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

,

compare these observations with model predictions.We also provide an assessment of the sources and sinks of these chemicals.

Discussion of Results
It is pertinent to note that during SONEX attention was focused on the UT region of the atmosphere that has been previously minimally studied.SONEX objectives also were such that much of the data were collected between 40ø-65øN latitudes with only one flight going as far south as 20øN.The stratosphere was frequently sampled although O_• concentrations never exceeded 450 ppb, and N20 was never below 300 ppb.This region of the stratosphere has been commonly defined as the "lowermost stratosphere" (LS).With this perspective in mind, a greater emphasis is placed on the UT/LS regions.Only SONEX data collected on the DC-8 are used in this study.POLlNAT Falcon data are occasionally considered for purposes of comparisons.
The SONEX NOy data were examined against several tracers that may be used to distinguish the stratosphere from the troposphere (03, N20 , and CO).It was determined that a simple filter (O3>100 ppb and Z>6 km) was adequate to define periods of stratospheric influences.This filter has been applied to all tropospheric/stratospheric observations discussed here.Tro-pospheric observations have also been compared with the Harvard 3-D global model (4 ø latitude x 5 ø longitude grid size), an earlier version of which has been described by Wang et al. [1998a, b].The present version has improved descriptions of chemistry and emissions, and higher vertical resolution (20 layers versus 9 layers) making it more appropriate for upper tropospheric simulations.The model transports 24 tracers, including odd oxygen (O x = O 3 + NO 2 + 2NO3), NO x, HNO3, N205, HNO4, three peroxyacyl nitrates, a lumped alkyl nitrate, CO, alkanes (ethane, propane, higher alkanes), alkenes, isoprene, carbonyls (acetone and higher alkyl ketones, formaldehyde, and higher alkyl aldehydes, methacrolein, methylvinyl ketone), and peroxides (H202, CH3OOH ).A present shortcoming is that the model contains 1993 meteorology.Compari- within the normal year-to-year variability, 1997 was a climatologically average year.In the following section, we shall discuss the distribution, sources, and sinks of selected organic species in the troposphere and the LS.

Tropospheric Distributions
Figure 1  Figure 2 provides a perspective on the abundance and distribution of total oxygenated chemicals (:SOx-organic) in relation to that of nonmethane hydrocarbons (:ENMHC)measured during SONEX.These NMHC included a large number of alkanes, alkynes, alkenes, and aromatics.It is evident from Figure 2 that in the remote troposphere, oxygenated organics are collectively as abundant as NMHCs.Because many of them are easily photolyzed at tropospheric wavelengths, their influence on upper tropospheric chemistry can be far greater than that of NMHC.As will be discussed in the following sections, the sources, sinks, and chemistry of these oxygenated molecules can differ greatly from each other.In many cases, significant biogenic sources appear to be present.

2.1.1.
Organic nitrates.An onboard gas chromatograph (GC)operated by the NASA Ames group measured peroxyacetyl nitrate (PAN), peroxypropionyl nitrate (PPN), methyl nitrate (MN), and ethyl/i-propyl nitrates in real time.The stable alkyl nitrates (methyl, ethyl, i-propyl, and 2-butyl) were also measured from the whole air samples collected by the University of California, Irvine group.Figure 3 shows a scatterplot of these measurements obtained by the two groups during the SONEX mission.Recognizing that 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 measurements is considered quite good.Statistical measures of ensemble averages indicated that these two data sets could be combined for all intents and purposes.PAN measurements were not duplicated on the DC-8, and it was also not measured on the POLINAT-2/Falcon, so no comparison is possible.
Figure 4 shows the tropospheric distribution of reactive organic nitrogen species along with that of NOy, a measure of the total available reactive nitrogen.Among this group of organic nitrates, PAN was the most abundant at --200 ppt in the lower troposphere and --50 ppt in the UT (Figures 4a and 4b).
Typically PAN accounted for 10-30% of the available NOy.
The higher concentrations in the lower troposphere are indicative of continental pollution influences over the Atlantic.distribution of PPN (Figures 4c and 4d)is similar to that of PAN, but its mixing ratios are significantly lower.This is largely due to the lesser abundance of PPN precursors as well as its somewhat shorter lifetime.Unlike PAN and PPN, methyl nitrate (MN)was nearly uniformly distributed in the troposphere (Figures 4c and 4d).It was clear that a large oceanic source, as observed in the tropical pacific (H.Singh, unpublished data, 1996), was not present in this region.Collectively, total alkyl nitrates (TAN) were present at relatively low concentrations with a distribution that was suggestive of surface, probably anthropogenic, sources (Figures 4c and 4d).Total alkyl nitrates on average accounted for less than 3% of reactive nitrogen (Nay).All of these species (pernitrates and nitrates) are thermally and photochemically dissociatable and can act both as a source or a sink of the available free NOx [Singh, 1987].

