SUMMERTIME DISTRIBUTION OF PAN AND OTHER REACTIVE NITROGEN SPECIES IN THE NORTHERN HIGH-LATITUDE ATMOSPHERE OF EASTERN CANADA

Aircraft measurements of key reactive nitrogen species (NO, NO2, HNO3, PAN, PPN, NO3-, NOy), C1 to C6 hydrocarbons, acetone, O3, chemical tracers (C2Cl4 CO), and important meteorological parameters were performed over eastern Canada during July to August 1990 at altitudes between 0 and 6 km as part of an Arctic Boundary Layer Expedition (ABLE3B). In the free troposphere, PAN was found to be the single most abundant reactive nitrogen species constituting a major fraction of NOy and was significantly more abundant than NOx and HNO3. PAN and O3 were well correlated both in their fine and gross structures. Compared to data previously collected in the Arctic/subarctic atmosphere over Alaska (ABLE3A), the lower troposphere (0–4 km) over eastern Canada was found to contain larger reactive nitrogen and anthropogenic tracer concentrations. At higher altitudes (4–6 km) the atmospheric composition was in many ways similar to what was seen over Alaska and supports the view that a large-scale reservoir of PAN (and NOy) is present in the upper troposphere over the entire Arctic/subarctic region. The reactive nitrogen budget based on missions conducted from the North Bay site (missions 2–10) showed a small shortfall, whereas the budget for data collected from the Goose Bay operation (missions 11–19) showed essential balance. It is calculated that 15–20 ppt of the observed NOx may find its source from the available PAN reservoir. Meteorological considerations as well as relationships between reactive nitrogen and tracer species suggest that the atmosphere over eastern Canada during summer is greatly influenced by forest fires and transported industrial pollution.


INTRODUCTION
Northern high latitudes are of great interest to atmospheric scientists because of the high predicted sensitivity of this region to climatic change and the paucity of available data. Despite the remoteness of the Arctic/subarctic atmosphere, significant human influences are beginning to impact this region, especially in winter and early spring [Schnell, 1984;Rahn, 1985;Barrie, 1986;Jaffe et al., 1991]. Only a few studies of the high-latitude environment have been conducted and these generally have focused on the winter/spring Arctic haze period. The NASA Global Tropospheric Experiment (GTE) seeks to understand the atmospheric chemistry associated with the earth's major ecosystems [McNeal et al., 1983]. Recent GTE field measurements are now providing a large data base to investigate atmospheric photochemical and transport processes impacting the northern high latitudes. For example, aircraft flights during the Arctic Boundary Layer controlled and organic nitrogen analysis was performed every 6 min. PAN calibrations were performed by using a diffusion tube filled with liquid PAN in a n-tridecane matrix held at icewater temperature. Air was passed over the diffusion tube at a constant rate of flow to provide a calibration gas stream with PAN mixing ratios in the 10 to 20-ppb range. PAN in the exit stream was measured by using an NOy detector equipped with a molybdenum oxide converter held at 375øC. A nylon filter was inserted in line to remove any traces of nitric acid. Calibration of this chemiluminescence system was accomplished by using both NO2 and NO standards that were intercompared with National Institute of Standards and Testing (NIST) primary standards. An on board dynamic dilution system was used to generate low parts per trillion mixing ratios of PAN. Instrument precision and accuracy are estimated to be +10% and +25%, respectively, at the 95% confidence level for PAN mixing ratios above 50 ppt. Details of this and similar techniques can be found elsewhere [Singh and Salas, 1983;Joos et al., 1986;Ridley et al., 1990;Singh et al., 1990Singh et al., , 1992a. NO, NO2, and NOy were measured by a laser-inducedfluorescence (LIF) instrument developed by Sandholm et al. [1990Sandholm et al. [ , 1992. In this instrument the 226-nm and 1.1/.tm laser beams are colinearly combined and passed through three separate sample cells dedicated to ambient NO, NO2, and NOy detection. Measurement precision is estimated to be + 16% NO, + 20% NO2, and + 5% NOy for 3 min integration.
Absolute accuracy is estimated to be + 15 % NO, + 18 % NO2, and + 15 % NOy (at the 95 % confidence limit). Nitric acid (HNO3) was measured by the mist chamber technique described by Talbot et al. [1990]. A recent intercalibration of the mist chamber HNO 3 instrument with the National Oceanic and Atmospheric Administration (NOAA) nylon filter method and the National Center of Atmospheric Research (NCAR) Lind instrument showed agreement to within +25% in the 100-to 1000-ppt range [Atlas et al., 1992]. Sampling times varied from 3 to 45 min for HNO3 with a limit of detection of 10 ppt. The overall uncertainty was estimated to be +20% for concentrations >100 ppt increasing to +30% for concentrations near the limit of detection (10 ppt). Instrumentation principles, sampling methods, sampling times, data precision and accuracy have been summarized in the overview paper by Harris et al [this issue].

