Effects of biomass burning on summertime nonmethane hydrocarbon concentrations in the Canadian wetlands

. Approximately 900 whole air samples were collected and assayed for selected C2-C10 hydrocarbons and seven halocarbons during the 5-week Arctic Boundary Layer Expedition (ABLE) 3B conducted in eastern Canadian wetland areas. In more than half of the 46 vertical profiles flown, enhanced nonmethane hydrocarbon (NMHC) concentrations attributable to plumes from Canadian forest fires were observed. Urban plumes, also enhanced in many NMHCs, were separately identified by their high correlation with elevated levels of perchloroethene. Emission factors relative to ethane were determined for 21 hydrocarbons released from Canadian biomass burning. Using these data for ethane, ethyne, propane, n-butane, and carbon monoxide enhancements from the literature, global emissions of these four NMHCs were estimated. Because of its very short atmospheric lifetime and its below detection limit background mixing ratio, 1,3-butadiene is an excellent indicator of recent combustion. No statistically significant emissions of nitrous oxide, isoprene, or CFC 12 were observed in the biomass-burning plumes encountered during ABLE 3B. The presence of the short-lived biogenically emitted isoprene at altitudes as high as 3000 m implies that mixing within the planetary boundary layer (PBL) was rapid. Although background levels of the longer-lived NMHCs in this Canadian region increase during the fire season, isoprene still dominated local hydroxyl radical photochemistry within the PBL except in the immediate vicinity of active fires. The average biomass-burning emission ratios for hydrocarbons from an active fire sampled within minutes of combustion were, relative to ethane, ethene, 2.45; ethyne 0.57; propane, 0.25; propene, 0.73; propyne, 0.06; n-butane, 0.09;/-butane, 0.01; 1-butene, 0.14; cis-2-butene, 0.02; trans-2-butene, 0.03; i-butylene, 0.07; 1,3-butadiene, 0.12; n-pentane, 0.05; i-pentane, 0.03; 1-pentene, 0.06; n-hexane, 0.05; 1-hexene, 0.07; benzene, 0.37; toluene, 0.16. and halocarbons were observed. These studies confirm that

use in the NASA ABLE 3A experiment conducted in Alaska . On first use, these canisters were quite satisfactory for alkanes, acetylene, and several halocarbons, but the light olefin concentrations tended to grow with time during repetitive measurements of successive aliquots from individual canisters. After the ABLE 3A experiments and prior to ABLE 3B the canisters were filled with ambient air and baked with the valves open following the temperature and time sequence previously mentioned. The alkene stability problems were solved by this ambient air baking.
Canister sample integrity studies included repetitive measurements of individual samples from the aircraft. Because the time delay between collection and analysis was always in the hours to days range for aircraft samples, a separate test of integrity was made on a ground level sample. Air from the pine woods immediately outside the Canadore laboratory was introduced at ambient pressure into an evacuated 2-L canister. Within 2 minutes a 250-ml (STP) sample was preconcentrated on the glass bead loop and injected onto the separation columns. This sample was assayed subsequently 1 hour, 24 hours and 3 weeks later. No significant deviations (+_ l•r) in any of the mixing ratios of the reported NMHCs and halocarbons were observed. These studies confirm that the canisters maintained sample integrity for at least several weeks. No direct information is available about possible alterations in trace gas concentrations occurring in the canisters during the 2 min between collection and analysis, but the consistency of the results suggests any such concentration variations to be negligible. The stability in these canisters of C2H 4, C3H6, and C5H8, all of which have significant reactivity with 03, suggests that ambient 03 is removed by the metal surfaces of the canister almost immediately after collection, without involving the olefins present in much smaller quantities. These results are consistent with those reported previously [Greenberg et al., 1992].

