) --80% NH i-Butane/i-pentaneoxidation I (1-2) --80% NH i-Butene/i-penteneoxidation 1 (1-2) --80% NH Monoterpene/methylbutenol 11 (7-15) Oxidation* Dead/decayingplant matter 6 (4-8) Biomassburning 5 (3-10) --70% SH

Total source strength 56 (37-80)

Total sink strength 40-60 removal via hv, OH, and slow deposition *NH,northern hemisphere' SH,southern hemisphere *Theseterrestrial biogenic hydrocarbons are known to havelarge yields of acetoneupon OH andO• oxidation (see text). 3800 SINGH ET AL.: DISTRIBUTION OF SELECTED OXYGENATED ORGANIC SPECIES

12 Thereis a weaklinear relationship (R2= 0.3) betweenmethanol and formaldehydeconcentrations for all cases when data 10 - Acetone are above LOD. This relationship might be used to suggest that methanol is a potential source of formaldehyde or con- E versely a potential interferent in its measurement. At this Methanol . point, neither the possibility of additional sourcesfor HCHO nor the potential for measurementerrors can be completely ,, ruled out. HCHO is a key intermediate of atmospheric oxida-

_ tion whose atmosphericfate cannot be fully reconciled until more sensitiveand accuratemeasurements are possible. 2.1.3. Peroxides. Peroxides are secondaryproducts of atmosphericoxidation (RO2 + HO2 --• ROOH + 02), andtheir direct emissions are extremely small in comparison [Heikes et 200 400 600 800 1000 1200 al., 1996; Lee et al., 1999]. The organic methylhydroperoxide Concentration, ppt (CH3OOH)is muchless soluble than H202 and can be trans- ported via moist cloud convectioninto the UT thus providing a 1 ooo local sourceof HOx [Jaegld et al., 1997; Crawford et al., 1999]. Both hydrogen peroxide and methylhydroperoxide 8OO were measuredduring SONEX. A substantial fraction of the CH3OOH data(28% within 4-8 km and 44% within 8-12 km) were below the limit of detection. Here we have used 0.5 LOD 600 as the mixing ratio for these. When 0.25 LOD was used for sensitivity purposes,the mean concentrationswere minimally 400 altered (<5 ppt). Figure 7 showsthe vertical and UT latitudinal o distribution of CH3OOH. The atmospheric abundanceof 200 CH3OOHis comparableto that of HCHO. Even though the model appearsto be in better agreement with observations in the UT, this must be considered fortuitous. Throughout the

10 20 30 40 50 60 70 LATITUDE, ON 12 I I I I Figure 6. Tropospheric vertical and latitudinal distribution of acetoneand methanolduring SONEX. Layoutis the sameas lO in Figure 1. E relativelyhigh concentrationsof acetone(10 3--10 4 ppb), but this is only a significant source in indoor environments [Lindinger eta!., 1998]. Table 1 provides an updatedsource inventoryof acetonewith a global sourceof 56 Tg yr" which is nearthe upperlimit of the 40-60 Tg yr" previouslypub- lished by $ingh et al. [1994]. Unlike acetone,formaldehyde has an extremelyshort life in the atmosphere(a few hours), and its sourceis expectedto be 0 50 100 150 200 250 largelydue to hydrocarbonoxidation. Althoughdirect emis- Concentration, ppt sions of formaldehydefrom automobile exhaust and biomass 300 burning are known to exist, these sourcesare small (<5%) in I I I I comparisonwith the =103Tg yr" of HCHOthat canbe glob- 25O ally producedfrom CH 4 oxidation alone. We note that a substantial fraction of the HCHO data (42% within 4-8 km and "'"' 200 52% within 8-12 km) were below the limit of detection (LOD). o -.• Here we have used0.5 LOD as the mixing ratio for these in- .• 15o stances. When 0.25 LOD was used for sensitivity purposes, mean concentrations in the UT were further reducedby 5-10 c 100 o ppt. It is evident from Figure 7 that substantialconcentrations o of HCHO are presentin the troposphere. Figure 7 also shows 5o that the observed mean HCHO concentrations are in reasonable agreementwith those predicted by the Harvard model. During typical SONEX conditions, we calculate that acetone pho- 10 20 30 40 50 60 70 tolysis may be responsiblefor 5-7 ppt of HCHO in the UT. It LATITUDE, ON should be noted that point models are unable to reconcile Figure 7. Tropospheric vertical and latitudinal distribution measuredconcentrations of HCHO which exceeded70 ppt and of formaldehyde and methylhydroperoxide during SONEX. shouldbe the most reliable [e.g., Jaegld et al., this issue]. Layout is the same as Figure in 1. SINGH ET AL.: DISTRIBUTION OF SELECTED OXYGENATED ORGANIC SPECIES 3801