RESULTS AND DISCUSSION
It is well known that nitrogen oxides play a central role in the chemistry of the atmosphere [Levy, 1971;Crutzen, 1979;Logan et al., 1981]. Reactive nitrogen chemistry is complex and is strongly influenced by the presence of organic reservoir species [Singh and Hanst, 1981;Calvert and Madronich, 1987;Singh, 1987;Kanakidou et al., 1991]. As stated earlier, a comprehensive set of reactive nitrogen measurements (NO, NO2, HNO3, PAN, PPN, NO3,-and NOy) were made during the ABLE3B deployment. Different sampling periods were employed in these measurements. Several merged files containing matching data averaged within the PAN, HNO3, and NOy sampling windows were created for the purposes of data analysis. Where fast response data were available (e.g., CO and 03) 1-min data averages were employed. For NO, NO2, and NOy both 1.5-and 3-min sampling sets were used. In the analysis and interprel•ation that follows, all of these data files were used as appropriate. Nitric acid sampling window was the longest and data merged to this window were used in all cases involving HNO 3. Missions 2-10 are used to describe North Bay (NB) measurements, missions 11-19 are used to describe Goose Bay (GB) measurements, and missions 20-22 were transit flights terminating at Wallops Island, Virginia. Since both ABLE3A and ABLE3B sampled Arctic/subarctic air in summer, largely over Alaska and eastern Canada, comparisons are made to reveal important differences in the large-scale atmospheric chemical structures.
The atmospheric composition of air over central/eastern Canada during ABLE3B was impacted by a variety of air masses containing forest fire smoke, industrial pollution, and stratospheric air. According to the meteorological classification for the first 18 ABLE3B missions, provided by Shipham et al. [this issue], forest fires impacted missions 4-9, 11-13, and 15-18, pollution possibly affected missions 4, 5, 10, 13, 16 and 17, and stratospheric influences could be seen in upper layers during missions 1-3, 5, 7-9, 12, 14, 16-18. Forest fires were both local (missions 7-9, 11) as well as remote (missions 4-7, 11-13, 15-17). Mission 6 provided a dramatic example of the influence of haze and smoke plumes transported over long distances and has been discussed in some detail by Shipham et al. [this issue]. High C2C14 mixing ratios further indicated that mission 11 was impacted by pollution as well but the flow of air was from the Northwest Atlantic. It is worth pointing out that C2C14 and CC13F (CFC 11), conventional tracers of urban pollution, were often inconclusive indicators although some enhancement in C2C14 could be seen in missions 10, 11, 13, 16, and 20-22. The reason for this was that small source signatures superimposed on a large background could not be easily separated. In rare instances (upper levels in missions 14, 15, 17), Pacific air of mixed tropical/mid-latitude origin was also sampled. In short the atmosphere over central/eastern Canada was strongly influenced by a variety of complex sources. A further discussion of selected events has been provided by Browell    The meteorological conditions that prevailed during ABLE3B were examined to assess their impact on measurements from different air masses and wind (air trajectory) patterns [Shipham et al., this issue]. Using height contour analyses of the lower troposphere 700 mbar (3 km msl) constant pressure surface, several characteristic weather patterns were recognized. Table 1 lists the principal features of these patterns and the mission numbers to which they apply. Typically, the prevailing synoptic patterns pointed to the flow of Arctic/subarctic air from north to northwest and the flow of potentially anthropogenically perturbed air from the south to southwest. Generally, clean conditions were encountered during northerly flow of Arctic air and NW flow of polar air except for intermittent low-level smoke and haze layers (e.g., mission 6). Figure 3 shows the vertical distribution of PAN, HNO3, and 0 3 for NB missions 5 and 9. The vertical profiles are obtained by using the available measurements of PAN, HNO 3 (15-min sampling time), and 0 3 (3-min sampling time) for one ascent and one descent made during mission 5 and for two ascents and two descents made during mission 9. The profile data were obtained between

