Summertime Tropospheric Observations Related to N xOy Distributions and Partitioning Over Alaska: Arctic Boundary Layer Expedition 3A

Measurements of the reactive odd nitrogen compounds NO, NO 2, peroxyacetyl nitrate (PAN), and NOy are presented for the summertime middle/lower troposphere (6.1-0.15 km) over northern high latitudes. In addition, the chemical signatures revealed from concurrent measurements of 03, CO, C2H2, C2H6, C3H 8, C2C14, and H20 are used to further characterize factors affecting the budget and distribution of Nx Oy in the Arctic and sub-Arctic tropospheric air masses sampled over Alaska during the NASA Arctic Boundary Layer Expedition (ABLE 3A) field campaign. Many of the compounds listed above exhibited a general trend of median mixing ratios increasing in proportion with altitude within the lower 6-km column. However, median mixing ratios of NO and NOx (NO + NO2) were nearly independent of altitude, having values of about 8.5 and 25 pptv, respectively. Median mixing ratios of NOy varied from about 350 pptv within the lowest altitudes to about 600 pptv within the highest altitudes sampled. PAN constituted the largest fraction of NOy (--(cid:127)50%) at the highest altitudes. In addition, PAN mixing ratios accounted for all of the approximate 60 pptv/km altitudinal dependency in NOy. The analyses presented implicate biomass burning in Siberia as the probable source of about one-third of the NOy abundance within the middle/lower troposphere over Alaska. These analyses also tropopause have

NOy convertors. NO2 was used to monitor NOy conversion etficiency. Cleaning of the converter tubes, when necessary, was accomplished via the procedures described by Fahey et al. [1985]. NO2 conversion etficiency was maintained at 98 -4% etficiency throughout the ABLE 3A field program.
To date, no interference has been identified in the TP/LIF measurement of NO, and interferences are believed to be -< ---2 pptv based upon nighttime measurements of NO mixing ratios at or below the limit of detection. The NO2 photolytic convertor was operated with a wavelength >350 nm. This cutoff wavelength was selected to minimize potential interference from the photolysis of NOx-containing organic compounds. The thermal decomposition of HO2NO2 and N205 yielding NO2 may represent a potential interference in the measurement of NO2 under cold high-altitude conditions, where these compounds have been predicted to have mixing ratios on the order of that of NO2 [Logan et al., 1981]. Attempts to quantify the magnitude of this potential interference are not yet definitive [Gregory et al., 1990a;Sandholm et al., 1990;Ridley et al., 1989]. If HO2NO2 and/or N205 decomposition took place in these short residence time, thermostated photolyric convertor systems, then it has likely taken place heterogeneously through reactions on the inlet and cell walls, not through homogenous decomposition in the gas phase. In this case, the magnitude of the interference would be relatively insensitive to modest variations (twofold to threefold) in sample flow rate (an often used test for possible interferences). Final resolution of any potential problem arising from the decomposition of unstable NO2containing compounds requires further study. Until these issues can be more clearly discriminated, the NO2 measurements presented here may need to be considered as upper limits under some sampling conditions (e.g., cold high altitude, and plumes rich in labile or photolytically active NOx-containing compounds). It should be noted, however, that the NOx/NOy and NOx/PAN ratios reported here and in a companion paper by Singh et al. [this issue (a)] represent some of the smallest free tropospheric values measured to date with the smallest ratios observed at the higher (i.e., colder) altitudes.
The TP/LIF sensor continuously measured the small background signal generated by the 226-nm laser beam, by tuning the laser wavelength off of the NO X2II --• A2E transition, and by blocking of the IR laser beam. This dominant background component of the instrument was subtracted for all measurements. Ambient nighttime clean air measurements made with our instruments yielded NO mixing ratios below, or at, the instrumental limit of detection. These nighttime values were often smaller than those measured from sampling zero-grade air gas cylinders. Similarly, mixing ratios of NO and NO2 measured in "clean" ambient air have been less than those measured in air from cylinders. In the case of NOy, a combination of chemical scrubbers has often reduced the mixing ratios of NOy measured from zero-grade air cylinders by factors of twofold to fourfold. Based on these observations, the data set presented here has not been altered by the NO, NO2, and NOy concentrations measured in the gas streams produced from zero-grade air gas cylinders and associated flow line hardware (where in these ground tests -•20 m of additional flow line was added to the system). These results suggest possible systematic differences of approximately 4 __ 4 pptv for NO, 8 --8 pptv for NO2, and 60 --60 pptv for NOy could exist between the data reported here and other data sets in which zero-grade air NO, NO2, and NOy "blank" values were subtracted from ambient air measurements unless some other collaborating background measurement has been made (e.g., the routine measurement of nighttime NO mixing ratios not influenced by local sources). The chemiluminescence-based NO, NO2, and NOy measurements made at the ABLE 3A ground site were also not corrected for zero-grade air "blanks." These NOx and NOy measurements were in good agreement when extrapolated to those made on board the aircraft. However, small (0-4 pptv) differences in NO were incapable of being discerned even though an offset of about +2 pptv was reported for the ground-based data [Bakwin et al., this issue].
Limits-of-detections (LOD's) for signal to noise ratios of 2/1 averaged 3.0 --1.5 pptv for NO and 8 --3 pptv for NO 2. Both LOD's had signal integration times of 3 min for data taken between July 8, and August 10, 1988. A slight increase in background noise counts was experienced in these earlier flights. This problem was corrected and resulted in improving the LOD's for the remainder of the program (August 11-17, 1988) to 1.4 ñ 0.4 and 5 ñ 2 pptv for NO and NO2, respectively. The absolute accuracy of the calibration transfer to the NO and NO 2 measurements was estimated at the 95% confidence limit at ñ 16% and ñ 18%, respectively [cf. Gregory et al., 1990a, b]. The NOy measurements were consistently above the detection limit and typically exhibited a measurement precision (at the 95% confidence limit) of ñ8.5% and ñ 15% at 700 and 200 pptv, respectively. NO and NOy data were reported for 1-min integration times. NO 2 data were reported using either 3-or 6-min signal integration time periods.