12 whose atmosphericfate cannot be fully reconciled with present knowledge. The' need for more sensitive measurementsin the 10 UT is clearly indicated. 2.1.4. Alcohols. Both methanol and ethanol were E measured during SONEX. Significant concentrations of methanol (400-800 ppt) were present in the Atlantic troposphereat all times (Figure 6). Ethanol concentrationswere often undetectableand nearly always below 50 ppt. The gas phase oxidation of methanol via OH leads to formaldehyde formation (CH•OH + OH + (02) --> CH20 + HO2 + H20), but this processis too slow in the UT to be a significant sourceof formaldehydeand free radicals[Singh et al., 1995]. Methanol was less abundant in the LIT relative to acetone despite its 100 i50 200 250 300 longer lifetime in the UT due to slow OH reactivity. Methanol NOx,ppt is preferentially partitioned in the gas phase even at very low

lOOO temperatures(K,= l0 sMatm 'l at 273øKin water).It canbe ar- gued that the greater relative decline of methanol comparedto acetonein the UT is due to the presenceof secondarysources of 8OO acetone throughout the troposphere. The presence of un- known, possibly heterogeneous,processes that could provide 600 effective sinks for methanol also cannot be ruled out. A weak linearcorrelation (R 2 = 0.4) of methanolmixing ratios with

400 thoseof acetoneand CO (moleculeswith comparablelifetimes) was observed suggesting some commonalty of sources. A similarrelationship (R 2 = 0.3) betweenmethanol and formal- 2OO dehyde (short-lived) may be used to suggestthat methanol is a potential source of atmospheric formaldehyde or conversely a potential interferent in its measurement. 10 20 30 40 50 60 70 A source inventory for methanol does not exist, but based LATITUDE, øN on removalby OH andslow deposition(=0.1 cm s-l), it has beencalculated that a globalmethanol source of 40-50 Tg yr'• Figure 8. Tropospheric vertical and latitudinal distribution of NOx during SONEX. Layout is the sameas Figure in 1. must exist. Recent studies of emissions from plants and leaves, and the presence of high concentrationsof methanol in forested and rural areas point to a sizable biogenic source [MacDonald and Fall, 1993; Goldan et al., 1995; Kirstine et troposphere,the simulation of CH•OOH is a substantial over- al., 1998; Reimer et al., 1998; Fall, 1999; Lainanna and Gold- predictionof observations.This overpredictionof CH•OOH is stein, 1999]. The measuredbiogenic emission rates are suffi- contraryto previousfindings from the South Atlantic [Jacobet cientlylarge for a methanolsource of >100 Tg yr-• [Guentheret al., 1996]. We believe a major reasonfor this is due to the fact al., 1995, 1999]. Large methanolsources from dead plant mat- that the observedNO x is nearly a factor of 2 larger than the ter via abiological processeshave been reported to be 1 8-40 model prediction throughoutthe troposphere(Figure 8), thus Tg yr'• [Warnekeet al., 1999]. Additionalsmaller sources allowing for a much faster rate of CH•OOH formation in the from biomassburning (3-17 Tg yr-l) anddirect anthropogenic model. Uncertainties in the CH•O2 + HO2 rate constant emissions(2-4 Tg yr •) are present[Singh et al., 1995; Holz- [Demote et al., 1997] also contribute to this disagreement. inger et al., 1999; Yokelson et al., 1999]. On the basis of the CH•OOH is a key intermediate of secondaryHOx production globel model of Lawrence et al. [1999], secondary source from