Sample Filling Procedure
Outside air was brought into the NASA Electra through a 1/2-inch stainless steel air intake mounted on the port side of the fuselage, forward of the wing section, and extending out 8 inches, beyond the boundary layer of the Electra. The inlet was attached to a two-stage metal bellows pump (Metal Bellows Company, MB-602) connected in series allowing for pressurization to 40 (psig). The canisters were configured in three columns of eight each via 1/4-inch stainless steel ultratorr tees equipped with VITON O rings and 1/4-inch stainless steel tubing. These 24-canister "snakes" formed the basic unit for handling operations on the NASA Electra and for air shipments when the Electra was not based at the North Bay airport, adjacent to the Canadore laboratory. A gas-handling manifold mounted on top of the rack directed the airflow to any one of the three 24-canister snakes connected in parallel. The exhaust line was mounted above the top of the fuselage. Once the Electra was airborne, the 200-ml manifold and snake were flushed with several hundred liters of ambient air prior to filling the first canister. This procedure adequately expelled any cabin air which had been incorporated in the lines during the installation of canister sets prior to the commencement of the mission. The time required to pressurize a sample to 40 psig ranged from 10 to 60 s depending on the ambient pressure associated with a particular altitude. Therefore depending on the speed and/or ascent/descent rate of the Electra, each sample represents an air mass collected over a linear distance of several kilometers or a vertical distance of as much as 150 m.
Prior to the beginning of the aircraft missions, pump integrity and contamination studies using humidified zero air were performed on the entire manifold system. The zero air was passed through the inlet, pump, and manifold before collection in the canister, exactly as ambient samples were collected on the Electra in flight. Test results showed no increase in any mixing ratios of reported NMHC and halocarbons. Additionally, pressure dependent studies were performed to simulate pump performance at different altitudes, and the results again showed no significant deviations in mixing ratios of the measured trace gases.
Samples were usually analyzed within 4 days and always within 7 days of collection. Some of the delay in analysis resulted from the necessity for shipping of air canisters via commercial air cargo from the Electra area of operation to the Canadore laboratory. The logistics for the simultaneous arrival of 72 newly filled canisters inevitably caused as much as a 30-hour delay in analysis for the last few canisters in a shipment. Once the samples were assayed, the canisters were evacuated to a pressure of 10 -2 torr and shipped back to the Electra for subsequent use.

Chemical Analysis
Each air sample was analyzed for NMHCs and halocarbons utilizing a trace gas analytical system composed of three separate gas chromatographic columns enclosed in two independently programmed temperature-controlled GC ovens (see Figure 1). Two of the columns were equipped with flame ionization detectors (FID) and the third with an electron capture detector (ECD). Approximately 650 torr of air was transferred from the canister to a 2-L stainless steel storage vessel permanently fixed to the vacuum line. The less volatile trace gases were preconcentrated on a loop (24 inches x 1/8-inch OD stainless steel) filled with 1-mmdiameter glass beads and immersed in liquid nitrogen. This procedure allowed the separation of the trace gases from the bulk N2, 02, and Ar but did not quantitatively trap CH 4. After 500 torr of sample in the storage vessel plus line (1275 cm 3 at STP) had passed through the preconcentration trap, the loop and its contents were isolated via an 1/8 inch UWP six-port switching valve with 0.03-inch orifices (Valco Instruments) and heated with hot water (40ø-60øC) to revolatilize the trace gases. The H2 carrier was redirected to flush the previously trapped contents from the loop to the splitter box which partitioned the gas flow to the three different columns. The 1/16-inch SS tubing from the switching valve to the splitter box was connected to a 1/16-inch SS cross. Different lengths of 0.25-mm deactivated fused silica were connected to the cross. The length depended upon how much restriction was necessary to yield the desired flows to each column. The precision obtained during this study indicated that the partitioning of flows was extremely reproducible. Forty-five percent of the flow was directed onto a 50-m 0.32-mm A120 3 plot column (Chrompak), 30% to a 60-m 0.53-mm DB-1 column (J & W Scientific) and 25% to a 75-m 0.53-mm DB-624 column (J & W Scientific). The plot column, used for C2-C7 analysis, was attached to one of the FIDs in GC 1 and the DB-1 column, used for C4-C•0 separations, was plumbed into the second FID of GC 1. The DB-624 column in GC 2 was attached to an ECD and was used to characterize the C•-C2 CFCs and several halocar-  The trace gas analytical system was interfaced to three Spectra Physics Chromjet computing integrators and an IBM PS2 computer for data acquisition, storage, and reduction. While the data output direct from the integrators was sufficiently useful for preliminary data, the occasional variations in detectable amounts in the working standard. Nonmethane hydrocarbon reference standards with a stated accuracy of 2% (95% confidence limit) were obtained from Scott Specialty Gases (Plumsteadville, Pennsylvania). The mixing ratio of propane in the Scott standard was confirmed by comparison to a methane-propane mixture provided by the National Bureau of Standards (SRM 1660A) which had a stated accuracy of 1% (95% confidence limit). The accuracy of the NMHC calibration procedure after dilution down to the low ppbv/high pptv range is estimated to be 5%. The halocarbon calibration gases were prepared on the same analytical systems as reported by Tyler [1983], Gilpin [1991], and Wang [1993] with an absolute accuracy for the various gases in the 2-10% range. Specific peaks detected by the FIDs were identified by comparison to the various Scott calibration standards, which contain four to nine gases, and to qualitative standards prepared at our home laboratory, each containing one gas diluted in helium. Because of the sensitivity in elution times of the plot column to variations in trace concentrations of H20 and CO2, the column was subjected during the peak   Each of the halogenated qualitative standards was in the low ppbv to high pptv range depending on the gas. Although retention times on the DB-624 column were affected less by varying amounts of H20 and CO2, qualitative standards were added to the whole air standard in the same manner as were the NMHCs. Only one separation column was used in the halocarbon analysis, but peak symmetry strongly suggests that coelution did not affect the mixing ratios of the seven reported halocarbons.