OVERVIEW OF FLIGHTS AND CONDITIONS
There were 33 aircraft missions conducted during the ABLE 3A program. The first five missions constituted the The summer of 1988 was somewhat warmer and drier than the climatological mean for the Bethel region, especially early in the season, whereas conditions at Barrow were close to the norm. During the ABLE 3A program, the air mass back trajectories over the Barrow region correlated reasonably well those of with the 5-year climatology reported by Miller [1981]. Notable, however, was an apparent lack of trajectories originating from the south to southwest (specified as sectors 22 and 23 in Figure 1 of Miller' s paper). These sectors represented a sizable percentage (---30-40%) of the 5-year climatology of July average air mass trajectory origins for the Barrow region. Air masses encountered near Barrow that originated from these sectors were only sampled on portions of missions 6 and 7. Air masses sampled over the Bethel region, however, had a large fraction of trajectories that originated from the south to southwest. Overall, the "snap-shot" view of the troposphere sampled over Alaska for the entire ABLE 3A program appears to represent somewhat typical conditions, as we sampled proportionally from all of the major air mass origin sectors for the region.

OBSERVATIONS AND DISCUSSION
During the ABLE 3A program, the middle/lower tropospheric column over Alaska was frequently stratified. This stratification suggests that the different air masses sampled could have been derived from a variety of source regions. Haze layers were encountered on several missions, and these layers were often apparent in the lidar aerosol data at altitudes from the mixed layer to the tropopause. Brock et al. [1989] observed similar discrete haze layers within the summertime middle troposphere over the high-latitude regions of North America and Greenland. They described particular haze layers, which were depleted in both 03 and odd nitrogen containing compounds (as measured by a luminol instrument) relative to adjacent air, which contained fewer particles. In these cases, an enhancement in the concentration of nucleation mode particles (<0.1 /am) was also observed within the haze layer along with significant enhancements of new particle production at the layer's boundary. Their findings suggested that these haze layers were probably derived from anthropogenic or biomass burning sources, or both. Based upon their observations and those made during the ABLE 3A program, the occurrence of haze layers appears to be a common phenomenon within the summertime middle troposphere over high northern latitudes.
During July and August, numerous "forest" fires were reported throughout most of Alaska lying south of the Brooks Range (see also discussion by Shipham  fraction of the tropospheric air masses that contained distinct haze layers. Their estimates suggest about 10% of the lower 4-km tropospheric column overflown during the ABLE 3A program contained discrete haze layers. They also estimated the 4-to 6-km region contained 1-4%, and the 6-to 8-km region 1-2%. In addition, Browell et al. estimated the fraction of high-latitude "background" air and air of suspected stratospheric origin. They described high-latitude background air as having 03 mixing ratios in the range of 25-35 ppbv just above the mixed layer (0.5-2 km) that increased in proportion to altitude by 5.5-7.5 ppbv/km. Using this characterization, they estimated that of the encountered air masses, background air was observed about 35% of the time in the 2-to 4-km region and about 50% for both the 4-to 6-km and the 6-to 8-km regions. Characterization of air masses of suspected stratospheric origin was based on 03 mixing ratios larger than values found in background air with coincident small values of aerosol scattering. Air of stratospheric origin was estimated to have been encountered about 35%, 50%, and 50% of the time for the 2-to 4-km, 4-to 6-km, and 6-to 8-km altitude regions, respectively. These estimates only apply to the air masses overflow, and their measurements were limited to cloud-free conditions. Several scientific issues merit clarification prior to our use of these classifications for assessing factors controlling the abundance of reactive odd nitrogen. In particular, air sampled during this program that was characterized as originating from the stratosphere usually contained mixing ratios of H20 that are more typical of the middle/upper troposphere (i.e., >> 15 ppmv). In addition, the representativeness of the carefully characterized haze layers derived from Kuskokwim Delta biomass burning emissions should be addressed prior to extrapolating emission factors to the majority of Alaskan or other sub-Arctic fire emissions. In order to further define the possible impact of haze layers and stratospheric/tropospheric exchange on the reactive odd nitrogen budget within the middle/lower tropospheric column (<6 km) over Alaska, several case studies warrant discussion below.

4.1.
