Reactive nitrogen oxides and ozone above a taiga woodland

Measurements of reactive nitrogen oxides (NOx and NOy) and ozone (03) were made in the planetary boundary layer (PBL) above a taiga woodland in northern Quebec, Canada, during June-August, 1990, as part of NASA Artic Boundary Layer Expedition (ABLE) 3B. Levels of nitrogen oxides and 03 were strongly modulated by the synoptic scale meteorology that brought air from various source regions to the site. Industrial pollution from the Great Lakes region of the U.S. and Canada appears to be a major source for periodic elevation of NOx, NOy m(cid:127)d 03. We find that NO/NO2 ratios at this site at midday were approximately 50% those expected from a simple photochenfical steady state between NOx and 03, in contrast to our earlier results from the ABLE 3A tundra site. The difference between the taiga and ttmdra sites is likely due to much larger emissions of biogenic hydrocarbons (particularly isoprene) from the taiga vegetation. Hydrocarbon photooxidation leads to relatively rapid production of peroxy radicals, which convert NO to NO2, at the taiga site. Ratios of NOx to NOy were typically 2-3 times higher in the PBL during ABLE 3B than during ABLE 3A. This is probably the result of high PAN levels and suppressed formation of HNO3 from NO2 due to high levels of biogenic hydrocarbons at the ABLE 3B site.

BAKWIN ET AL.: NOx, NOy AND 03 ABOVE A TAIGA WOODLAND northeastern Canada and the eastern United States (e.g., Talbot et air at the inlet of the converter of =2 cm 3 STP of a National Instial., Sandholm et

