Convective injection and photochemical decay of peroxides in the tropical upper troposphere: Methyl iodide as a tracer of marine convection

. The convective injection and subsequent fate of the peroxides H202 and CH3OOH in the upper troposphere is investigated using aircraft observations from the NASA Pacific Exploratory Mission-Tropics A (PEM-Tropics A) over the South Pacific up to 12 km altitude. Fresh convective outflow is identified by high CH3I concentrations; CH3I is an excellent tracer of marine convection because of its relatively uniform marine boundary layer concentration, relatively well-defined atmospheric lifetime against photolysis, and high sensitivity of meas-urement. We f'md that mixing ratios of CH3OOH in convective outflow at 8-12 km altitude are enhanced on average by a factor of 6 relative to background, while mixing ratios of H202 are enhanced by less than a factor of 2. The scavenging efficiency of H202 in the precipitation associated with deep convection is estimated to be 55-70%. Scavenging of CH3OOH is negligible. Photolysis of convected peroxides is a major source of the HO(cid:127) radical family (OH + peroxy radicals) in convective outflow. The timescale for decay of the convective enhancement of peroxides in the upper troposphere is determined using CH3I as a chemical clock and is interpreted using photochemical model calculations. Decline of CH3OOH takes place on a timescale of a 12 days, but the resulting HOx converts to H202, so H202 mixing ratios show no decline for -(cid:127)5 days following a convective event. The perturbation Means and standard deviations of concentrations measured at 0ø-30øS latitude during PEM-Tropics A. The following selection criteria were applied: ex-hibits similar traits to CH3I and could also serve as a tracer of marine convection. that concentrations of CH3OOH Photochemical model results for the PEM-Tropics A conditions indicate a 50% enhancement of HOx in convective water vapor and half is


Introduction
Updrafts in deep convective clouds can raise boundary layer air into the upper troposphere in a matter of minutes; outflow occurs primarily from the cloud top and anvil [Chatfield and Crutzen, 1984]. Model studies have suggested that convective injection of the peroxides H202 and CH3OOH could provide an important source of hydrogen oxide radicals (HOx = OH + peroxy radicals) to the upper troposphere, resulting in enhanced production of 03 and gas-phase H2804 [Chatfield and Crutzen, 1984;Prather and Jacob, 1997]. Jaegld et al. [1997,1998] found that recent observations of OH and HO2 in the upper troposphere Wennberg et al., 1998] are consistent with a major source of HO• from convective injection of peroxides and formaldehyde. In the present study we use aircraft observations of CH3I, H202, and CH3OOH taken up to 12 km altitude over the tropical South Pacific to investigate the convective injection and Copyright 1999 by the American Geophysical Union.
Paper number 98JD01963. 0148-0227/99/98JD-01963509.00 subsequent chemical decay of peroxides in the upper troposphere. As previously shown by Davis et al. [1996] and further demonstrated here, CH3I provides a sensitive tracer of deep marine convection.
The peroxides H202 and CH3OOH are produced in the atmosphere by combination reactions of the HO• radicals, HO2 + HO2 and CH302 + HO2, respectively. They photolyze on a timescale of the order of 1 day to regenerate HOx radicals and thus serve as reservoirs for HO•. Water vapor is the main source of HO• in most of the troposphere, so the abundance of peroxides is correlated in general with humidity. Mixing ratios of H202 and CH3OOH are of the order of 1000 parts per trillion by volume (pptv) in the marine boundary layer and 100 pptv in the upper troposphere . The large concentration gradient between the boundary layer and the upper troposphere, combined with the 10-day characteristic time for overturning of the upper troposphere with boundary layer air in the tropics [Prather and Jacob, 1997 CHaOOH to be scavenged in the precipitation associated with the convective updraft [Chatfield and Crutzen, 1984]. In the upper troposphere the peroxides photolyze to release HO•, and subsequent cycling takes place within the HOy family (sum of HO• and peroxides). Eventual conversion of HOy back to water vapor terminates the process. A schematic of the resulting life cycle for HOy is shown in Figure 1.

The observations analyzed in this paper are from the Pacific Exploratory Mission-Tropics A (PEM-Tropics A) flown in
September-October 1996 [Hoell et al., this issue]. PEM-Tropics A used two aircraft, a DC-8 and a P-3B, to survey atmospheric composition over a broad .expanse of the Pacific from 45øN to 72øS. Most of the data were collected between 0øS and 30øS and extended zonally across the South Pacific. We limit our attention to data from the DC-8, which had a higher ceiling (12 km)than the P-3B (7 km). Measurements aboard the DC-8 included H202, CH3OOH, CH3I, and a number of other species. We use CH3I together with high relative humidity as a tracer of fresh outflow of marine convection in the upper troposphere (section 2). From there we examine the enhancement of peroxides and other species in the convective outflow and estimate scavenging efficiencies in the precipitation associated with deep convection (.section 3). We then use CH3I as a chemical clock to determine the timescale for decay of HOy in the upper troposphere following convection and interpret the results with a photochemical model calculation (section 4). Conclusions are in section 5.