Case Studies 4.1.1. Case study mission 6. The air mass sampled near Barrow, Alaska, during mission 6 (on July 12-13, 1988) contained strata that could be characterized as haze layers, air of stratospheric origin, and back•ound air based upon the previous descriptions. Plates 1 and 2 illustrate the lidar soundings of relative aerosol scattering and 03 mixing ratios measured along a portion of a 6.1-km flight leg during this mission (6). The region between about 4.8 and 5.5 km produced small values of relative aerosol scattering in the near-infrared spectral region (IR = 1.06/xm). These values fell within the range of values used to describe air of stratospheric origin. In addition, 03 mixing ratios within this air parcel were enhanced (i.e., 03 > 70 ppbv) compared to mixing ratios in background air. The region near 3 km appeared to consist primarily of background air. Enhanced values of relative aerosol scattering in the visible wavelength spectral region were observed between 7 and 8 km (zenith IR lidar not operational on this mission). This haze layer extended down to the aircraft altitude and was sporadically sampled along with aerosol-depleted/O3-rich air parcels, during both the 6.1-km flight leg and the spiral descent made near point A (as depicted in Plate 1). The vertical soundings produced from in situ measurements of CO, 03, relative humidity (% RH), and NOy taken during this descent are Several meteorological factors contributed to the characteristics of the midtropospheric air mass sampled on this mission. High pressure persisted over central and northern Alaska at the 500-mbar level for several days prior to the flight (July 9-11, 1988). During this period, regions of low pressure were centered over western Siberia and the Gulf of Alaska. The high pressure over Alaska migrated to the Queen Elizabeth Islands from July 11 to 13, 1988, while a broad area of low pressure developed off the northeastern coast of Siberia. This resulted in a southerly flow of air along the high-pressure ridge that extended north through most of western Alaska. Calculated values of Ertel potential vorticity (Epv) at the 500-mbar pressure level exceeded the nominal threshold values often associated with air of stratospheric origin (Epv -> 1 x 10 -5 K hPa -• s -1) in areas near the Gulf of Alaska low and a strong low-pressure system north of the Aleutians (183øW, 55øN). These regions of enhanced potential vorticity covered areas of about 2-4 x 105 km 2 and were centered near the indicated origin of the sampled air mass (15 IøW, 5 IøN on July 8 and near 178øW, 49øN on July 10-11). Three-day isentropic back trajectories along the O = 310øK potential temperature surface (--•500 mbar) indicated an almost due north flow of air along the high-pressure ridge. These trajectories indicated the air mass originated near 168øW, 57øN at the northern end of the Aleutians between the two low-pressure systems on July 10, 1988. These back trajectories also crossed regions where several large fires were burning in the southwestern foothills of the Brooks Range (near 160øW, 66øW) about 24-32 hours prior to reaching the Barrow area (see Shipham  Given these conditions, it may be reasonable to tentatively assign the origin of the encountered haze layers to fires burning in the southwestern foothills of the Brooks Range. The mechanism responsible for transport of the portion of the air mass (air parcel) suspected to have been of stratospheric origin is currently less clear. Several mechanisms, consistent with the meteorological conditions, offer possible explanations: (1) the air parcel was a remnant of a cutoff region formed in association with the Gulf of Alaska low, (2) the air parcel was influenced by jet streak activity on the southern edge of the Bering Sea low, or (3) the air parcel was produced by downward transport of upper tropospheric/ lower stratospheric air following tropopause relocation. Due to the mixing in of tropospheric air, all of these mechanisms could result in air parcels with chemical signatures significantly modified from those characteristic of stratospheric air.
A similar example of O3-enhanced air was found during mission 18 (on July 31, 1988) near Bethel. In this case, Plate 1. Aerosol and 0 3 distribution measurements made on July 12, 1988, during mission 6. The zenith airborne differential absorption lidar (DIAL) aerosol (top panel) and 03 (bottom panel) data are shown in false color display with the relative amount of atmospheric back scattering and 03 mixing ratio in parts per billion by volume (ppbv) defined at the top or bottom of the respective displays. In either case, black represents values greater than the maximum given on the color scale. The altitude is in kilometers above sea level (ASL). Local time is at the top of the aerosol display, and the aircraft latitude and longitude information is given in degrees at the top of the 03 display. however, the O3-enhanced region appeared in the form of a distended tropopause rather than a more well-defined 03enhanced stratum (cf. Plate 3). Back trajectory analysis also indicated that movement of this air parcel was controlled by a ridge of high pressure, which had built up along a line from Norton Sound to Anchorage, concurrent with a low-pressure system over the Aleutians and a strong Polar low located north of Prudoe Bay. Calculated Epv values were larger than the stratospheric threshold value in the vicinity of the Aleutian low, near the region where this air mass appears to have originated. A complete meteorological analysis (in progress), capable of providing a detailed characterization of mechanisms affecting these and other ABLE 3A case studies, is beyond the scope of this paper and will be presented in a subsequent paper(s).
The ratio NOy/O 3 had a value of about 0.12 for air suspected of having a stratospheric origin, which was encountered sporadically along the 6.1-km flight leg of mission 6. This NOy/O 3 ratio was larger than ratios expected from high-latitude middle-stratospheric air (--•0.005 [Murphy et al., 1992]). This larger ratio could suggest a large fraction (--•50%) of the NOy abundance within the depleted-aerosol/ enhanced-O3 air parcel may have originated from sources other than the stratosphere, unless significant air mass aging resulted in an increased loss of 03 relative to NOy during the downward transport of this air from altitudes well above the tropopause (i.e., >4 km above the tropopause). Several compounds exhibited enhancements, which correlated with those of CO, within the haze layers on mission 6. Enhancement factors relative to CO (i.e., AM//XCO) for these layers differed somewhat from those characterized for several of the other haze layers encountered during the ABLE 3A program [Wofsy et al.,  . Enhancement factors for C2H 6 and C3H 8 relative to CO also differed from those found in the haze layers over the Kuskokwim Delta. However, the mixing ratios of these hydrocarbon compounds and CO found outside of the haze layer on mission 6 did fall within the range characterized as typical for background high-latitude air observed during the ABLE 3A program [Blake et al., this issue]. These larger enhancement factors may have resulted from differences in the ecosystem types burnt (e.g., taiga versus tundra/Boreal fores,t), or the ecosystem ages, or the ages of the plumes, or a combination of all these factors. This suggests that a larger range of enhancement factors might be implied for high-latitude haze layers produced by biomass burning than the limited range found over the Kuskokwim Delta region.