Moosonee and Churchill for mission 5 and between North
Bay and Moosonee for mission 9. Lines connecting the PAN and HNO 3 data points are drawn for visual aid. Mission 5 is characteristic of clean NW flow, whereas mission 9 represents S-SW flow. Mission 5 was downwind of distant fires in Alaska and may have been minimally impacted. However, local fires did influence mission 9 at low altitudes and biomass burning was visible during the mission. Tropospheric (2-6 km) C2C14 mixing ratios (not shown) were also slightly larger during mission 9 (11_+3 ppt) compared to mission 5 (9_+1 ppt). (If the data are assumed to have a normal Gaussian distribution, the probability of finding C2C14 mixing ratios of > 11 ppt would be 50% in the case of mission 9 but only 2% in the case of mission 5.) In Figure  3b and 3c the combined affect of forest fires and pollution is seen in the enhanced mixing ratios of HNO 3 and 0 3 below 1km altitude observed during mission 9. Figure Table 2 lists mean mixing ratios (-+1•) of species measured during ABLE3A and ABLE3B for the mean boundary layer (0-2 km), the transition layer (2-4 km), and the free tropospheric layer (4-6 km). Median mixing ratios are shown in Figure 7 for easy comparison. The measurements represent those made at the primary sites of Barrow and Bethel, Alaska, and North Bay and Goose Bay, Canada. Some of the salient differences are noted as follows: 1. In the boundary layer, mean (or median) mixing ratios of PAN for the NB and GB missions were larger than those for the Alaska missions. At the free tropospheric altitudes of 4-6 km, PAN mixing ratios were comparable at all sites and

Mean (or median) NOy mixing ratios were highest at NB
and lowest at GB. This behavior was also reflected in the O3/NOy ratios (Table 3). Standard deviations in the 2-to 4km and 4-to 6-km layers were very large during the GB missions.
3. NOx values in the various tropospheric layers are comparable, except for the NB missions when mixing ratios were relatively large in the lower troposphere (0-4 km) with considerable variability. Unlike NOy however, NOx mixing ratios at GB were not the lowest, and at high altitudes (4-6 km) were comparable to those at NB. In this respect NOx behavior was similar to that of PAN, CO and C2H2 .

Despite substantial variabilities, HNO 3, CO, and C2H2
show significantly higher mixing ratios in the boundary and transition layers for the NB missions compared to other sites. Ozone values, on the other hand, are quite comparable for all altitude layers at all sites.
Overall, the atmosphere over eastern Canada during ABLE3B was perturbed by forest fires and surface pollution plumes significantly more than over Alaska (ABLE3A). These

Relationships Between Reactive Nitrogen and Tracer Species
The relationships between PAN, NOy, CO, 03, select hydrocarbons, and C2C14 were explored to see if these can shed any light on the nature of reactive nitrogen sources. Figure 9 shows  Figure 9d were measured during GB mission 11 when polluted air was sampled in turbulent cloudy weather (see also Figure 5). Similarly, the high concentration data points in Figure 9c were associated with missions affected by smoke and haze layers (e.g., NB mission 6). Although C2C14 data showed the existence of sporadic pollution, it was not possible to find distinct correlations among PAN, NOy, and C2C14 largely because of the extremely small variability in C2C14 mixing ratios. Both CO and C2H 2 are tracers of biomass burning and urban air pollution. As stated earlier, no unique tracers of biomass burning (possibly CH3C1) were measured. Figure 9 is largely consistent with the view, also supported by meteorological analysis, that both forest fires and pollution impacted this region. PAN was also correlated with 03 both in the boundary layer and aloft ( Figure 10) in a manner similar to that observed over Alaska [Singh et al., 1992a]. The correlation at GB was somewhat stronger than that at NB. In NB, because of the influence of smoke plumes, a greater scatter in the data is seen. However, when such cases are excluded the PAN-O3 correlation at NB is comparable to that at GB. These PAN-O 3 relationships are indicative of the influences of continental precursors [Singh et al., 1990[Singh et al., , 1992a. The NOy-O3 correlations are also shown. It is noteworthy that the mean NOy mixing ratios at GB are much smaller than those at NB for a comparable amount of 03. Mean O3/NOy ratios at NB (107+54) were about 40% smaller than at GB (179+86). This behavior was observed at all altitudes but was more significant in the free troposphere (Table 3). It is noted that O3/NOy ratios observed here and over Alaska [Singh et al., 1992b] are significantly smaller than those observed in the lower stratosphere [Murphy et al., 1993]. This provides little support for the suggestion made by Jacob et al.
[1992] that stratospheric transport may provide a large source of Arctic/subarctic reactive nitrogen.

