Influence of convection and biomass burning outflow on tropospheric chemistry over the tropical Pacific

. Observations over the tropics from the Pacific Exploratory Mission-Tropics A Experiment are analyzed using a one-dimensional model with an explicit formulation for convective transport. Adopting tropical convective mass fluxes from a general circulation model (GCM) yields a large discrepancy between observed and simulated CH3I concentrations. Observations of CH31 imply the convective mass outflux to be more evenly distributed with altitude over the tropical ocean than suggested by the GCM. We find that using a uniform convective turnover lifetime of 20 days in the upper and middle troposphere enables the model to reproduce CH3I observations. The model reproduces observed concentrations of H202 and CH3OOH. Convective transport of CH3OOH from the lower troposphere is estimated to account for 40-80% of CH3OOH concentrations in the upper troposphere. time step for transport and chemistry is 30 s. The model is run for 60 days to obtain a steady state solution. mass continuity equation for a one-dimensional model


Introduction
Convection is a dominant process for the vertical distribution of chemical species in the troposphere. This distribution process, unlike diffusion, is strongly direction oriented. Updrafts bring up air from the lower troposphere into the free troposphere, and compensating subsidence takes free tropospheric air back down into the lower troposphere. Over industrialized continents, convection leads to efficient export of pollutants into the free troposphere and greatly enhances photochemistry outside the polluted continental lower troposphere [Gidel, 1983;Chatfield and Crutzen, 1984;Dickerson et al., 1987;Pickering et al., 1988Pickering et al., , 1992Jaegld et al., 1998].
Over tropical continents, convection exports large amounts of pollutants from biomass burning [Crutzen and Andreae,

Model Description
The model is based on the one-dimensional model by Trainer etal. [1987Trainer etal. [ , 1991 where n i is the concentration of species i, Pi is the chemical production rate, L i is the loss rate by chemical reactions and in the lowest model layer by dry deposition, and •i is the flux due to vertical transport. We consider two types of vertical transport, diffusion and convection. Diffusion, which is the only transport mechanism used by Trainer etal. [1987], represents mixing by random motion of air parcels. In contrast, convection represents direct transport of the lower tropospheric air into the free troposphere followed by compensating subsidence. The need for invoking convective transport in our model becomes apparent in the simulation of CH31 in section 3. Including convective transport also allows us to explicitly simulate wet scavenging of soluble species in the model. Previous one-dimensional models usually invoke a pseudo first-order loss rate constant to account for wet scavenging of soluble species [e.g., Logan where N is the concentration of air. The value of K z in the surface layer (lower than 80 m in altitude) is calculated from similarity theory [Trainer et al., 1987]. The value of K z above the surface layer will be specified. Convective mass fluxes in the model are from the Goddard Institute for Space Studies (GISS) general circulation model (GCM) (version II') as previously used by Prather and Jacob [ 1997]. We will discuss the model vertical transport constrained by CH3I observations in section 3.

Observations
We use observed CH3I concentrations to adjust the parameters of our vertical transport scheme. Methyl iodide is emitted largely from the ocean [Andreae, 1990;Bates et al., 1992;Happell and Wallace, 1996]. It is lost primarily by photolysis. Its photochemical lifetime is about 3 days near the surface and 2 days in the upper troposphere. Its vertical distribution therefore depends critically on the vertical transport of CH3I from the marine lower troposphere. Furthermore, the detection limit of CH3I during PEM-Tropics A is low (0.01 parts per trillion by volume (pptv)) [Cohan et al., 1999], making CH3I observations a valuable constraint on the model transport scheme. Blake et al. [1996] estimated that CH3I emissions from biomass burning during TRACE-A are much smaller than the oceanic source. The vertical profiles of CH3I compiled for the lower, middle, and upper one-third quantiles of C2H2, a tracer for biomass burning, also do not exhibit enhancements of CH3I associated with upper quantiles of C2H 2, suggesting that observed CH3I originates primarily from the ocean. We specified the CH3I concentration at 500-m altitude in the model to the observed value.   [1997]. The convective mass outflux below 10 km is significantly larger in Table 1 than reported by Prather and Jacob [ 1997] due to the addition of entraining convection in the statistics. Figure 1 shows the simulation of CH3I concentrations based on these convective statistics. The value of K z is specified as 10 m s -2 above the surface layer to reproduce the observed CH3I vertical gradient in the lower troposphere. This simulation is much closer to the observations compared with diffusion-only simulations. However, it overestimates CH3I concentrations at 9-12 km but underestimates at 4-7 km, reflecting the "C"-shaped profile of the GISS convective statistics. Similar discrepancies   Table 2. Upper 34 Integrated over the air column of 0-12 km ( [Singh et al., 1995]. We therefore specify acetone concentrations of 200 and 400 pptv for data groups of the lower and upper one-third quantiles of C2H2, respectively. We specify for the middle one-third quantile of C2H2 a concentration of 300 pptv based on a global three-dimensional model simulation by Wang et al. [1998a, b]. The concentration of CH4 is specified at 1.7 ppmv. The DC-8 flight ceiling was at about 12 km; we assume constant mixing ratios above 12 km to be the observed medians at 11-12 km for most species except for water vapor, which decreases with altitude to 6 ppmv at the tropopause. transport agree better with the observations above 10 km than photolysis of CH3OOH and CH20 transported from the lower without convection. Figure 5 shows the fraction of CH3OOH troposphere by convection (for the middle one-third quantile attributed to convective transport in our standard model. The of C2H2). Photolysis of 03 followed by the reaction of fraction is large in the upper troposphere, increasing from 40% OlD+H2 ¸ dominates the primary HOx sources up to 10 km.