The thin O3-depleted stratum occurring near 10-10.5 km (Plate 1) was observed on several occasions. This type of layer was usually observed just below, or in association with, a thin layer of enhanced aerosol scattering. These strata predominantly occurred within about 1 km of a generally well-defined (via measured 03 profile) tropopause. This trend was also observed over high-latitude regions of Canada during the ABLE 3B program (July/August 1990). On several occasions during the ABLE 3B program, portions of this type of stratum were believed to have been sampled following these air parcels' downward transport into the middle troposphere. NOy was found to be enhanced by about 1.5-to 2-fold within these O3-depleted air parcels, whereas CO mixing ratios were near background values (R. W. Talbot  back trajectory analyses along the © = 310 K potential temperature surface (--•500 mbar) indicated northerly air flow extending to about 83øN. Past this point, two possible air mass origins were indicated: the Taymyr Peninsula and the Queen Elizabeth Islands. Analyses along the © = 300 K surface (--•625 mbar) followed a similar path but only toward the Queen Elizabeth Islands. As in mission 6, calculated values of Epv (at the 500-mbar level) exceeded the threshold value for stratospheric air in the vicinity of both of the indicated air mass origins (i.e., the Taymyr Peninsula and the northern end of Baffin Bay, 75øW, 75øN). Although certainly not conclusive, this tends to support the speculation that the 4-km air parcel was of stratospheric origin. km yielded ANOy/AO 3 --• 0.01. This value was once again larger than that expected for stratospheric air. As in the mission 6 case study, the magnitude of the NOy/O3 ratio in conjunction with the general wetness of the air parcel suggests that this air parcel was modified with tropospheric air during its downward transport. Such chemical modification would have been necessary for this air parcel's origin to have been from regions well above the tropopause.
Mixing ratios of C2H2, C2H6, C3H8, and CO within the air parcels near 4 and 3 km were near or slightly smaller than values characterized as typical for high-latitude background air [Blake et al.,  = 300 K potential temperature surface indicated descending northerly air flow for the 3-day period prior to this mission (13). This flow pattern was preceded by cyclonic air flow around a strong Polar low during the fourth and fifth days prior to the mission. These trajectories indicated an air mass origin near 105øW, 75øN (at---550 mbar, on July 20, 1988) that was transported to 162øW, 61øN (at ---700 mbar, on July 25, 1988). Calculated Epv values for the 500-mbar pressure level exceeded threshold values for stratospheric air, on July 21 and 23, 1988, near the center of the low located just west of Banks Island (at about 130øW, 74øN). peared to have had an apparent "lifetime" of about a day, whereas mixing ratios of 03 and perhaps other compounds may have changed by only a small amount (except for possibly H20, which could have been significantly perturbed by even small (5-20%) contributions of tropospheric air containing 0.5-1.0 parts per thousand (pptv) H20). In situ measurements of CO and CH 4 also suggested this air parcel was aged with respect to typical mid-latitude tropospheric conditions. Mixing ratios of both CO and CH 4 were the smallest measured during the program (CO -60 ppbv, and CH 4 -1652 ppbv). These mixing ratios were about 35 and 45 ppbv smaller than those measured in air adjacent to the O3-enhanced air parcel. CO mixing ratios near 60 ppbv are close to those given by Warnek [1988] for the northern hemisphere's lower stratosphere, but are about twofold larger than those found by Murphy et al. [1992] for stratospheric air containing -300 ppbv 03 over Darwin, Australia. This difference most likely reflects the interhemispheric gradient in background CO mixing ratios of approximately 30-40 ppbv. The small CO and CH 4 mixing ratios found in this northern hemispheric air parcel support the contention that this O3-enhanced air parcel originated from the lower stratosphere. general, these events appeared to have covered spatial scales of the order of 105 km 2. Most of the active areas shown in Figure 5 lie within the range of 3-to 5-day back trajectories from Alaska. These areas also exhibited a larger frequency of occurrence (i.e., stronger activity) near northern Siberia during July, and near Greenland during August. This shift followed the eastward migration of the strong Polar low.
Although admittedly circumstantial, these cases do support the idea of significant downward transport into the middle/lower troposphere over Alaska of air that resided within the vicinity of the tropopause ( Inputs from anthropogenic sources: Relatively fresh (<1 day) anthropogenically produced emissions were encountered, but only rarely. Plumes from small villages and the downwind plumes from Bethel and Barrow were encountered on occasion. These sources are believed to contribute negligibly to the regional budgets of NOx, NOy, CO, or NMHC's, although mixing ratios very near the source (<20 km) were as much as 1.5-fold larger than background air values within the mixed layer. Emissions from petroleum industry activity at Prudoe Bay apparently enhanced free tropospheric NMHC mixing ratios to the northeast of this area and were the only important source of fresh anthropogenic perturbations found in the free troposphere over the areas studied [ Shipping lanes still, for the most part, follow fairly narrow corridors along trade wind routes. Within these corridors, ship trails (plumes) can remain relatively well defined over distances of several hundred kilometers [Porch et al., 1990]. The anthropogenic emissions from these multimegawatt (tens to hundreds of megawatts) mobile power plants may provide a locally potent source of NO x to remote and otherwise pristine oceanic regions. These sources could also inject a significant burden of NOx directly into the trade winds, especially at night. This large NOx/NOy fraction could then be converted to HNO3 above the marine mixed layer, in a region with possibly less HNO3 loss due to deposition. This mechanism of NOx transport and transformation may explain at least a portion of the apparently anomalous Mauna Loa data treated by Levy and Moxim [1989] and Moxim [1990]. Active shipping lanes exist along the isentropic and three-dimensional trajectory paths they examined. The NOx/NOy enhancements measured off the Aleutians suggest that the world's ocean surface should not necessarily be treated as a vast, sourceless region for NxOy compounds.