Convection
Methyl iodide (CH3I) is emitted ubiquitously by the oceans. Though biological production of CH3I may be important in coastal and upwelling regions, photochemical reactions of methyl radicals and iodine atoms in seawater are thought to be the dominant marine source [Moore and Zafiriou, 1994; Happell and Wallace, 1996; Manley and de la Cuesta, 1997]. Inde.ed, in the marine boundary layer during PEM-Tropics A, concentrations of methyl iodide were not correlated with concentrations of biologically produced marine tracers such as DMS. Biomass burning is thought to be a much sma.!ler source of CH3I emissions globally [Andreae et al., 1996], but as discussed below, its impacts are nonnegligible even over the remote Pacific.
Methyl iodide is removed from the atmosphere mainly by photolysis, with a mean lifetime of 4 days in the tropical troposphere ( Figure 2). Oxidation by OH accounts for only -• 1% of the loss from photolysis. The Henry's Law constant of CH3I is sufficiently low (K• = 0.14 M atm '• at room temperature [Moore et al., 1995]) that rain-out is negligible.
Observations in PEM-Tropics A indicate relatively uniform concentrations of CH3I in the marine boundary layer (MBL) over the tropical South Pacific, with an interquartile range of 0.21-0.44 pptv at 0-2 km altitude ( Figure 3). This is a much narrower range than that observed for other marine tracers, such as DMS, which have highly variable biological sources [Andreae et al., 1985]. Atmospheric measurement. s of CH3I concentrations can be made with high sensitivity (detection limit of 0.01 pptv) [D. . The combination of relatively uniform boundary layer concentra- Interference from biomass burning must be considered when using CH3I as a tracer of marine convection. Most measurements of elevated CH3I at 8-12 km altitude (>0.11 pptv, top octile) displayed corroborating signs of recent marine convection: high humidity, high mixing ratios ofbromoform, and low O3 mixing ratios (Table 1). However, as shown in Table 1, some of the high-CH3I measurements were associated with low humidity (<10% with respect to ice) and high C2H2 mixing ratios (80-300 pptv), in-dicating biomass burning pollution rather than marine convection as the source of CH3I. To distinguish recent marine convection from biomass burning pollution in the PEM-Tropics A data at 8-12 km altitude, we used relative humidity as a corroborating tracer of convection (Table 1). All points with both CH3I and relative humidity in the top octile (CH3I > 0.11 pptv, relative humidity >50% with respect to ice) also had elevated mixing ratios of bromoform as well as low C2H2, 03, and NO.

a NO x = NO + NO9_; NO9_ is calculated with a photochemical steady-state model [Schultz et al., this issue].
b Relative humidity is defined with respect to ice. c Detection limit for the DMS measurements.
The role of convection in enhancing CH3OOH concentrations in the upper troposphere is evident from Table 1. The mean CH3OOH mixing ratio above 8 km is 6 times higher in convective outflow than in background air. The mean H202 mixing ratio in convective outflow is also elevated but by less than a factor of 2. The CH3OOH/CH3I concentration ratio in fresh convective outflow is similar to that in the boundary layer, indicating no significant scavenging of CH3OOH in the precipitation associated with deep convection.
We estimate the scavenging efficiency of H202 in deep convection by modeling the observed composition of the fresh convective outflow (cony) in Table 1 (

1-[•)YBL
Scavenging efficiencies for H202, H20, and SO2 in deep convection are given in Table 2 using the mean observations in Table 1 and either of the reference tracers CHBr3, CH3OOH, or CH3I to derive the dilution factor 13. Changes in c• depending on the reference tracer used give some measure of the uncertainty of the approach. We derive in this manner an H202 scavenging efficiency of 55-70%. A higher scavenging efficiency is found for SO2 (65-95%), presumably reflecting oxidation by H202 in convective clouds. Water vapor is even more efficiently scavenged (90%) because of its low vapor pressure at convective outflow temperatures.