The specific NMHCs presented in this paper are the   The precision of the air sample measurements, based on repeated analysis of the NOAA working standard, was estimated to be better than 2% for the C2-C5 NMHCs and 10 and 20%, respectively, for benzene and toluene. The halocarbon precision in the NOAA standard was better than -+ 1% for all reported gases. The standard deviation for CC12F2 in the 37 canister measurements of mission 18 was _+2.9 pptv on a 484 pptv average, or _+0.6% including any atmospheric variability among the samples. The precision in CCI3F measurements in the same 37 air samples was _+ 1.7 pptv, or _+0.7%. The precision in the measurement of N20 was better than _+2%. The three C2 hydrocarbons were present in all samples collected during this project. Propane and propene suffered from CO2 peak broadening on the plot column so their detection limits are estimated to be 10 pptv. Because of split ratios and column efficiencies we estimate that the C4-C 7 hydrocarbons have a lower limit of detection of 2 pptv. The measurements with the plot column (Figures NMHCs, nonmethane hydrocarbons. aThe measured yields for cis-2-butene, trans-2-butene, 2-methylpentane, and 3-methylpentane were below 2 pptv in all air samples. All of these have average yields < 0.01 versus C2H6. longitude.

Fuel and Burn Description Sites
The type of fuel or vegetation consumed by a fire can play a significant role in determining the partitioning among species of NMHCs emitted. The predominant vegetation type found in the latitudinal belt from 45øN to 70øN is the boreal forest, or taiga. The composition of the boreal forest is mainly pine (Pinus), spruce (Picea), latch (Larix), and fir (Abies), mixed, usually after disturbance, with deciduous  fen, to bog and swamp, and finally peat. Swamps decrease northward, fens increase northward (and coastward), and tundra and permafrost dominate with increasing latitude [Riley, 1982]. To the south, the boreal forest zone is succeeded by temperate forests or grasslands. Since 1920, detailed forest fire statistics have been archived in Canada. Currently, about 10,000 Canadian fires are reported annually, representing a significant increase over previous decades. This is a reflection not only of a growing population and increased forest use but also of an expanded fire detection capability. Lightning accounts for only 35% of Canada's fires, yet these fires result in 85% of the total area    1.85 ___ 0.09 1.63 ___ 0.03 1.62 -0.04 1.64 Figure  t0). The back trajectories suggest that the air mass passed directly over fires prior to arriving at the sampling location. Thus the plume encountered between 1300 and 2200 m probably went aloft around sunset the previous day. Therefore as the main removal process for most NMHCs is reaction with hydroxyl radicals only produced during the daytime, it is likely that such removal occurred for just a few hours prior to sample collection. Thus only the shortestlived gases had undergone significant removal. The absence of c/s-2-butene and trans-2-butene, both highly reactive with hydroxyl, support this conclusion.
