Factors influencing atmospheric composition over subarctic North America during summer

Elevated concentrations of hydrocarbons, CO, and nitrogen oxides were observed in extensive haze layers over northeastern Canada in the summer of 1990, during ABLE 3B. Halocarbon concentrations remained near background in most layers, indicating a source from biomass wildfires. Elevated concentrations of C 2 C14 provided a sensitive indicator for pollution from urban/industrial sources. Detailed analysis of regional budgets for CO and hydrocarbons indicates that biomass fires accounted for = 70% of the input to the subarctic for most hydrocarbons and for acetone and more than 50% for CO. Regional sources for many species (including CO) exceeded chemical sinks during summer, and the boreal region provided a net source to midlatitudes. Interan-nual variations and long-term trends in atmospheric composition are sensitive to climatic change; a shift to warmer, drier conditions could increase the areas burned and thus the sources of many trace gases.


INTRODUCTION OBSERVATIONS
Industrial and urban emissions provide massive inputs of pollutants to boreal and subarctic latitudes in winter and spring; rates for deposition and degradation are slow, and high pollutant concentrations are observed [Rahn, 1981;Bartie et al., 1985;Battle, 1986;Stonehouse, 1986;Bottenheim et al., 1986;Li and Winchester, 1990]. Pollutant concentrations are much lower during summer [Hatriss et al., 1992]. Emissions from midlatitude sources appear to be efficiently scavenged during summer, and the influence of long-range transport is relatively weak. The composition of the atmosphere is most strongly affected by regional emissions, particularly boreal fires.
This Tetrachloroethylene represents a sensitive indicator for anthropogenic emissions. Its lifetime (-90 days in summer) is long use data for selected halocarbons and hydrocarbons to characterize enough to allow transport on a hemispheric scale but short enough emissions from urban/industrial sources and fi,• ..... ildfires, to maintain a low background concentration. When air with providing information on emission ratios for key species on the elevated concentrations of CO and hydrocarbons was sampled, the regional scale. We quantitatively assess the relative contributions anomaly could be attributed to long-range transport of pollution if of these sources to regional budgets, using a photochemical concentrations of C2C14 were also elevated, and vice versa. model. Wildfires appear to provide a major source for these gases Concentrations of C2C14 measured by the on-board during summer, as suggested earlier by data from Alaska [Blake et chromatograph and in grab samples were in harmony at high al., 1992; Harriss et al., 1992;Wofsy et al., 1992]. The subarctic concentrations (> 20 parts per trillion by volume (pptv)), but is a net global source for many trace species, reversing the disagreed at low concentrations, reflecting different sampling source-sink relationships observed in other seasons. times and possibly measurement and/or sampling artifacts. Correlations between C•CI•, other halocarbons, and hydrocarbons were preserved in the grab samples at low concentrations, as shown in Figure 1, and we therefore focus on these measurements 1Division of Applied Sciences and Department of Earth and Planetary in our analysis. Three areas were sampled extensively: Hudson Bay lowlands (HBL), Ontario (flights 4 -9), central Labrador and Quebec LBQ), and eastern Labrador (ELO) with adjacent coastal areas (flights 11 -20). These areas were selected to allow study of atmospheric composition for a range of biomes, dry and fire prone in HBL, somewhat wetter and cooler in LBQ and ELO, both remote from large urban complexes. Identifiable smoke plumes from large fires were sampled over HBL on flights 4, 6, 7, and 8. Industrial/urban pollution was sampled twice over Canada (flights 10 and 16) and over the East Coast of the United States (flights 21 and 22), providing a signature for urban/industrial emissions.

Characteristic Emission Ratios for Tundroll'alga Fires Versus GH2
Hydrocarbon emissions from biomass fires and ratios to CO and C2H2 are summarized in Table 3a, determined from haze layers with unperturbed halocarbon concentrations (flights 4, 6-9, 11; see Figure 5 for examples). Haze layers sampled on flights 18 and 19 (Table 3a, bottom panel) showed enhancements for long-lived species similar to those in flights 4, 6-9, and 11, but no enhancement of shorter-lived hydrocarbons. These haze layers had apparently aged long enough for short-lived species to be removed by photochemistry and for significant quantities of (CH3)2 CO to be generated.
The ratio C2H2:CO was remarkably consistent among flights (e.g., Figure 5) of most other hydrocarbons are more uniform, relative to C2H2, in was reflected in enhanced concentrations relative to background levels, all seasons. Therefore in the present paper we use emission ratios throughout the column 0-6 km. Largest enhancements were observed at from Table 1, relative to C2C14, neglecting anomalous points for the lowest altitudes, indicating rapid low-level advection of pollutants from C2H 6 and C3H8 . Thus we neglect these fuel-related sources and source regions to the south. emission ratios for most hydrocarbons are remarkably similar, the prevalence of smoldering combustion; the associated low relative to CO or C2H2, from biomass fires in a wide range of NOx/hydrocarbon ratio is responsible for the negligible rates of environments, including tundra, taiga, midlatitudes, and the ozone photochemical production observed in boreal haze layers tropics (see Table 3b). It appears that smoldering combustion [Jacob et al., 1992]. provides the dominant source for emissions of these reduced In the discussion below, we use concentrations of C2H2 and the gases, even though the fraction of fuel consumed in smoldering ratios from  Table 3b, representing aged haze layers to account for methane are much higher in boreal fires than in fires at secondary production from oxidation of labile olefins (e.g., midlatitudes [Laursen et al., 1992] or in the tropics [Andreae et propene). The observed variance ratio forC2H2:C2C14 from Table  al., 1988], reflecting release of biogenic methane from peat soils 1 was employed to account for input of C2H2 from by pollution in boreal wildrites (Table 3b). Emissions of NOx are much lower sources (for flights with enhancements of C•C14). We selected than in most other biomass combustion or in pollution, reflecting C2H2 as the index species, instead of CO, to preserve correlations on gases with lifetimes of 1-12 weeks, long enough to establish a generally uniform background in the subarctic but short enough so that regional background concentrations are regulated significantly by chemical losses in the region. Careful statistical analysis of observed concentrations is required in order to define regional background concentrations.