Fate of Peroxides in the Upper Troposphere
When injected in the upper troposphere, the peroxides decay and supply a source of HOx (Figure 1). The timescale for decay of the peroxide enhancement following a convective event can be estimated in the PEM-Tropics A data by plotting the peroxide mixing ratios versus the CH3I mixing ratio taken as a time coordinate (Figure 6). Also shown in Figure 6 are results from a time-dependent, zero-dimensional photochemical model simulation [Schultz et al., 1998] for an air parcel initialized with the fresh convective outflow composition in Table  1. Acetone and CH20, not measured in PEM-Tropics A, are initialized with mixing ratios of 400 pptv and 60 pptv, respectively [Wang et al., 1998]. The NOx concentration is held constant in the simulation and no dilution of the air parcel with time is allowed. The 03 column is 6.7 x 10 •8 molecules cm -2, average of tropical observations in PEM-Tropics A. Circles plotted every 24 hours on the model curves in Figure 6 convert the CH3I coordinate to time. The trend of peroxide versus CH3I concentrations computed with the model is roughly consistent with the observations although the scatter in the observations is large. We see from Figure 6 that observed concentrations of CH3OOH decline on a timescale of 1-2 days following a convective event. In contrast, observed concentrations of H202 show no significant decline for---5 days following convection and then decline to steady state. We explain the longer persistence of H202 as reflecting the cycling within the HOy family ( Figure 1) days in fresh convective outflow to 1.5 days in aged air. The reason is that photolysis of the peroxides in the outflow produces OH, which oxidizes CH4 to produce CH20; photolysis of this CH20 then regenerates HOy, thus sustaining the HOy enhancement in the outflow for longer than would be expected from the standard calculation of HOy lifetime. The effective efolding lifetime of the HOy enhancement for the PEM-Tropics A conditions is still much shorter than the 6-day value reported by daegld et al. [1997] in model calculations for the wintertime upper troposphere over Hawaii. Higher Sun angles in PEM-Tropics A are the principal factor for this difference; lower 03 columns and higher NOx concentrations (which facilitate reaction (5) by increasing the OH/HO2 concentration ratio) also contribute.
We find in our model that HOx concentrations in convective outflow are 50% higher than in the background upper troposphere (Table 3). About half of this enhancement is due to convected water vapor, and half is due to convected peroxides (we do not account for convective enhancements of acetone or formaldehyde due to lack of observations). Larger relative enhancements of HOx in convective outflow were found in previous studies [daegld et al., 1997, 1998]. The weaker effect in PEM-Tropics A is due to high Sun angles, low 03 columns, and high humidities, which maintained a

Summary and Conclusions
We used aircraft observations over the tropical South Pacific up to 12 km altitude to examine the deep convective injection of peroxides to the upper troposphere and the subsequent chemical decay of these peroxides. Convective outflow at 8-12 km was identified by a combination of elevated CH3I and elevated relative humidity. CH3I is an excellent tracer of marine convection in the upper troposphere because of its relatively uniform marine boundary layer concentrations, its relatively well-defined atmospheric lifetime (photolysis is the main sink, with a lifetime of about 4 days in the tropics), and its low detection limit. Interference from biomass burning pollution is a problem but can be screened out using concurrent observations of high relative humidity and C2H2. Though not used extensively in this study, CHBr3 exhibits similar traits to CH3I and could also serve as a tracer of marine convection.
We found that concentrations of CH3OOH in convective outflow at 8-12 km altitude were elevated by a factor of 6 relative to the upper tropospheric background, while concentrations of H202 were elevated by less than a factor of 2. Scavenging by precipitation in the convective updraft was negligible for CH3OOH and 55-70% for H202. Photolysis of convected peroxides was a major source of HOx in the convective outflow.
Formaldehyde is an additional HOx precursor injected to the upper troposphere by deep convection, but no measurements of CH20 were made in PEM-Tropics A. If boundary layer CH20 mixing ratios were 600-1000 pptv, as found in some shipboard measurements over the equatorial Pacific JArlander et al., 1990], then convective pumping of CH20 would be of comparable importance to peroxides as a source of HOx in convective outflow. However, models and most observations indicate CH20 mixing ratios of 100-300 pptv in the tropical marine boundary layer . At these levels, transport of CH20 in deep marine convection provides only a minor source of HOx in the upper troposphere [Prather and Jacob, 1997;daegld et al., 1997].
We estimated the rate of chemical decay of the peroxides in the upper troposphere following convective injection using concurrent observations of H202, CH3OOH, and CH3I concentrations, with CH3I serving as a photochemical clock. Results were compared to a photochemical model simulation. Concentrations of CH3OOH declined with an e-folding lifetime of 1-2 days due to losses from photolysis and reaction with OH. The HOx produced from photolysis of CH3OOH was recycled to H202. Concentrations of H202 thus did not decline for about 5 days following convective injection. The perturbation to HOx from the convective injection of peroxides decayed on a timescale of 2-3 days; this timescale is relatively short because of the intense radiation over the tropical South Pacific. Photochemical