Mixing ratios of CO were not significantly enhanced in the ship plumes encountered on mission 23 (G. W. Sachse, private communication, 1990) and followed a similar trend of large ANOy/ACO enhancement factors (i.e., small CO production) observed in other shipping lanes and in high-altitude aircraft corridors. These relationships emphasize the efficient combustion nature of these high-temperature NOx sources. Thus, deconvoluting the influence of these sources on the remote troposphere may require careful examination of nonstandard combustion tracers. out into four altitude regions centered at 0.75, 2.25, 3.75, and 5.3 km, which cover the altitude ranges of 0.15-1.5 km, 1.5-3.0 kin, 3.0-4.5 kin, and 4.5-6.1 km, respectively. These data are presented in pseudo-three-dimensional form, where the z axis represents the fraction of measurements occurring within a particular mixing ratio bin for each altitude range (Figures 6, 8, and 11). The base data set for these composites was generated by averaging temporarily overlapped measurements of other chemical/meteorological variables into a 3-min time base consistent with the NOx-NOy measurements. These composites used all of the measurements made during each sounding, including chemically enhanced regions or layers. All constant altitude flight legs were, however, excluded from this analysis. In addition, four soundings were excluded from the analyses because they represented duplicates, containing nearly identical vertical structure information as soundings used within the same air mass. The rationale for these exclusions was based on creating composites that would not be overly biased by any one day's events or a particular sampling altitude.  1.5, 1.5-3.0, 3.0-4.5, and 4.5-6.1 km. The relative probability of occurrence for a species mixing ratio within each altitude range is presented on the z axis. ratios of NO, NOx, PAN, NOy, and the individual fractions of NOx/NOy and PAN/NOy. Statistical information for the mixing ratio distributions of each altitude range are given in Table 1. Mixing ratios of HNO3 were not analyzed by our approach due to the long (30-90 min) sampling times used by this instrument. The general altitude trend for HNO3 mixing ratios is discussed in the companion paper of Singh et al.

Composite Chemical Characterization of the
Mixing ratios of NO (Figure 6a) were filtered to include only those measurements made when the solar zenith angle was <67.5 ø. This limited the range of NO2 photolysis rates (JNO2), due solely to solar zenith angle, to about a factor of 2 for the 38 ø to 67.5 ø range of solar zenith angles occurring in this filtered data set. A composite of the JNO2 values calculated for this filtered data set is given in Figure 10f. These JNO2 values were calculated using an equation of the form given by Chameides et al. [1990] to extrapolate zenith and nadir viewing Eppley UV-photometer measurements. The larger than twofold range of JNO2 values resulted from clouds lying above, below, or to the side of the aircraft, where either enhanced (for clouds below and to the side) or depressed (for clouds above) values of JNO2 were encountered relative to clear-sky conditions. In some instances the increased albedo from the pack ice also enhanced JNO2. Further filtering of NO mixing ratios based upon JNO2 values was not done, as the remaining natural variance in JNO2 is believed to represent a composite of typical daytime conditions occurring within about -+5 hours of solar noon.
NO mixing ratios measured at or below the instrumental limit of detection (LOD) represented < 10% of this data set. Composites made from data sets in which these LOD data were omitted, or divided by factors of 1.5 or 2 yielded nearly identical distributions, means, and medians. Mean and median values differed by much less than the standard deviations between the different treatments of LOD data.
Mean and median NO mixing ratio values (-8.5 -+ 2 pptv) were nearly identical and were close to the most probable mixing ratios measured in each altitude range except for possibly the lowest altitude range (0.15-1.5 km). At altitudes >1.5 km the vertical gradient in NO mixing ratios with altitude, implied from this composite, was small (<0.5 pptv/km).
The composite of NOx (NO + NO2) mixing ratios ( Figure  6b), which was not solar zenith angle filtered, had a similar trend to that found for NO. Various treatments of the small fraction (<20%) of NO x mixing ratios measured at or below the instruments LOD also had little effect on the tabulated values or mixing ratio distributions. Mean and median NOx mixing ratios were also close in value (-25 -+ 2 pptv) and exhibited little altitude dependency (<0.5 pptv/km). The slight tails and secondary maxima occurring in the distributions were also indicated in the distributions of several other compounds (e.g., CO and C2H2) and will be discussed in more detail later.
Unlike the mixing ratios of NO or NOx, those of PAN and NOy increased significantly in proportion to altitude (cf. Figures 6a and 6b versus 6c and 6d) Figure 8a). The vertical gradient implied from the median mixing ratios (-9 ppbv/km) was slightly larger than that indicated by Browell et al. [this issue] for background air (5.5-7.5 ppbv/km). The difference between median 03 mixing ratios in the highest two altitude ranges was, however, similar to that defined by Browell et al. for background air.  The tendency toward correlation between mixing ratios of 03 and NOx, at small NOx mixing ratios (i.e., NOx -< 40 pptv) was not anticipated from the trends in 03 and NOx mixing ratios with altitude. This on-average correlation could suggest a nearly linear 1 ppbv AO3/1 pptv ANOx relationship for N Ox mixing ratios <40 pptv. A similar tendency toward correlation occurred between mixing ratios of 0 3 and CO, for CO mixing ratios -<110 ppbv. The nearly 1 ppbv AO3/1 ppbv ACO relationship implied from the regression (for CO -< 110 ppbv) also suggests air mass aging processes may have been important factors controlling the abundance of these compounds. The apparent falloff of the on-average relationship at NOx mixing ratios larger than about 40 pptv could result from an insufficient time for NO x to have photochemically impacted 03 abundance. Indeed,

Estimate of photochemical 03 production versus destruction. A simplified analysis based on in situ measurements can be used for an assessment of the balance between
photochemical 03 production versus destruction for air masses described by this composite data set. Following the general treatment recently summarized by Ridley [1991], photochemical 03 production is controlled by the reactions: where each NO 2 molecule formed via (RI) and (R2) is assumed, upon photolysis, to yield an 0 3 molecule. The photochemical 0 3 production rate, PR(O3) , is directly dependent on the concentration of NO and can be expressed by the equation

The loss processes of 03, under conditions typical of these air masses, is primarily controlled by the reactions
where the O(1D) branching ratio BR is given by

From (1) and (2) the equivalent concentration of NO,
[NOeq], needed for a zero net rate of 03 production (i.e.,

--• 4 x 10 -12 exp (210/T) with kl-2 representing the average of kl and k 2 (rate coefficients from DeMore et al. [1990]). Equation (5) represents the smallest abundance of NO necessary for net photochemical production and provides a useful reference point for comparisons involving the additional assumptions necessary to evaluate (4).