Table 2. An Estimate of the Global Source Inventory of Methanol

Source Type Global Annual Comments Emissions,Tg yr'•

Primary anthropogenic 3 (2-4) =95% NH Primary biogenic 75 (50-125) not well quantified Methane oxidation 18 (12-24) 3-D model estimate Dead/decayingplant matter 20 (10-40) Biomassburning 6 (3-17) =70% SH 999 Oceanic source ß . . likely source

Total sourcestrength 122 (75-210)

Total sink strength 40-50 removal via OH and slow deposition 3802 SINGH ET AL.: DISTRIBIJTION OF SELECTED OXYGENATED ORGANIC SPECIES

1500 12 i i i i i i [NOy,ppt]=18067-55.70 [N20,ppb]; R2=0.58 10

->, 1000 o z

I I 5OO 50 100 150 200 295 300 305 310 315 Concentration, ppt N20,ppb lOO Figure 10. Therelationship between NOy and N20 in the "lowermost"stratosphere. The stratosphereis definedby 80 03>100 ppb andZ> 6 km.

T 60 km and 85% between 8-12 km) were below the limit of detection (LOD). As stated earlier, in Figure 9 we have used IX)D 40 data at a value of 0.5 LOD. The UT acetic acid abundance is clearlyhighly influenced by the treatmentof LOD data. When 20 0.25 LOD was used for sensitivitypurposes, the UT concentrationswere reducedby about10 ppt with no perceptiblechange

10 20 30 40 50 60 70 Table 3. Mixing Ratios of SelectedTrace LATITUDE, ON Constituentsin the "Lowermost"Stratosphere Figure 9. Troposphericvertical andlatitudinal distribution (03> 100 ppb;Z>6 km) During SONEX of acetic acid and formic acid during SONEX. Layout is the same as Figure in 1. Trace Chemicals Mixing Ratios { Mean+(7 (Median, n) }

03, ppb 191+82( 165, 685) CH4/NMHC oxidationpathways (2CH302 --> CH3OH+ CH20 + N20, ppb 309+3 (310, 476) 02) canbe calculatedto be =18 Tg yr-' (R. von Kuhlmann, H20, ppm 46+26(40, 614)* Max-Planck-Institute for Chemistry, private communication, CH4, ppb 1722+20(1728, 582) 1999). Although CH3OH is easily synthesized in seawater CO, ppb 48+14 (48, 581) NOy,ppt 874+194(881,592) from the hydrolysisof methyl halides, no estimate of its oce- NO, ppt 67+58 (51,586) anic sourceis presentlyavailable. Table 2 summarizesa first NOx, ppt 136+83 (133, 477) inventoryof the globalsources (=122 Tg yr-') of methanolin- HNO 3, ppt 649+255 (628, 178) dicatingthat its sourcesfar exceedits known sinks. If this es- PAN, ppt 61+29 (54, 115) PPN,ppt 1.7+1.3(1, 53)* timate of the global methanol sourceis found to be correct, CH3ONO2, ppt 1.3_+0.6(1.3, 218) then removal mechanisms other than OH oxidation (and slow C2H5ONO2, ppt 1.1_+0.6(1.0, 220) deposition) must play an important role. In the boundary i-C3H7ONO2, ppt 0.7_+0.8(0.4, 220) layer, aerosolmedia containingOH, C1, and SO4 could react HCHO, ppt 31+19 (25, 78)* with methanol to form complex products. In the middle and Acetone, ppt 336+225 (296, 55) CH3OH, ppt 104+102(71, 55) UT, methanolmay also react in clouds to oxidize into organic CH3OOH,ppt 16+14(12, 232)* acidsor reactwith U2SO4 in clouddroplets to form methanesul- H202, ppt 31+24 (26, 232) phonic acid [Murphy et al., 1998]. Suffice it to say that CH3COOH,ppt 23+10(21, 172)* methanolis a globally abundantorganic specieswith limited HCOOH, ppt 39+27(30, 178)* C2H6 , ppt 380!-_160 (381,220) atmosphericdata and poorly understoodsources and removal C3H8, ppt 38+44 (24, 220) processes. C2H2, ppt 43+26 (37, 220) 2.1.5. Organic acids. Formic and acetic acids are C2H4, ppt 2+4 (1, 97)t ubiquitoustrace gasesin the atmosphereand contribute a large C6H6 , ppt 4+5 (2, 178) fraction of the free acidity in precipitation in remote areas ZOx-organic, ppt 563+262 (507, 40) •ENMHC, ppt 476+245 (220, 449) [Keeneand Galloway, 1988]. During SONEX, formic acid was CN-fine, numbercm -3 1037+659(896, 685) presentin concentrationsof =45 ppt in the UT and =120 ppt in the lower troposphere(Figure 9). The mean abundanceof ace- *Aninstrumental offset (= 10 ppm) has been detected. tic acid was less than half that of formic acid. Unlike formic Actual concentrations are lower. acid, a large fractionof the acetic acid data (50% between4-8 *Generallynear the limit of detection. SINGH ET AL.: DISTRIBUTION OF SELECTEDOXYGENATED ORGANIC SPECIES 3803