Background concentrations for species measured in grab samples were derived by examining the subset of grab samples with corresponding CO concentrations within +20% of median CO, exploiting the more representative sampling for CO. We adopted the trimmed mean (average excluding the highest and lowest 10%) of this subset to define background concentrations; other conditional selection procedures yielded indistinguishable
The background concentration is the characteristic value in the results (Table 4). As expected, background concentrations were region, upon which anomalies reflecting local inputs or losses, or notably lower than means of all grab samples and slightly lower unusual transport events may be superimposed. The inputs, (5-10%) than corresponding medians. outputs, and chemical sources and sinks for a gas in a defined Estimates for changes in background concentrations during the geographical region determine the regional budget. Both concepts mission are needed for the budget analysis. Only a rough are defined unambiguously for species with small spatial variance; determination is possible due to the limited geographic area neither is well defined for gases with short lifetimes and covered and the bias in selecting grab samples.

. Even the 10th percentile of data from the smaller set of grab samples appears to be affected by biomass fires on Flights 6-9. We adopt the difference between flights 4 and 5 and 17-19 as an estimate of the change in background concentrations over the period; the corresponding rates of change ((A[i]o/At) for species i)
are shown in Table 5. relatively little effect on NOx or HNO3 (Table 3).  Table 5. Means assumed =0 for short-lived, high-variance species.

Figures 8 and 9 show meridional gradients for a variety of gases above the boundary layer (> 4 km) during transit flights on ABLE 3B and ABLE 3A, respectively. Concentrations of most species were higher in the subarctic than at midlatitudes during both missions for halocarbons and hydrocarbons but not for nitrogen oxides. This surprising result is consistent with the budget analysis given below. The latitudinal distribution for halocarbons indicates the influence of both European sources at high latitudes
REGIONAL BUDGETSOFTRACESPECIESINTHESUBARCTIC where the terms on the right hand side represent regional photochemical loss, net exchange with midlatitudes, and The regional mass balance for C2C14, at latitudes > 56øN, may concentration change during the observation period, respectively.    Table 5 shows our analysis of the budget for C2C14 for latitudes > 56øN. A mean input of 0.2 ppt/d is required to balance losses due to reaction with OH, transport to midlatitudes, and time is the mean variance ratio for i relative to C2H 2 from Table 3;

The input of C2H 2 from biomass fires, Pbb © (= 3.8 ppt/d),
dependence. This is equivalent to 120 kt/yr or about one fourth of was obtained by assuming balance in equation (2) for i = C2H 2 European and North American emissions (Table 2a), a reasonable (see Table 5). The source due to biomass fires is about twice as fraction since most European sources are located between 48 ø and large as inputs from industrial/urban pollution. Since C2H 2 has a 60øN, while most North American sources are south of 48 ø. shorter chemical lifetime than C2C14 (5 times faster reaction with Emission rates for C2C14 at high latitudes evidently exceed rates OH), transport and time-dependence terms are considerably of consumption due to reaction with OH. The quantity of C2C14 smaller for C2H 2 than for C2C14, relative to regional chemical exported from the subarctic cannot be determined accurately from loss. our data due to the restricted spatial domain of the measurements, but the direction of net exchange is evident.
Table 5 summarizes regional inputs and losses for a number of species, given by

0[iPind C2Ch + [•iPbb ©+Pin situ = [i]oki[OH] + A[i] + A[i]o Xex At
(2) where oti is the mean variance ratio for pollutant i from Table 1; Table 5 using observed variances relative to C2C14 and C2 H2, combined with budgets for C2C14 and C 2H 2 constructed above using model OH distributions [Spivakovsky et al., 1990]. We can test how well the sources thus defined account for observed background concentrations by examining the ratio of terms on the right-hand side in equation (2) Table 5 to the emission inventory in Table 2. The industrial source for CO to the subarctic corresponds to annual input of 85 x 106t yr -• , 36 % of North American and European sources. This appears to be a reasonable value in light of the geographic distribution of sources.

•-,C2C14
Values for (gil"in d appear to correspond to similar fractions of emission inventories for hydrocarbons (Table 2b), except for (CH3)2CO (see below).
The results in Table 5  there is a small additional source associated with oxidation of nonmethane hydrocarbons (Table 6a), mainly isoprene.
Industrial/urban emissions play a smaller role, contributing 1/4 -1/3 of total input (except 50% for n-C4H•0). Rates for photochemical removal are larger than loss terms associated with time dependence or flux divergence, indicating that the composition of the atmosphere does not depend strongly on antecedent conditions or on exchange with lower latitudes. We were surprised initially to observe consistent excess concentrations of trace species in the "clean" subarctic, as compared to midlatitudes, above the boundary layer. According to our analysis, this distribution reflects inputs from biomass fires that exceed the regional rate of scavenging by OH. Note that areas burned in the boreal zone in 1988 were about 2 times larger than in 1990 (Table 6b); the latitude gradient for CO was correspondingly larger (Figure 9), providing strong support for this analysis. human activities. Hence CO concentrations appear to be strongly perturbed by global anthropogenic changes, even though direct pollution inputs are not large in the region during summer. Acetone is a major source of acetylperoxy radicals, and it therefore plays an important role in producing PAN from NOx. Direct industrial sources are small (Table 2), but observed variance ratios (Table 1)