For the NOx mixing ratios measured over Alaska (i.e., NOx < 100 pptv), the concentration of peroxy radicals has been predicted to be nearly independent of NOx concentra- where a and b are linearly fit functions of albedo (a) and altitude (2) (e.g., a = c + da + e2) and X is the solar zenith angle. These clear-sky JO3 values were normalized to JNO 2 values calculated from the Eppley UV-photometers using similarly parameterized clear-sky estimates of JNO 2. This normalization produced results equivalent to those described by Chameides et al. [1990] for the normalization of clear-sky two-stream model photolysis rate estimates. This normalization, based on nadir and zenith Eppley UVphotometers, was assumed to provide a first-order correction for varying albedo. This correction was also assumed to be wavelength independent over the range of 300-400 nm. In addition, JO 3 values were also corrected for daily total 0 3 column over the study region using the tabulated total ozone mapping spectrometer (TOMS) 1988 data. For clear-sky conditions, comparison of values obtained from these parameterized photolysis rate equations to those derived from a two-stream model have agreed to better than _+20% over the range of solar zenith angles contained within this filtered data set (38 ø < x < 67.5ø).

. Median values calculated for the ratio O3eq/O3mea s within the lowest altitude range suggest a small contribution from peroxy radicals on the photostationary state of the mixed layer. This estimate is in general agreement with the analyses by Bakwin et al. [this issue] for surface measurements of NO 2 and NO near Bethel. Larger median mixing ratios of HO 2 + RO 2 (-75 pptv)
were, however, implied from our calculations for the highest altitude range. This latter results, in conjunction with the large difference between mean and median O3eq/O3mea s ratios, suggests a more complete understanding of the NO2/NO photostationary state relationships preliminarily investigated here may require (1) refined model calculations, which incorporate actual conditions (e.g., the use of actual hydrocarbon abundances, representation of actual cloud fields, and changes in total 03 column), and (2) reinvestigation of small possible interferences in measured NO 2 (e.g., 5-10 pptv of thermally/photolytically derived interference from 100-fold larger mixing ratios of unaccounted for NOy compounds). These results also suggest that without such reanalyses, which are in progress, comparisons of the model-calculated mixing ratios of NOx necessary for photochemical 03 production to mixing ratios of NOx that were measured may be less interpretable than similar analyses based on NO mixing ratios. The mixing ratios of NO were more directly measured than those of NO 2, allowing higher immunity to interferences, and absolute accuracy. Based on the preceding discussions, we believe the NO-based analyses more accurately describe the role of NO on the photochemical lifetime of 03. We find that on average the NO mixing ratios measured over Alaska were about twofold smaller (within the 1.5-to 5-km altitude range) than those necessary to balance the photochemical rates of 03 production and loss.

4.2.4.
Distributions and trends of representative carboncontaining compounds. The compounds CO and C2H2, which are primarily products of combustion, also exhibited altitude dependencies in their median mixing ratios (--•1 ppbv/km for CO and --•5 pptv/km for C2H2; cf. Figures 8c and 8d). However, the most probable portion of the mixing ratio distributions for these compounds varied only slightly with altitude. The pronounced tails and secondary maxima in their distributions contributed to most of the differences (--•10-15% for CO and 20-30% for C2H2) between the most probable and the median mixing ratios for these compounds. Similar tendencies were also measured for mixing ratios of C2H6, and C3H8 (cf. Figures 8e and 8f), even though these latter compounds have both combustion-and noncombustion-related sources. These similarities may represent the effects of either common sources (or source regions) or sinks for these compounds. The ratios formed from various carbon-containing compounds are useful as indicators of relative air mass age when the compounds chosen share common sources and sinks would have been similar to that estimated for C2H6, and the lifetime of C2H 2 (--•14 -4 days) would have been similar to that estimated for C3H8. The overall uncertainty in these lifetime estimates would most likely be much larger than indicated by the pressure and temperature dependencies of the rate coefficients, due to uncertainties in the estimates of average OH concentrations and the exclusion of surface sink terms. Even so, these estimates could be useful in describing the general trends that might be expected between compounds and the order of magnitude of their lifetimes in relation to transport time (At) from distant sources. The relative difference in elapsed time (equated here to transport time At) for compounds with different chemical lifetimes can be approximated by

In [(CA/CB)tl/(CA/CB)t2 ] At = (12) (k A -kB)CoH
where C^, CB are the concentration of compounds A and B at times t 1 and t2, having reaction rate coefficients k^ and kB, and Con is the OH concentration.