The mixing ratios given in Tables ta and lb Table 2. With its atmospheric lifetime of a few hours or less, the isoprene concentrations in the mission 4 spiral were certainly the consequence of very recent injections, presumably mixed upward prior to the arrival of the biomass-burning plume. While isoprene, or 2-methyl-l,3-butadiene, is a plausible compound to have been emitted from a burning forest, as its close chemical relative 1,3-butadiene definitely is, the isoprene levels within the plume samples show no evidence of any special enhancement. In fact, the major plume canister at 1970 m had a lower isoprene concentration than most of the other plume canisters. Table 2, for the six plume canisters, using the assumption that the background concentrations in the absence of bio-mass burning would have been the average value for the 10 canisters surrounding them, five above and five below. These excess NMHC values have then been converted, also in Table 2, into relative excesses versus AC2H6. (Emission factors are normally reported relative to CO or CO2. However, CO2 was not an on-board measurement and the carbon monoxide instrument was in a calibration mode during some of the plume events. Because of this we have chosen to report our NMHC emission factors relative to ethane.) The errors of these relative excesses have been estimated by the arbitrary assumption that the standard deviation for the individual canister in the plume is the same as for the background canisters, e.g., at 2060 m [C2H 6] = (1043 -+ 6) -  The relative amounts of excess hydrocarbon in these six canisters are very consistent as shown by the near constancy of all of the ratios in Table 2. This consistency holds quite well, not only for the components such as C2H6, C2H 4 and C2H2 but also for several of the less abundant hydrocarbons, even for those with single-digit pptv excesses. The calculated errors for toluene are clearly too large and are reduced to the _+0.02 level if a background of 15 -+ 3 is used, ignoring the sample at 1110 m.

Excess NMHC mixing ratios have been calculated in
In contrast to the observations for NMHC the plume canisters from the mission 4 spiral show no statistically significant excesses of N20 or of the man-made chlorocarbons, as summarized in Table 3. The average mixing ratios of these have been calculated for nine canisters each above and below the plume canisters, and the average of all 18 has been subtracted as the background for the plume samples. None of these calculated "excesses" approaches even the 1 rr level of significance. Furthermore, 33% of the total C2H 6 excess occurs in one canister (1970 m) and 73% in three (1970, 1890, 1730 m) of the six canisters, and no differences exist in concentrations observed for these three versus the other three plume canisters. No evidence, therefore, exists for the release of N20 or the reported halocarbons during forest fires in northern Ontario. This conclusion differs from that of Hegg et al. [1990], whose data were largely drawn from fires in the vicinity of Los Angeles. As a minimum, the lack of any enhanced halocarbon emission indicates that extrapolation from the Los Angeles vicinity as globally typical is inappropriate. Further data from other locations are highly desirable, including a recheck of Los Angeles fires.
Missions which took place later in the fire season revealed elevated background levels of CO, NOx, NOy, and several of the longer-lived NMHCs. The enhanced background levels of these gases precluded the same treatment as that used for mission 4 samples because the emission ratio determinations are all based on differences with respect to background. This problem is minimized when the measured in-plume mixing ratios are large. Biomass plumes were encountered during many of the missions but often were mixed with urban air, as inferred from elevated concentrations of C2C14, thus distorting the NMHC ratios. For this study, only plumes which were visually verified to be of biomass-burning origin or plumes for which the trajectory analyses are unequivocal have been used for emission factor calculations.
In both cases the assignment to biomass burning has been confirmed by the lack of change in halocarbon concentrations, especially C2C14, from background levels. In addition to the mission 4 spiral, samples obtained from missions 9 and 12 meet these three stringent criteria for biomass plumes. The low-altitude samples of mission 9 were taken in a plume rising directly from a visible active fire and were, thus, only a few minutes old. The observed yield of 1,3-butadiene in the most impacted canister of mission 9 had a concentration ratio of 0.13 versus C2H 6. The emission yields from this canister are, because of the proximity in both time and space of the sampling to the fire, probably the best estimate we have of the initial yields in an active Canadian forest fire. In none of these other occurrences were as many separate samples     Tables 4a and   4b for four canisters on a short ascent on mission 9 and for three successive canisters in level flight on mission 12. Two of the canisters contained plume samples in each set, while the others serve as background. These data have been calculated in the manner used in Table 2. The emission ratios relative to C2H 6 are summarized for all three missions in Table 5.
On a number of missions, plumes were encountered that through trajectory analysis were later attributed to biomass burning. However, these plumes were "aged", having last come in contact with a fire at least five days before the mission was flown. These older plumes have a very different NMHC signature, as seen in Figures 19 and 20. The more reactive gases show little or no enhancement, while the longer-lived gases are observed in appreciable amounts. Clearly, emission factors for the most reactive gases such as 1,3-butadiene are usually no more than lower limit unless the plume is sampled in a time short compared to the individual atmospheric lifetime. However, emission factors for the longer-lived gases, ethyne and propane, are generally in good agreement with those measured in fresh plume samples. Emission factors for samples collected in aged plumes, mission 9 near Moosonee and mission 11 over Schefferville, are given in Table 5 for comparison.