150 • • • • • • • 150

120 ß 120

E• e ß ß ß

0 0 60

lOOO 5OO

ß 8OO 400

ß ß ß 600 • 300 ß ß ß o

ß z 400 _•..•_• ee ß ",- 200 - ee• • ß

•e ß 2OO lOO ß •

o el el I ß I I i o 100 150 200 250 300 350 400 450 100 150 200 250 300 350 400 450 03, ppb 03, ppb Figure 11. Mixing ratiosof selectedoxygenated species as a functionof Os in the "lowermost"stratosphere.

below 7 km. Primary emissions from terrestrial vegetation ppb]= -447 + 7.0[N20, ppb];R 2 = 0.9) arein goodagreement and from combustion of fuel and biomass, and secondaryfor- with previouslyreported results from the Pacific[e.g., Singh mation from gas phase and liquid reactions are known to be et al., 1997]. present [Talbot et al., 1990; Fall; 1999; Yokelson et al., In nearly all cases,mixing ratios of oxygenatedorganic 1999]. Although acetic acid could react with OH radicals to speciesdeclined rapidly above the tropopause.Deeper into the form HCHO (CH•COOH + OH ---> CH302 ---> CH3OOH and stratosphere,mixing ratiosof acids,peroxides, formaldehyde, HCHO), the processis quite slow, and this is at best a minor and methanol were extremely low and near the limit of detec- sourceof HOx radicals. In most cases, organic acids are re- tion as 03 approached400 ppb. Figure 11 showsthe distribu- moved from the atmosphereby wet and dry deposition proc- tion of someselected oxygenates as a functionof Os. PAN and esses,and their chemistryis at best highly uncertain[Talbot et acetone declined at a slower rate than methanol and other al., 1990, 1995; Jacob et al., 1996]. An accurateenough NMHC (e.g., C2H6). In this region, which is believed to have sourceinventory for theseorganic acids is not currently avail- --10s molecules cm -3 of C1atoms, C1 chemistry instead of OH able to model their global distribution. dominatesthe oxidationprocess. PAN, however,reacts negli- gibly slowly with C1 atoms and may have a potential local 2.2. Stratospheric Distributions sourceresulting from C3Hs/C2H6-C1and acetone photochemis- The stratospherewas frequentlypenetrated during SONEX. try [Demoreet al., 1997]. Acetoneitself can also be produced Chemical indicators of the stratospheric air suggestedthat from C3H8-C1oxidation in this region. The slower net loss these instances representedthe "lowermost" stratosphere rate of acetonecompared to methanol is evident from the fact (03<450ppb; N20>300 ppb). Figure10 showsthe NOyand that the acetoneto methanolratio in the stratosphereincreased N20 relationship for the stratospheric data set defined by fromabout 2 to 8 asO s increased from 100 to 400 ppb. There O3>100 ppb and Z>6 km. There is substantial scatter largely is someindication, based on extremelysparse data, that acetic associatedwith the fact that sampled air is a mix of tro- acid increasedin the LS from 40 ppt (03 -- 100-150 ppb) to 8 0 posphericand stratosphericinfluences. The slope of this line ppt (03 =300-400 ppb). This is consistentwith the view that ([NOy,ppt] = 18,067-55.7[N20, ppb]; R2= 0.6) is compara-acetone photolysis followed by reaction of peroxyacetyl radi- ble to but somewhat less steep than previously reported data calswith HO2 (CH3CO 3 + HO2 ---)CH3COOH + 03) is a possible [Singh et al., 1997; Keim et al., 1997]. N20 correlation with in situ sourceof the lower stratosphericacetic acid [Singh et HNO3 ([HNO•,ppt] = 25,431 - 80.2[N20,ppb]; R 2 = 0.6), 03 al., 1997]. Table 3 summarizesthe mixing ratios of a number ([Os, ppb] = 6924- 21.9[N20,ppb]; R 2 = 0.9), andCH4 ([CH4, of selecttrace species as measuredin the-LSduring SONEX. 3804 SINGH ET AL.: DISTRIBUTION OF SELECTEDOXYGENATED ORGANIC SPECIES