Mixing ratios of C2H 2 were significantly correlated (r 2 0.87) with those of CO (slope --•1.7 pptv/ppbv; Figure 12a) as might be expected from compounds that share both common sources (i.e., combustion) and sinks (i.e., oxidation via O atom addition reactions involving OH). The composite of individual C2H2/CO ratios (Figure 11a) exhibited an increase in median values in proportion with altitude. This trend could suggest an influx of less aged (At --20-30%) air into the 4-to 6-km altitude range. Average mixing ratios of C2H 6 and C3H8 were also correlated with those of CO (cf. Figures 12b and 12c). In addition, the average ratio of C3Hs/C2H 6 was correlated with ratios of C2H2/CO , which suggests common factors (either sources or sinks) were controlling the mixing ratios of all of these compounds. The regression of In (C3H8/C2H6) versus In (C2H2/CO) yielded about a threefold smaller slope than expected from (12), based upon equivalent transport times (At), OH concentrations, and rate coefficients representative of the tropospheric conditions encountered (i.e., expected slope --• 1.1 versus --•0.4; cf. Figure 12d). Although very qualitative, this could suggest various source emission factors (signatures) were a more important factor than the OH photochemical sink in controlling the average relationships observed for these compounds. Relative emission factors (e.g., AC2H2/ACO and AC3H8/C2H6) were found to  Figure 13). These similarities in altitudinal trends may suggest these compounds shared either common source(s) (e.g., the result of industrial/urban emissions) or common sinks (e.g., lower altitude sinks such as surface deposition and low-altitude thermally induced loss of PAN). These trends could also simply represent a tendency for long-lived compounds to accumulate in the Arctic middle troposphere.
Unlike the correlations with PAN and NOy, average mixing ratios of C2C14 were only slightly correlated with those of NOy and CO. In addition, average mixing ratios of CO were correlated with those of NOy, but not with those of PAN (cf. Figures 14a and 14b). This correlation between CO and NOy was in contrast to other lower-latitude remote tropospheric measurements that have generally found little correlation between CO and NOy for NOy mixing ratios <1 ppbv [e.g., Parrish et al., 1991;Hiibler et al., 1992]. The slope of the regression between average CO and NOy mixing ratios (slope --•0.04 ppbv/pptv) was also about twofold to fourfold larger than the slope obtained from the portion of their mid-latitude measurements that did exhibit correlation for NOy mixing ratios > 1 ppbv [Parrish et al., 1991]. This As might be expected, the air masses originating from the Gulf of Alaska possessed a chemical signature with median mixing ratios of nearly all trace gases that were smaller than those in other air mass categories or in the overall composite (Comp). The significantly smaller (twofold to threefold) mixing ratios of PAN accounted for most of the reduction in median NOy mixing ratios, as evidenced by the small median values of the ratio PAN/NOy. The smaller median PAN mixing ratios found the Gulf of Alaska air masses are believed to represent the general lack of Nx Oy sources from this region, since median temperatures do not support enhanced thermal decomposition rates of PAN within these air masses. In general, the smaller median mixing ratios of C2H2, C2H6, and C3H8 also support the idea that these air masses were relatively disconnected from these compounds' sources. During summer, approximately 45% of the air masses influencing western Alaska originate from sectors covering the North Pacific Ocean. Taking this fraction, in conjunction with the somewhat smaller median NOy mixing ratios, suggests that about 35% of the middle tropospheric NOy burden may have originated from these regions. This fraction may represent the contribution of relatively wellaged lower-latitude background air.
In 'contrast to the Gulf of Alaska air masses, those categorized as plumes possessed a chemical signature with median mixing ratios of nearly all trace gases that were larger than those in the overall composite or in most of the other air mass categories. Median PAN mixing ratios were particularly enhanced within the lowest altitude range. This enhancement, despite the warmer air temperatures, suggest that NxOy emissions from short-range sources were responsible for these observations. In addition, the significantly enhanced median NOy mixing ratios within this air mass category indicate a large fraction of the NOy budget consisted of compounds other than NO x and PAN.
The air masses categorized as originating from mid-Siberia possessed a chemical signature nearly identical to that found for the plumes. In particular, median values of potential temperature, CO, and H20 are remarkably similar in comparison to median values for the other air mass categories.
The smaller median value of PAN within the lowest altitude range may represent the thermal loss of this reservoir compound during transport over the relatively NxOysourceless Bering Sea. This result also supports the earlier suggestion that some of the enhancement in PAN mixing ratios, in the plume categorized air masses, may have originated from more localized N xOy emissions. The larger median values of the ratios C2H2/CO and C3H8/C2H 6 also imply that these two air mass categories are similar. These ratios also suggest a relatively young age for these two air mass categories.
In Regionally, however, high-latitude biomass burning could have a substantially larger impact. In particular, approximately 50% of high-latitude biomass burning is believed to occur in Siberia [Stocks, 1991] ANOy/ACO emission factor applies to taiga fires, as might be suggested by the plumes encountered on mission 6, then nearly 80% of the total dry and wet deposition flux could be balanced from the Siberian biomass burning source. These estimates suggest biomass burning may have been an extremely important regional-scale source for air originating from central Siberia.
During summer, approximately 20% of the air masses influencing western Alaska originate from sectors that could be affected by mid-Siberian emissions. Considering this contribution along with the nearly 1.5-fold larger median NOy mixing ratios found in air masses from this sector suggests that about 30% of the NOy burden over western Alaska could be due to biomass burning in mid-Siberia. This burden could be in addition to any general enhancement background in high-latitude NOy mixing ratios due to biomass burning. This estimate of the NOy burden due to biomass burning is in agreement with the multivariant analyses of Wofsy et al. [this issue]. This NOy burden is, however, a larger contribution than might be implied from the analyses of Browell et al. [this issue], in which the fraction of air masses containing discrete haze layers was of the order of 10%. However, the fraction of this composite data set containing chemically enhanced air with CO mixing ratios > 115 ppbv was approximately 22%, suggesting that cloud processing of aerosol (i.e., removal) may have contributed to an underestimate of "plumes" based on lidar aerosolscattering measurements. Indeed, several chemically enhanced layers were encountered that contained very small enhancements in aerosol number densities.