Concentrations of OH or HO2 radicals were not measured on the DC-8 during PEM-Tropics
at 10 km up to 80% at 15 km. Below 6 km the fraction drops to The rapid decrease of this source with altitude reflects decreasless than 20%. In this region the increase of CH3OOH from ing H20 levels (Figure 4). The HO x sources from photolysis of convective transport is not enough to offset the decrease due to acetone and convected CH3OOH are less variable with altitude subsidence of air with lower CH3OOH concentrations, result-and surpass that from OlD+H2 ¸ above 11 km. Convective ing in lower CH3OOH concentrations compared with the sim-transport of CH3OOH more than doubles the total primary ulation without convection (Figure 3). source of HOx in the upper troposphere. The source from pho-Photolysis of CH3OOH convected from the lower tropo-tolysis of convected CH20 is a factor of 5-10 smaller than that sphere provides a significant primary HOx source that drives from convected CH3OOH. chemistry faster in the upper troposphere. Figure 6 compares The total HOx production is larger than the sum of its priprimary HOx sources, OlD+H2 ¸, photolysis of acetone, and mary sources because of photolysis of H202 and CH3OOH produced in situ and photolysis of CH20 from CH4 oxidation. Figure 7 compares the total HOx production with and without convection. The increase of total HOx production due to convective transport of CH3OOH is apparent above 10 km by up to a factor of 3. Below 8 km, HOx production with convection tends to be lower than without convection due to lower H202 and CH3OOH concentrations in the convective case ( Figure  3). The concentration of OH shows corresponding changes but with smaller magnitudes (Figure 7); the maximum increase is less than 50% in the upper troposphere. The increase of HOx production due to convection of CH3OOH has a negligible effect on the column mean OH concentration.

Ozone
The shows simulated production and loss rate and the chemical lifetime of 03 for the middle one-third quantile of C2H2. The production rate of 03 increases slightly with altitude and is more uniformly distributed vertically than the loss rate. The loss of 03 occurs primarily in the lowest 5 km. Integrated over the air column of 0-12 km, about 70% of 03 loss occurs at 0-3 km and 20% occurs at 4-5 km. The column 03 production (0-12 km) is 1.7x10 ll molecules cm -2 s -1, less than half of the column 03 loss of 4.5x10 TM molecules cm -2 s -1 (Table 2). About 70% of the 03 loss is due to the reaction of OlD+H2 ¸. The remaining fraction is due to 03 reactions with OH and HO2. The chemical lifetime of 03 increases from 4.5 days near the surface to over 300 days at 15 km.
Air masses with higher concentrations of C2H2 have higher concentrations of NO (Figure 2), which is produced most likely from biomass burning and lightning during convection. Higher NO concentrations tend to increase in situ production of 03. The column 03 production of 0-12 km are 1.4x10 ll, 1.7x10 ll, and 3.0x10 ll molecules cm -2 s -] for the lower, middle, and upper one-third quantiles, respectively, of C2H2 (Table 2). Assuming as previously that the lower one-third quantile of C2H2 data represent air masses not significantly influenced by ouffiow from biomass burning regions, we estimate that the column production of 03 is increased by 45%, or 6.3x10 lø molecules cm -2 s 'l, due to biomass burning outflow.
Our estimate of the column 03 production of 1.7x10 TM molecules cm -2 s -1 compares well with 2.0x10 TM molecules cm -2 s -1 over the tropical North Pacific calculated by Davis et al. concentrations of 03 and H20 selected in this manner tend to be higher than the median concentrations for all the data. The difference is much larger for H20 than 03. For instance, median H20 concentrations are factors of 2 and 6 higher at 2-3 and 4-5 km, respectively, for the middle one-third quantile of C2H2 than for all the data. High H20 concentrations at these altitudes are reflected in the 03 loss rate in Figure 9. The uncertainty in the model estimate of the column 03 loss tends to be larger than that of the column 03 production since the former estimate relies largely on a subset of observations (below 3 km) that also have large variability. The net column 03 deficits (0-12 km) computed in our model are 2.3x10 ll, 2.8x10 l•, and 2.4x10 • molecules cm -2 s -• for the lower, middle, and upper one-third quantiles, respectively, of C2H2 . The largest deficit is for the middle one-third quantile of C2H2 due to a combination of high relative humidity and relatively high 03 concentrations (Figures 2 and 4). Schultz et al. [1999] suggested that the large 03 deficit is supplied mostly by longitudinal transport of 03 into the region. We conduct a sensitivity simulation; the concentration of 03 at 11-12 km is specified as observed, below which altitude 03 concentrations are determined by chemical production and loss and a downward flux of 03 from the upper troposphere due to subsidence. The resulting 03 concentrations are much lower than the observations (Figure 10). The 03 column of 0-12 km in the sensitivity simulation is 50% lower, suggesting large amounts of lateral import of 03 in the region, in agreement with Schultz et al. [1999].