Here, as in the troposphere, Ox-organics are as much or more Jaeg16,L., et al., ObservedOH and HO 2 in the upper tropopsheresug- abundant than NMHC. gesta major sourcefrom convective injection of peroxides,Geo- phys.Res. Lett., 24, 3181-3184, 1997. 3. Conclusions Jaeg16,L., et al., Photochemistryof HOx in the upper troposphereat northernmidlatitudes, J. Geophys.Res., this issue. Keene, W. C., and J. N. Galloway, The biogeochemicalcycling of for- Oxygena•ed organic species are important sources of at- mic and acetic acids throughthe troposphere:An overview of cur- mospheric free radicals and are intricately linked with proc- rent understanding,Tellus, Ser. B, 40, 322-334, 1988. essesof troposphericozone formation. With the exception of Keim,E. R.,et al., Measurements of NOy-N20 correlation in the lower formaldehyde, the interest in oxygenated hydrocarbon is rela- stratosphere:Latitudinal and seasonalchanges and modelcompari- sons,J. Geophys.Res., 102, 13,193-13,212, 1997. tively recent. This study makes a case that these oxygenated Kirstine,W., I. Galbally, Y. Ye, and M. Hopper, Emissionsof volatile species are ubiquitousand collectively are present in concen- organiccompounds (including oxygenated species) from pasture,J. trations that are comparableto those of nonmethane hydrocar- Geophys.Res., 103, 10,605-10,619, 1998. bons. They can participate in photochemistry of the tro- Lamanna, M. S., and A. H. Goldstein,In situ measurementsof C2-C•0 posphere in important ways and have complex biogenic VOCs abovea Sierra-Nevadapine plantation(including oxygenated species)from pasture,J. Geophys.Res., 104, 21,247-21,262,1999. sourceswhich require better quantificationand further study. In Lawrence, M.G., P.J. Crutzen, P.J. Rasch, B.E. Eaton, and N.M. Ma- many cases, the sourcesand sinks of this group of chemicals howald, A model for studiesof troposphericphotochemistry: De- are poorly known. It is likely that important new oxygenated scription,global distributions, and evaluation,J. Geophys.Res., in speciesare presentbut have not yet been identified. press,1999. Lee, M., B. G. Heikes, and D. W. O'Sullivan, Hydrogen peroxide and organichydroperoxides in the troposphere:A review, Atmos.Envi- Acknowledgments. This researchwas fundedby the NASA Atmos- ron., in press,1999. pheric Effects of Aviation Project (AEAP)/Subsonics Assessment Lindinger,W., A. Hansel,and A. Jordan,Proton-transfer-reaction mass (SASS) Program and the NSF Atmospheric Chemistry Program. We spectrometery(PTR-MS): On-line monitoringof volatile organic thank R. Friedt, S. Kawa, and D. Anderson, SASS Project Scientists,for compoundsat pptv levels,Chem. Soc. Rev., 27, 347-354, 1998. their encouragementand support. Special thanks are due to James A. Loyd, A., Tropsphericchemistry of aldehydes,NBS Spec. Publ. U.S. Eilers (SONEX Project Manager) and the NASA Ames DC-8 flight and 557, 27-48, 1979. groundcrew for making this experimenta success. MacDonald, R. C., and R. Fall, Detection of substantial emissions of methanolfrom plantsto the atmosphere,Atmos. Environ., Part A, 27, 1709-1713, 1993. References Murphy, D. M., D. S. Thomson,and M. J. Mahoney, In situ measurementsof organics,meteoritic material, mercury, and other elements Alvardo, A., E. Tuazon, S. M. Aschmann, J. Arey, and R. Atkinson, in aerosolsat 5 to 19 kilometers, Science, 282, 1664-1669, 1998. Productsand mechanismsof the gas phase reactions of OH radicals and O• with 2-methyt-3-buten-2-ol,Atmos. Environ., in press,1999. Reissell,A., C. Harry, S. Aschmann,R. Atkinson, and J. Arey, Formation Arnold, F., et at., Acetone in the upper troposphereand lower strato- of acetonefrom the OH radical- and O•-initiated reactionsof a se- sphere:Impact on trace gasesand aerosols,Geophys. Res. Lett., 24, ries of monoterpenes,J. Geo?hys.Res., 104, 13,869-13,879, 1999. 