Hydrocarbon and halocarbon measurements as photochemical and dynamical indicators of atmospheric hydroxyl, atomic chlorine, and vertical mixing obtained during Lagrangian flights

. Nonmethane hydrocarbons and halocarbons were measured during two Lagrangian experiments conducted in the lower troposphere of the North Atlantic as part of the June 1992, Atlantic Stratosphere Transition Experiment/Marine Aerosol and Gas Exchange (ASTEX/MAGE) expedition. The first experiment was performed in very clean marine air. Meteorological observations indicate that the height of the marine boundary layer rose rapidly, entraining free tropospheric air. However, the free tropospheric and marine boundary layer halocarbon concentrations were too similar to allow this entrainment to be quantified by these measurements. The second Lagrangian experiment took place along the concentration gradient of an aged continental air mass advecting from Europe. The trace gas measurements conf'u'm that the National Center for Atmospheric Research (NCAR) Electra aircraft successfully intercepted the same air mass on consecutive days. Two layers, a surface layer and a mixed layer with chemically distinct compositions, were present within the marine boundary layer. The composition of the free troposphere was very different from that of the mixed layer, making entrainment from the free troposphere evident. Concentrations of the nonmethane hydrocarbons in the Lagrangian surface layer were observed to become depleted relative to the longer-lived tetrachloroethene. A best fit to the observations was calculated using various combinations of the three parameters, loss by reaction with hydroxyl, loss by reaction with chlorine, and/or dilution from the mixed layer. These calculations provided estimated average concentrations in the surface layer for a 5-hour period from dawn to 11 UT of 0.3+0.5 x10 6 molecules cm -3 for HO, and 3.3+1.1 x10 for 4


Introduction
Despite their low atmospheric abundance, hydroxyl radicals (HO) play a crucial role in the oxidation of many important trace atmospheric gases including hydrocarbons, hydrogencontaining halocarbons, and dimethyl sulfide.

Direct measurements of HO have been made by various techniques.
Recently, good agreement has been reported between measured and modeled HO levels [Poppe et al. 1994;Eisele et al., 1994]. However, HO has been directly characterized for a very limited range of field situations. Average concentrations of HO have been calculated indirectly from the atmospheric distribution of species such as methyl chloroform [Singh et al., 1979;Prinn et al., 1995]. In addition, hydroxyl has been indirectly determined from temporal changes in nonmethane hydrocarbon (NMHC) concentrations in well-defined air NOy(gas) + NaCl(aq) ---> XCl(gas) + products (2a) XC1 + hv ---> C1 + products (2b) 4331 The HC1 in (1) is thought to be degassed more efficiently from marine aerosol under polluted conditions, when sulfuric and nitric acids have raised the aerosol acidity [Clegg and Brimblecombe, 1985;Singh and Kasting, 1988]. Reaction (2a) requires nitrate or another NO•, species such as NO, NO0., N9.O 5, or HNO3 to react with NaC1 associated with marine aerosol [Green, 1972;Finlayson-Pitts, 1983;Zetzsch et al., 1988;Finlayson-Pitts et al., 1989;Laux et al., 1994]. In the subsequent reaction (2b), XC1 is an easily photolyzed C1 species such as C12, NOC1, C1NO9., or HOC1, all of which dissociate within about 1 hour after sunrise. Another proposed mechanism for the production of C1 atoms involves the reaction of ozone with sea-salt aerosol in the presence of sunlight Zetzsch, 1989, 1995;Keene et al., 1990] and is consistent with the measurements of Keene et al.
[  Blake et al., , 1994Blake et al., , 1995 In addition to balloons, chemical species can be used as tracers to follow the motion of an air mass advecfing relative to the ground or other air masses. Ideally, a conservative tracer is used that moves passively with the air mass and has no additive or removal processes that are significant over the timescale of the experiment. Changes in the concentration of this tracer then will indicate where the observation was made relative to any spatial gradients present. Tetrachloroethene