2.1.2.
Carbonyls.Acetone and formaldehyde, two of the most important carbonyls in the atmosphere, were measured during SONEX.In recent years, acetone has been shown to be a globally abundant species that has the potential to both sequester reactive nitrogen as PAN and provide an upper tropospheric source of HO x radicals [Singh et al., 1994[Singh et al., , 1995;;Arnold et al., 1997].Similarly, formaldehyde is a well known product of methane oxidation and a major secondary source of HOx radicals [Crawford et al., 1999].Acetone was measured on the DC-8 with an onboard GC operated by the NASA Ames group and on the Falcon by the Max-Planck-Institute Heidelberg group who used a technique involving Chemical Ionization Mass Spectrometry (CIMS) [Arnold et al., 1997;Wohlfrom et al. 1999].Both instruments quoted substantial error bars of_+25% (GC) and _+50% (CIMS).The CIMS instrument was not directly calibrated, and the reaction of product ions with water vapor was recognized as a potential source of error.The GC method collected a sample over a 150-s period, while the CIMS instrument had a much faster response time (10 s). Figure 5 shows an example of the acetone distribution measured by the two aircraft when they were in the vicinity of each other during the two missions.The altitudes of the two aircraft are also shown, and the CIMS data have been averaged over the sampling time of the DC-8 instrument.During mission 5 (Figure 5a), the two instruments produced comparable data and saw acetone concentration decrease in the stratosphere (at 11 km).However, the decline observed by the Falcon was less than that observed by the DC-8.Similarly,' during mission 7, the DC-8 measurement is some 30% lower than the CIMS measurement.These results are within the quoted error bars of these instruments.Figure 6 1999).Although CH3OH is easily synthesized in seawater from the hydrolysis of methyl halides, no estimate of its oceanic source is presently available.Table 2 summarizes a first inventory of the global sources (=122 Tg yr-') of methanol indicating that its sources far exceed its known sinks.If this estimate of the global methanol source is found to be correct, then removal mechanisms other than OH oxidation (and slow deposition) must play an important role.In the boundary layer, aerosol media containing OH, C1, and SO4 could react with methanol to form complex products.In the middle and UT, methanol may also react in clouds to oxidize into organic acids or react with U2SO 4 in cloud droplets to form methanesulphonic acid [Murphy et al., 1998].Suffice it to say that methanol is a globally abundant organic species with limited atmospheric data and poorly understood sources and removal processes.

2.1.5.
Organic acids.Formic and acetic acids are ubiquitous trace gases in the atmosphere and contribute a large fraction of the free acidity in precipitation in remote areas [Keene and Galloway, 1988].During SONEX, formic acid was present in concentrations of =45 ppt in the UT and =120 ppt in the lower troposphere (Figure 9).The mean abundance of acetic acid was less than half that of formic acid.Unlike formic acid, a large fraction of the acetic acid data (50% between 4-8  below 7 km.Primary emissions from terrestrial vegetation and from combustion of fuel and biomass, and secondary formation from gas phase and liquid reactions are known to be present [Talbot et al., 1990;Fall;1999;Yokelson et al., 1999].Although acetic acid could react with OH radicals to form HCHO (CH•COOH + OH ---> CH302 ---> CH3OOH and HCHO), the process is quite slow, and this is at best a minor source of HOx radicals.In most cases, organic acids are removed from the atmosphere by wet and dry deposition processes, and their chemistry is at best highly uncertain [Talbot et al., 1990[Talbot et al., , 1995;;Jacob et al., 1996].An accurate enough source inventory for these organic acids is not currently available to model their global distribution.

Stratospheric Distributions
The stratosphere was frequently penetrated during SONEX.Chemical indicators of the stratospheric air suggested that these instances represented the "lowermost" stratosphere (03<450 ppb; N20>300 ppb). Figure 10  In nearly all cases, mixing ratios of oxygenated organic species declined rapidly above the tropopause.Deeper into the stratosphere, mixing ratios of acids, peroxides, formaldehyde, and methanol were extremely low and near the limit of detection as 03 approached 400 ppb. Figure 11 shows the distribution of some selected oxygenates as a function of Os.PAN and acetone declined at a slower rate than methanol and other NMHC (e.g., C2H6).In this region, which is believed to have --10 s molecules cm -3 of C1 atoms, C1 chemistry instead of OH dominates the oxidation process.PAN, however, reacts negligibly slowly with C1 atoms and may have a potential local source resulting from C3Hs/C2H6-C1 and acetone photochemistry [Demore et al., 1997].Acetone itself can also be produced from C3H8-C1 oxidation in this region.The slower net loss rate of acetone compared to methanol is evident from the fact that the acetone to methanol ratio in the stratosphere increased from about 2 to 8 as O s increased from 100 to 400 ppb.There is some indication, based on extremely sparse data, that acetic acid increased in the LS from 40 ppt (03 --100-150 ppb) to 8 0 ppt (03 =300-400 ppb).This is consistent with the view that acetone photolysis followed by reaction of peroxyacetyl radicals with HO 2 (CH3CO 3 + HO 2 ---) CH3COOH + 03) is a possible in situ source of the lower stratospheric acetic acid [Singh et al., 1997].Table 3 summarizes the mixing ratios of a number of select trace species as measured in the-LS during SONEX.

Figure 1 .
Figure 1 shows the vertical distribution of 03 and CO (a useful tracer) as observed during SONEX over the North Atlantic

FigureFigure 4 .
Figure 4 also shows a comparison of modeled PAN and NOy with SONEX observations with reasonable agreement.The

Figure 7 .Figure 8 .
Figure 5.A comparison of acetone as measured on the DC-8 (solid circles) and the Falcon (solid squares) using two independent techniques.The altitude profile DC-8 (solid line) and Falcon-20 (dashed line) is shown.

Figure 11 .
Figure 11.Mixing ratios of selected oxygenated species as a function of O s in the "lowermost" stratosphere.