Budget of Reactive Nitrogen Species
Issues relating to reactive nitrogen budget, shortfall and the associated variability have been studied in detail by Sandohm  (Table 3), NOy i is nearly always less than NOy in NB and accounts for 55to 77% of measured NOy. Some 100-400 ppt of reactive nitrogen is unaccounted for.

In GB the picture is quite different and NOy i accounts for nearly all of NOy and actually exceeds it by a small amount in some instances (Table 3). Given the error bars in individual measurements and the inexact overlapping of sampling windows, it is not surprising for NOy i to exceed NOy some of the time. What is surprising is that this result (NOy i --NOy)
is contrary to what was seen at NB where similar air masses were sampled, and in Alaska. Indeed, reactive nitrogen

Role of PAN as a Source of NO x and 0 3
Data presented here show that significant mixing ratios of PAN and correspondingly small mixing ratios of NOx coexist in the troposphere over eastern Canada. Whereas combustion sources of NOx were clearly present, a certain fraction could also be produced from the PAN reservoir. During the ABLE3A Alaska effort it was calculated that 50 to 70% of the median N Ox mixing ratios present could have come from the PAN reservoir alone [Jacob et al., 1992;Singh et al., 1992b]. What fraction of the available NOx could have come from PAN decomposition source alone in NB and GB during ABLE3B?
To explore this possibility, a one-dimensional model of NM H C-O 3-NOx photochemistry in the troposphere was adapted to the physical and chemical environment encountered during ABLE3B (August 1, 50øN) and a calculation similar to that in the work of Singh et al. [1992b] was performed. In comparison to the previous effort, larger K values (by a factor of 2) were used in the present computations, as these values provided a better fit to measured PAN and 0 3, especially in the lower troposphere. Mean atmospheric temperature as well as mixing ratios of key species were chosen so as to reproduce the mean observed values during ABLE3B (e.g., 400 ppt for PAN and 60 ppb for 0 3 at 6 km). Model simulations, under clear sky conditions, were run when all sources of NOx (e.g., lightning, soil) were set equal to zero and zero flux was Was there sufficient NO x present in the atmosphere over eastern Canada to cause net 0 3 synthesis? The (NOx)eq value (dashed curve in Figure 12) is calculated based on chemistry previously described by Singh et al., [1992b] and a faster diffusion as stated above. The latter change had the effect of increasing calculated NO x and (NOx)eq in the lower troposphere. Figure 12 shows that 25 to 45 ppt of NO x is CONCLUSIONS The two field studies performed over Alaska and eastern Canada have shown that large abundances of PAN and NOy are present in the Arctic/subarctic troposphere of the northern hemisphere during the summer. Coincident PAN and 0 3 atmospheric structures suggest that continental precursors define the behavior of both PAN and 0 3 . In the free troposphere, PAN was found to be the single most abundant  +GIT aircraft data. NO 3-is very low (= 5% of NOy) and is expected to be largely in the coarse aerosol fraction which is probably not sampled. _ ?ND, no data; U, uncorrected for NO 3-' C, corrected data assuming that all NO 3 is sampled by the tower instrument as NDy.
reactive nitrogen species constituting a major fraction of NOy and was significantly more abundant than NOx and HNO3. The nitrogen budget based on missions conducted from the North Bay site showed a shortfall, as expected, but a similar budget from the Goose Bay operations showed essential balance, contrary to what has been previously reported from remote locations. At remote locations, reservoir species such as PAN have the ability to provide an important source of NOx. Meteorological considerations as well as relationships between reactive nitrogen and anthropogenic tracer species suggest that the atmosphere over eastern Canada during ABLE3B wa• strongly influenced by forest fires and transported industrial pollution.