Various types of fires from different regions have been studied by numerous groups [Bonsang et al., 1991;Radke et al., 1991;Blake et al., 1992]. Emission factors from some of these studies are given in Table 6. Although the emission factors from these other studies exhibit considerable scatter, a significant amount of which probably originates from differently aged plumes, the values agree fairly well.
The impact of biomass-burning plumes on local photochemistry is illustrated in Table 7. The relative removal rate by HO radical reaction has been calculated for each of the NMHC compounds measured by us for several typical biomass-burning samples. When isoprene is present, its reactions tend to dominate the NMHC hydroxyl budget, allhough in a few instances the incremental additions from biomass burning play a comparable role in total reactivity. The reactions of HO radicals with CH 4 and CO are major contributors in all cases. However, NMHCs released by biomass burning can play a significant role in regional photochemistry long after isoprene and the reactive alkenes, etc., have been removed.

World Average Calculation
Biomass-burning emission factors for CO and NMHCs are dependent on many factors, particularly the fuel and whether the fire is flaming or smoldering. Because the efficiency of a flaming fire is greater than a smoldering one, the CO2/CO emission ratio is greater in the former case [Einfeld et al., 1991]. In this study we assume that the encountered plumes were a result of both flaming and smoldering fires and that the emission factors obtained in this study represent the average. Savanna fires would burn hot and fast and are therefore very efficient and produce relatively high amounts of CO2 [Einfeld et al., 1991]. Thus although savanna fires represent a significant amount of total biomass burned, it is likely that biomass burning of forested areas produce the majority of CO emitted during fires. If it is assumed that NMHC and CO emissions from biomass burning in different regions of the world are similar and that 350 x 1012 g of CO are released globally each year during biomass burning [Andreae, 1991], then the contribution of biomass-burning emissions to the tropospheric budgets of some of the longer-lived gases can be estimated. The ACO/AC2H 6 biomass-burning ratio obtained in this project is 260, while the average of the four geographically different studies given in Table 6 is 240 -+ 120 [Stocks, 1991; Blake and Rowland [1986], from which a percent contribution to annual C2H 6 emissions from biomass burning of 9-15% is calculated. Using emission ratios obtained in this study, the global input of C2H 2, C3H 8, and n-C4H10 were calculated in the same manner as was C2H 6 and are given in Table 8. Using latitudinal profiles of our unpublished data and those from Singh and Zimmerman [1992], global input values for ethyne, propane, and n-butane of 7-10%, 3-4%, and 1-2%, respectively, are obtained.

Conclusions
In conclusion, the following observations were made during the ABLE 3B campaign which took place during July and August 1990 over the northern Canadian wetlands. Based on many vertical characterizations of the background conditions encountered during this experiment, the general trend was for the concentrations of halocarbons and NM-HCs to be nearly constant or slightly increasing with increasing altitude (up to 5400 m). The vertical profile of isoprene implies that mixing within the boundary layer was rapid to altitudes as high as 3000 m. Stratified layers denoting longrange transport were evident. Based on the levels of C2C14 and related anthropogenic halocarbons, the origin of these stratified layers or plumes may be further characterized as having either biogenic and/or urban/industrial sources.
Emission factors for NMHCs were calculated and ethene was found to be the most abundant species released. During biomass burning, alkanes, alkenes, and alkynes are all emitted in significant quantities. The correlation coefficients for the longer-lived gases are in excellent agreement and for the shorter-lived species are very good. Compounds such as CFC 12, N20, and isoprene, whose emissions from biomass burning have been reported in previous studies, were not found to be enhanced significantly in the fire plumes encountered. The estimated percent contribution of biomassburning emissions to the worldwide total were calculated for ethane, ethyne, propane, and n-butane to be 9-15%, 7-10%, 3-4%, and 1-2%, respectively. Finally, the less reactive NMHCs associated with biomass burning tend to accumulate as the fire season progresses. Although these emissions can effect regional photochemistry, within the PBL, isoprene is still the dominant hydroxyl sink.