Nitrogen Oxides
Simulated HNO3 concentrations are generally within the range of observations for the lower and middle one-third quantiles of C2H2 but are about a factor of 2 higher above 4 km for the upper one-third quantile of C2H2 (Figure 11). Using rate constants for reactions of OH with NO2 and HNO3 from Brown et al. [1999a, b] H202 by 50-100% is necessary in the model to match observations. Model results indicate that 40-80% of CH3OOH in the upper troposphere is due to convective transport from the lower troposphere. Photolysis of convected CH3OOH more than doubles the primary source of HOx in the region. The reaction of OlD+H2 ¸, which is the dominant primary HOx source in the lower and middle troposphere, is surpassed by photolysis of convected CH3OOH and photolysis of acetone above 11 km. The latter two sources are also much larger than that from photolysis of convected CH20. Convection of CH3OOH increases the total HOx source and OH concentration by up to a factor of 3 and 50%, respectively, in the upper troposphere; it increases the production rate of 03 by about 0.4 ppbv d -1 above 11 km. Integrating over the air column of 0-16 km, we find that convection of CH3OOH has a negligible effect on the column mean OH concentration and increases column 03 production by only 4%.
In comparison, the effect of biomass burning outflow is much more significant in changing chemistry over the tropical Pacific. We group observations into three equally divided quantlies of C2H2 concentrations, because C2H2 is a good tracer for biomass burning outflow during PEM-Tropics A. Using the lower one-third quantile of C2H2 data as the air masses least influenced by biomass burning outflow, we estimate that 60% (or 9 DU) of 03 and 75% NOx enhancements in the air column of 0-12 km are due to import of pollutants from biomass burning regions. The NOx enhancement increases column 03 production (0-12 km) by 45%. The import of pollutants increases the column mean OH concentration by a moderate 7% due in part to the offsetting effects of concurrent increases in NOx and CO concentrations.
The effect of pollutant import into the region is reflected in the budget and concentrations of long-lived chemical species. Whereas short-lived H202 and CH3OOH concentrations are generally reproduced based on in situ chemistry and local convective transport within the PEM-Tropics A region, the budget of longer-lived 03 shows a column 03 loss a factor of 2 larger than column 03 productionß A model sensitivity study indicates that 50% of 03 is supplied by lateral import, similar to the finding by Schultz et al. [1999]. Peroxyacetylnitrate has a lifetime of about 1 month in the tropical upper troposphere. Model simulations are in much better agreement with PAN observations for the lower one-third quantile of C2H 2 than those for the middle and upper one-third quantlies of C2H2.
The underestimate of PAN concentrations for the middle and upper one-third quantlies of C2H 2 is consistent with pollutant import into the region. Nitric acid also has a long chemical lifetime but is removed efficiently by wet scavenging. As a result, significant direct import of HNO 3 from biomass burning regions is unlikely. Model results are in reasonable agreement with observations for the lower and middle one-third quantlies of C2H2. However, the model overestimates HNO3 concentrations by about a factor of 2 above 4 km for the upper one-third quantile of C2H 2, as observed previously in other regions of the globe. This discrepancy cannot be explained by recent kinetic data for the reactions of OH with HNO3 and NO2 [Brown et al., 1999a, b]. There are two likely explanations: (1) the outflow from biomass burning regions has not reached steady state with respect to HNO3; and (2) HNO 3 is recycled rapidly back to NOx by some heterogeneous mechanism in biomass burning outflow.