3017-3020, 1997. Riemer, D., et ai., Observationsof nonmethanehydrocarbons and oxy- Crawford,J., et al., Assessmentof uppertropospheric HO x sourcesover genatedvolatile organiccompounds at a rural site in the southeastern the tropical Pacific based on NASA GTF_,/PEMdata: Net effect on United States,J. Geophys.Res., 103, 28,111-28,128, 1998. Schumann, U., et al., Pollution from aircraft emissions in the North At- HOx and other photochemicalparameters, J. Geophys.Res., 104, 16,255-16,273, 1999. lanticflight corridor: Overview on the POLINAT projects,J. Geo- Demore, W., et al., Chemical kinetics and photochemicaldata for use in phys.Res., this issue,1999. stratosphericmodeling, Evaluation 12, JPL Publ., 97-4, 1997. Singh,H. B., Reactive nitrogenin the troposphere,Environ. Sci. Tech- Fall, R., Biogenicemissions of volatile organic compoundsfrom higher nol., 21,320-327, 1987. plants,in Reacative Hydrocarbonsin the Atmosphere,edited by N. Singh,H. B., et al., Acetonein the atmosphere:Distribution, sources and Hewitt, pp. 41-96, Academic,San Diego, Calif., 1999. sinks,J. Geophys.Res., 99, 1805-1819, 1994. Ferronato,C., J. J. Orlando, and G. S. Tyndall, Rate and mechanism of Singh,H. B., M. Kanakidou,P. J. Crutzen, and D. J. Jacob,High con- the reaction reaction of OH and C1 with 2-methyl-3-buten-2-ol, J. centrationsand photochemicalfate of oxygenatedhydrocarbons in Geophys.Res., 103, 25,579-25,586, 1998. the globaltroposphere, Nature, 378, 50-54, 1995. Fuelberg,H., et al., A meteorologicaloverview of the SONEX period,J. Singh,H. B., et al., Trace chemicalmeasurements from the northern Geophys.Res., this issue. midlatitudelowermost stratosphere in early spring:Distributions, cor- Goldan, P. D., W. C. Kuster,.F. C. Fehsenfeld, and S. A. Montzka, Hy- relations,and fate, Geophys.Res. Lett., 24, 127-130, 1997. drocarbon measurements in the southeastern United States: The Ru- Singh,H. B., A. Thompson,and H. Schlager,SONEX airbornemission ral Oxidantsin the SouthernEnvironment (ROSE) program 1990, J. and coordinatedPOLINAT-2 activity: Overview and accomplish- Geophys.Res., 100, 25,945-25,963, 1995. ments,Geophys. Res. Lett., 26, 3053-3056, 1999. Guenther,A. B., et al., A global model of natural volatile organic com- Talbot,R. W., et al., Sourcesand sinksof formic, acetic,and pyruvic poundemissions, J. Geophys.Res., 100, 8873-8892, 1995. acidsover central amazonia,J. Geophys.Res., 95, 16,799-16,811, Guenther,A. B., et al., Natural emisionsof non-methanevolatile organic 1990. compounds,carbon monoxide and oxides of nitrogen from North Talbot,R. W., et al., Carboxylicacids in the rural continentalatmos- America, Atmos.Environ., in press, 1999. phereover the easternUnited States during the ShanandoahCloud Harley, P., et al., Emissionsof 2-methyl-3-buten-2-olby pines:A poten- And PhotochemistryExperiment, J. Geophys.Res., 100, 9335-9343, tially large sourceof reactivecarbon to the atmosphere,J. Geophys. 1995. Res., 103, 25,479-25,486, 1998. Wang,Y., D. J. Jacob,and J. A. Logan,Global simulation of tro- Heikes, B., et al., Hydrogenperoxide and methylhydroperoxidedistribu- posphericO'•-NOx-hydrøcarbøn chemistry, 1, Modelformulation, J. tions related to ozone and odd hydrogenover the North Pacific in the Geophys.Res., 103, 10,713-10,726,1998a. fall of 1991, J. Geophys.Res., 101, 1891-1905, 1996. Wang,Y., D. J. Jacob,and J. A. Logan,Global simulation of tro- Holzinger, R., C. Warnke, A. Hansel, A. Jordan, and W. Lindinger, posphericO'•-NOx-hydrøcarbøn chemistry, 2, Modelevaluation and Biomassburning as a sourceof formaldehyde,acetaldehyde, metha- globalozone budget, J. Geophys.Res., 103, 10,727-10,755, 1998b. nol, acetone,acetonitrile, and hydrogencyanide, Geophys.Res. Lett., Warneke,C., et al., Acetone,methanol, and otherpartially oxidized 26, 1161-1164, 1999. volatileorganic emissions from deadplant matter by abiological Jacob,D. J., et al., Origin of ozoneand NO x in the tropicaltroposphere: processes:Significance for atmosphericHO x chemistry,Global A photochemicalanalysis of aircraft observationsover the South Biogeochem.Cycles, 13, 9-17, 1999. Atlantic basin,J. Geophys.Res. 101, 24,235-24,250, 1996. Wennberg,P. O., et at.,Hydrogen radicals, nitrogen radicals, and the SINGH ET AL.: DISTRIBUTION OF SELECTED OXYGENATED ORGANIC SPECIES 3805