Air masses originating from north of Siberia and from the Arctic pack ice should have been relatively free from the short-range effects of biomass burning. The larger median mixing ratios of C2C14 found in higher altitudes suggest these air mass categories may reveal the chemical signature associated with long-range transport from European and North American industrial regions. This, coupled with the average frequency of occurrence of air masses originating from these sectors (--•30%), could suggest that about 35% of the NOy burden over western Alaska could have also been due to the long-range transport of pollutants from lower latitude industrial/urban regions. However, downward transport of tropopausal NOy may have contributed to a portion of the NOy in these and other air mass categories, which originated from regions with larger frequencies of occurrence of enhanced potential vorticity.
The air categorized as originating from downward transport of tropopausal air (Strat in Table 2) did not exhibit a chemical signature markedly different from the other air mass categories, except for the plume and mid-Siberian types. This could represent the loss of a distinct stratospheric signature as the air was advected during transport and mixed with upper/middle tropospheric air. Alternatively, the lack of a distinct signature could imply that high-latitude tropopausal air possesses a chemical signature closer to that of tropospheric air, with the exception of enhanced 03 mixing ratios derived from the stratosphere.
We can only speculate on the chemical characteristics of the tropospheric air residing above 6 km over these regions. However, if the springtime high-latitude NOy profiles measured by Dickerson [1985] are paradigmatic, then continued increases in proportion to altitude in the mixing ratios of many compounds, especially NOy, should not necessarily be expected for the upper troposphere. Based on Dickenson's measurements, a substantial reduction in NOy mixing ratios may occur between 6 km and the air above. This coupled with the trends described in the case studies suggest the influx of tropopausal air could be an important source of NOy into the region. Even so, it would seem unlikely that the middle stratosphere is the ultimate source of this NOy, based on the case studies presented, unless an efficient means of converting stratospheric HNO 3 into other NxOy compounds exists. This adumbrates other NxOy sources  of NOy mixing ratios analyzed as stratospheric that were highly correlated with those of 03 but not with those of CO The corollary argument could, however, imply that an equally large fraction may be due to sources such as highaltitude aircraft emissions.
Significant uncertainty (perhaps twofold) exists in all of the estimates discussed above. Even so, these analyses are believed to depict at least a semiquantitative estimate of the factors controlling the summertime abundance of NxOy compounds over western Alaska. In particular, the largest degree of uncertainty is involved in assessing the relative contribution from anthropogenic sources. If the median mixing ratios of CO measured within the Gulf of Alaska air masses represent a relative baseline for comparing the impact from lower latitude sources, then for CO mixing ratios of 90-100 ppbv, approximately 200-400 pptv of NOy could be attributable to anthropogenic sources based on the regression discussed earlier. This would predict that somewhere between one-and two-thirds of the abundance of NOy could have originated from such sources. In conjunction, about one-third to one-half of the NOy is indicated as having originated from biomass burning and another one-third to one-half as having originated from the downward transport of tropopausal air.

SUMMARY
We have attempted to assess the factors that control the abundance of NxOy compounds in the lower 6-km tropo-spheric column over Alaska. Our analyses identified several potential factors that may significantly influence tropospheric chemistry over this region. The most prominent of these are posed below in the form of questions.
1. What are the chemical characteristics of the summertime high-latitude tropospheric column above 6 km, and of the tropopausal region? 2. Is there an efficient mechanism for converting stratospherically derived HNO3 into other NOx-containing reservoir compounds?
3. To what extent does the 3-to 6-km altitude region at high latitudes represent a "stable" regime for accumulating surface-emitted pollutants and 03 from aloft? 4. How variable are emission factors of trace gases from high-latitude biomass burning, especially throughout the vast regions of Siberia? 5. How important are the anthropogenic NxOy source inventories for Siberia (or Russia)? 6. What compounds constitute the large fraction of NOy not accounted for by NOx, PAN, and HNO37 How "reactive" a form of odd nitrogen are they? 7. How extensive and variable are quasi-localized regions of enhanced potential vorticity and how well do these regions reflect patterns of stratospheric/tropospheric exchange? What is the climatology of summertime highlatitude stratospheric/tropospheric exchange (see also similar question by Staehelin and Schmid [1991])? How would quasi-localized areas of enhanced exchange affect the interpretation of current ozonesonde data bases? 8. How important are the observed long-term trends in high-latitude biomass burning to the chemical climatology over high latitudes ? Are the temporally coincident long-term tendencies in high-latitude biomass burning and enhanced middle-tropospheric ozone potentially caused by climate changes [cf. Van Wagner, 1988;Logan, 1985]? Are these trends in burned areas associated with climatological patterns that may suggest large-scale atmosphere/biosphere couplings?
These questions are important, and their answers should be addressed in future modeling and field program activities. patience in enduring the many slow spiral descents that are featured in this article. We are especially grateful to Roger Navarro, Jim Hoell, Richard Bendura, Joseph Drewry, and Helen Thompson for coordination of the operational and logistical support that was so crucial to the success of this program. Our highest commendation goes to Bob Harriss for his ability as mission scientist to juggle the desires of the science team, the demands of the program goals, and the diversity of nature. We are also grateful to J. Ward, J. Bradbury, S. Simms, S. Shurling, and D. Yang for their contributions toward data analysis and preparation of this manuscript. This research program was sponsored by the National Aeronautics and Space Administration, Tropospheric Chemistry Program Office under the Program Directorship of Robert J. McNeal.