The detailed meteorology for this experiment can be found in the work by Bretherton and Pincus [1995] and Bretherton et a/. [1995]. In summary, the MBL over the ocean was generally stratified into a surface layer (SL) below about 500 m, and a mixed layer (ML) (between about 650 and 1700 m).
The boundary region, typically between 500 and 650 m, was frequently marked by scud or broken cumulus (the surface cloud layer). The SL was usually unstable or conditionally stable.
The ML was conditionally stable, indicating that significant convective turbulence could only be observed in the ML when the air was at its saturation point, or associated with cloud dynamics. The boundary between the ML and the free troposphere (FT) was usually marked by thick sheets of stratocumulus clouds. At the beginning of flight 11 the wind in the FT was from the WNW, while the ML wind was predominantly out of the north, producing wind sheer between the two layers. Surface layer winds were initially from the NE but were out of the north by the beginning of flight 12, minimizing any wind sheer between the mixed and surface   in the Lagrangian SL. The first results are derived from changes over the approximately 11 hours between around midnight during flight 11 (from 2246 to 0215, with a median time +1 sigma of 2231 :•.58 rain) and the first part of flight 12 (from 1050 to 1131, median time 1106 +16 min). Even though the initial collection period was relatively long (3.5 hours), the samples could be considered together because little photochemistry is expected to occur at night. The lack of nighttime photochemistry also meant that the HO and C1 concentrations calculated for this time period were scaled to the 5 hours of daylight, i.e., from sunrise at 0606 to 1106. The second set of results was derived from changes over a daytime period of 2 hours and 19 rain between the first part of flight 12 (from 1050 to 1131, 1106 +16 min) and the second part of flight 12 (1322 to 1332, 1325 +3 min). Local solar noon was 1324.
Model description. The theoretical model used to fit the NMHC losses assumed that the SL and ML were vertically well mixed within their respective layers. After removal of samples contaminated by exhaust and ML samples recently affected by entrainment from the FT, the measurements were consistent with this assumption. The vertical column above the ocean did not remain intact; i.e., between flights 11 and 12, the ML moved about 35 ø more to the west than did the SL, thus, only mixing into the SL from the initial or final observed ML could be accounted for. As shown in Table 1 The surface and mixed layers were compared vertically for mixing and temporally for chemical loss, using C2C14 as the principal tracer. The latitudinal gradient of C2C14 and its good correlation with the NMHCs and CH3CC13 (R 2 > 0.9) allowed the Cartesian coordinate system of latitude and longitude to be transformed to a Lagrangian or chemical coordinate system via C2C14. The mixing ratio of C2C14 was then used to refer to a     .............. ................ .... ................ ................ , ............
where w i is the statistical weight based on a linear least squares fit of the hydrocarbon or CH3CC13 and is defined in (6),
The initial C2C14 concentration calculated by the program is obtained by minimizing $ for all of the correlations for the NMHCs and methyl chloroform with C2C14. For example, the initial concentration of C2C14 for the first part of flight 12 is 24.5+0.1 pptv, which corresponds to an initial n-butane concentration of 61.4+0.4 pptv ( Figure 5 and Table 3). The self-consistenfiy calculated C2C14 concentration from the second part of flight 12 was 24.2+0.4 and corresponds to an nbutane concentration of 52.9+2.4 pptv ( Figure 6 and Table 3).
As a guide to how well the various combinations of the above parameters fit the observed losses, the Akaike information criterion (AIC) was calculated for each of these scenarios [Sakamoto et al., 1986]. The AIC takes into account the number of parameters use, d in fitting the data and how well the predicted values agree with observations, as in (7), AIC = n x In(S) + 2m (7) where n is the number of species being fitted and m is the number of parameters used to fit the data. A smaller AIC implies a better fit. However, some caution should be used when relatively few species are being fit by so many parameters. Although the scenarios that best fit the data are quite clear, ranking these scenarios by AIC alone is less certain. Thus, some judgment when interpreting the results is necessary.