productionof ozone in the upper troposphere,Science, 279, 49-53, D. Blake and N. Blake, University of California, PhysicalSciences 1998. Building, Irvine, CA 92697. Wohlfrom, K.-H., T. Hauler, F. Arnold, and H. Singh, Acetone in the Y. Chen, Y. Fukui, H. Singh, and A. Tabazadeh, NASA Ames Re- free troposphereand lower stratosphere:Aircraft-based IMRMS and searchCenter, Moffett Field, CA 94035. ([email protected]) GC measurementsover the North Atlantic and a first comparison, G. Gregory,G. Sachse,and S. Vay, NASA Langley ResearchCen- Geophys.Res. Lett., 26, 2849-2852, 1999. ter, Hampton, VA 23681. Yokelson,R. J., et al., Emissionsof formaldehyde,acetic acid, methanol, B. Heikes and J. Snow, University of Rhode Island, Center for At- and other trace gasesfrom biomassfires in North Carolina measured mosphericChemistry Studies, Narragansett, RI 02882. by Airborne Fourier Transform Infrared Spectroscopy(AFTIR), J. Y. Kondo, Nagoya University, Honohara 3-13, Toyokawa, Aichi Geophys.Res., in press,1999. 442-8507, Japan. R. Talbot, University of New Hampshire,Durham, NH 03824. F. Arnold and K. Wohlfrom, Max Planck Institute ftir Kernphysik, Postfach103980, D-69029 Heidelberg, Germany. E. Atlas and F. Flocke, National Center for AtmosphericResearch, 1850 Table Mesa Dr., P.O. Box 3000, Boulder, CO 80307. I. Bey, D. Jacob, and R. Yantosca, Harvard University, 29 Oxford (ReceivedApril 22, 1999; revisedJune 29, 1999; St., Cambridge,MA 02138. acceptedJuly 7, 1999.)