HO, CI, and mixing scenarios between flight 11 and the first part of flight 12. Table 4 shows the best fits to the seven different scenarios describing the NMHC and CH3CC13 concentration changes observed during the 11-hour period (5 hours of daylight) between flight 11 (2331:1:58 min) and the first part of flight 12 (1106 +16 min). The HO-only scenario gives a poor fit to the observed losses as evidenced by its high AIC of 30. The AIC for the C1 only case was also poor at 29. Thus HO or C1 chemistry alone does not predict the observations well. When considering mixing only, the best fit (AIC of 24) was obtained when the SL air was completely replaced by ML air.
Varying both HO and C1 in a combined scenario, the AIC value was 31, which is no better than for the HO-or Cl-only chemistry scenarios. However, the HO chemistry plus mixing  Combining all three HO, C1, and mixing terms resulted in a slightly higher AIC of 10. However, this combination is the most reasonable scenario, and the calculated average daytime concentrations for HO and C1 of 0.3+0.5 x106 molecules cm '3 and 3.3+1.1 x104 molecules cm '3, respectively, with mixing estimated at 50+9%, yield a good fit to the observations. In this case the C1 reactions and mixing processes were seen to dominate the aging of this air mass, with the relative unimportance of the calculated HO levels leading to its wide range of possible HO concentrations.
Average HO, C!, and mixing between the first and second parts of flight 12. The second set of fitted Lagrangian results were calculated from NMHC and CH3CC13 concentration changes observed during the 2.3-hours period just before solar noon between the 1106 +16 rain and 1325 +3 rain sampling segments of flight 12. Seven samples were collected in the SL during the first part of flight 12, and these are compared for changes in concentrations with five SL samples from the second part. Table 4 shows that the HO-only case, with an AIC of 1.8, fits most of the observations well.
The C1 only loss mechanism (AIC of 3.2) gives a slightly worse AIC but still yields a good fit to the observed data. In the mixing-only scenario, the best fit was obtained when no mixing was involved. This is because the SL was slightly lower in concentration in four of the six gases, while CH3CC13 and ipentane had similar concentrations as the ML. Mixing did not improve the HO scenario, but the C1 and mixing scenario (AIC of 3.0) predicts significant mixing.
Considering both chemical removal processes together yields an AIC of-0.5. However, with an AIC of-6.1, the combination of HO, C1, and mixing together by far gives the best fit to the observations. This fit was obtained with an HO concentration of 2.6+0.7 x106 molecules cm -3, C1 at 6.5+1.4 x104 molecules cm -3, and mixing of 33+14%. The differences between the expected and observed losses shown in Table 3 were no more than 1 pptv for any of the gases. The average [HO] for this noontime period is much higher (about a factor of 9) than the average for the 5 daylight hours following dawn. This increase is as expected for a photochemically produced radical such as HO. The chlorine concentrations were also higher for the later time period (by about a factor of 2). This indicates that photochemistry plays an important part in the production of C1 and is consistent with the field results of Pszenny et al. [1993] and with recent laboratory experiments by Behnke and Zetzsch [1995] that produced C1 precursors from light, ozone, and sea-salt aerosol in the absence of NOx. However, our levels of CI for the morning period are also consistent with a significant nighttime buildup of C1 atom precursors and a morning release of C1. Additional experiments with better time resolution and under varying conditions would help to elucidate the diurnal C1 cycle.
Other possible loss processes. NMHC loss by reaction with nitrate and bromine radicals was also examined. for the midnight to 1100 period were calculated to be 0.3+0.5 x106 molecules cm -3 for HO, and 3.3+1.1 x104 molecules cm -3 for C1. Noontime concentration estimates were 2.6+0.7 x106 molecules cm -3 for HO and 6.5+1.4 x104 molecules cm -3 for C1. The HO levels vary diurnally as expected, and the higher noon C1 levels also indicate a daytime photochemical source.

Insignificant
However, the fact that the C1 variation was much smaller than for HO is also consistent with a nighttime precursor buildup of easily photolyzed C1 species. This is the first field experiment for which HO, C1, and mixing have been quantified simultaneously. The results suggest that, under the conditions encountered during the second Lagrangian experiment of ASTEX/MAGE, all three processes were important in the photochemical and dynamical aging of this air mass.