al., Wofsy et al.). tute for Standards and Technology (NIST) (Gaithersburg, Mary-Talbot et al. [this issue] segregate the aircraft data obtained in land) traceable standard gas containing 4.42 ppmv NO in N2. various altitude intervals with air mass type, on the basis of CO
Checks on the conversion efficiency for NO2 were performed mixing ratios. We take a different approach and average the PBL twice daily. For an efficiency check the NO standard was added data for each aircraft mission to determine the dominant influences to 50 cm 3 STP of "zero" air containing =2 ppmv 03 in a 4 cm 3 on the chemical composition of PBL air during the time period of volume just before being introduced into the converter, so that each flight. We found a high correlation between C2H2 and CO in =99% of NO was converted to NO2. The conversion efficiency biomass burning and industrial pollution plumes and therefore remained >95% for the full observation period. ha -l, and the mean canopy height was 5-6 m. A 31-m-high tower (Rohn 25G) was erected and instrumented for chemical and mi-of the conversion efficiency for higher N oxides were done in the field. In laboratory tests using HNO3 in humidified "zero" air, efficient conversion to NO was obtained at sample flow rates up to 5000 cm 3 STP. Also, conversion efficiency for NH3 in humidified air (>20% relative humidity (RH) at 23 ø C) was found to be negligible in agreement with results from other laboratories ((3. Hltbler, personal communication, 1990).
Occasionally local pollution (mainly from the generator) was sampled at the tower. During these periods the variance of the NOy measurements was greatly increased (coefficient of variation > 2 for a 5-min period), so that such intervals were easily identified and were removed from the data set.
A separate detector was used to measure mixing ratios of NO and NO2. Nitric oxide was measured by chemiluminescence with 03, and NO2 was measured following photolysis to NO [Bakwin et al., 1992] in a 165 cm 3 quartz cell at a sample airflow rate of 800 crometeorological measurements. Chemical analyzers and data cm 3 STP and pressure of 300 torr. The reduced pressure in the acquisition and control computers were located in a tent about 20 photolysis cell minimized the conversion of NO to NO2 by amm southeast from the base of the tower. Electrical power was pro-bient 03 [Ridley et al., 1988;Bakwin et al., 1992]. During samvided by a 12.5-kVA diesel generator located 300 m southeast of pling from each altitude, 2 min were spent in an NO mode and 2 the tower.
The experimental design was similar to that described by Bakwin et al. [1992]. Mixing ratios and turbulent fluxes of NO v and 03 were measured at 29 and 31 m height, respectively. Mixing ratios of 03, NO and NO2 were measured by sampling through 0.635 cm OD. Teflon tubes with inlets fixed at 0.05, 0.85, 2.8, 6.2, 9.5, 18.2, and 30.8 m height on or near the tower. The sampling sequence was from highest to lowest, each location was sampled for 4 min during each profile. The NOx detector was zeroed following sampling from the 0.05-m tube.
Reactive nitrogen oxides (NOv) were converted to NO by goldcatalyzed reaction with H9• at 300øC [Fahey et al., 1986] and quantified using chemiluminescence with 03. The converter consisted of a 90-cm-long, 0.635 cm ID gold-plated (2.5 gm thickness) copper tube and was located at the inlet end of the sampling tube. Calibration gases and H9• were added to the sample air through two 0.16-cm-OD stainless steel tubes that intruded several centimeters into the inlet of the converter tube. The converter design minimized instrument response time to NOy species by minimizing contact of sample air with nonconverting surfaces [Bakwin et al. 1992; M93]. The instrument responded to a pulse input of HNO3 with a 90% risetime under 2 s.
After passing through the converter, sample air passed through =40 m of 0.635-cm-OD Teflon tubing to the NO analyzer. Sample flow rate was 900 cm 3 min -• STP (cm 3 STP). An instrument zero was obtained every 40 min by addition of 50 cm 3 STP of "zero" air containing =100 ppmv (parts per million by volume) 03 (generated using a Hg vapor lamp) to the sample air just downstream of the converter, so that NO was converted to NO9.. Calibrations for NO were carried out every 3 hours by addition to the sample •nin were spent in an NO2 (photolysis) mode.
To zero the NOx detector, a solenoid valve was used to introduce a flow of 50 cm 3 STP "zero" air containing =100 ppmv of 03 to the sample air just upstream of the quartz cell, with no photolysis, so that NO was removed. At night, reaction with ambient 03 is expected to completely remove NO, providing a check on the zeroing procedure. The mean (standard deviation) nighttime NO mixing ratio for 2.8 to 30.8 m was 0.2 (1.1, n=2124) pptv (parts per trillion by volume), with no significant difference between sampling heights (see Figure 3). Nitric oxide mixing ratios at 0.05 m were elevated somewhat due to emission of NO from the surface (see below). Previously, we reported an artifact of --1.7 pptv for our NO measurements [Bakwin et al., 1992]. Introduction of a solenoid valve to switch the O3-1aden zeroing air in and out of the sample airstream appears to have eliminated this artifact (<0.5 pptv).
Calibrations for NO and NO2 were performed every 3 hours using the methods described by Bakwin et al. [1992]. The NO (compressed gas) and NO2 (permeation tube) calibration standards were compared in the field using gold-catalyzed reaction with H2 to completely convert NO2 to NO. The NO standard was considered the primary standard in the field and was referenced before and after the field campaign to two NIST standards maintained in our laboratory. No significant change was observed in the working sta•,dard between these two comparisons. Estimates of the precision and accuracy of the NO and NO2 measurements are given by Bakwin et al. [1992].
An ultraviolet (UV) photometer (Dasibi 1003-AH) was used to determine 03 mixing ratios. The zero level of the photometer was determined frequently by passing the sample air through a screen impregnated with MnO2 and was found to be stable to better than +0.5 ppbv (parts per billion by volume) during the field experiment. To ensure a consistent calibration for ground and aircraft ozone observations, the photometer was compared in the field to a similar instrument calibrated at NASA Langley Research Center. Ozone was also measured continuously at 30.8 m height using a Bendix C2H4-chemiluminescence detector that was modified for fast response, and the data were used to compute turbulent fluxes of 03 (M93). Solar UV radiation was measured at the tower site using a radiometer identical to those flown on the Electra (Eppley Laboratory, Newport, Rhode Island). The radiometer was mounted on a pole at 6.1 m height, just above the tops of most nearby trees (5-6 m high). The radiometer output was used to compute the photo- ppbv, respectively, were observed near midday, coincident with the maximum rate of vertical mixing. At mght, NOy and 03 levels in the surface layer were depressed due to deposition and decou-Fitzjarrald and Moore [this issue] discuss climatological pling from the atmosphere above (M93). Mixing ratios of NO, changes that occurred during the period of our observations. They were also somewhat lower at night than in the daytime. The report a shift in the synoptic regime around Julian day 197, from a nighttime loss rate for NO, (about 1.3 pptv h 4 on average) prob-cool period of frequent precipitation and westerly to northwesterly ably reflects net deposition to the surface (see below) as well as flow to a warmer, drier period characterized by southwesterly flow reaction of NO2 with 03 to form NO3, which may be lost by depo-and generally deeper afternoon boundary layers. A rerum to sition or via further reactions. cooler weather and westerly to northwesterly flow occurred about Statistics for NOx, NO v, and 03 mixing ratios during midday day 221. These changes in climate are reflected in the NO, and (1000-1600 eastern standard time (EST)) for Julian days 178 to NOy mixing ratios ( Figure 1); on average, higher daytime levels 229, 1990, are given in Table 1. Probability distributions for NOx were observed during the period of generally southwesterly flow and NOy were somewhat skewed due to a relatively small number than during the cooler periods (Table 1). Ozone levels were not of observations with very high mixing ratios: median NO, and significantly correlated with the climatological changes. NO v mixing ratios were lower than the means. The 03 data were Figure

Pollution Influences
The time series of NO,, NOy, and 03 (Figure 1)clearly show coherent variations, =1 (e.g., day 195) to 8 (e.g., days 193 to 201) days duration, correlated with surface pressure changes indicating the influence of synoptic scale air mass characteristics. Typically, NOx, NOy, and 03 were enhanced during periods of falling surface pressure, which tended to be associated with air originating from the southwest quadrant, according to 5-day back trajectories for surface air. Mixing ratios were relatively low during periods of rising pressure, most often associated with trajectories from the N hemicircle, indicating Arctic air. Regions to the southwest were apparently significant sources of NOx, NO v, and 03 at the surface.
Mixing ratios of NO, and NOy measured at the tower appear to have been somewhat higher than those measured si[nultaneously aboard the aircraft. Direct comparison is difficult since the aircraft data represent means taken over [nany kilometers, and because of the small number of data points. The ground site may have been located in a zone of somewhat elevated N oxide levels due to natural processes or local anthropogenic pollution. However, the overall agreement of NOdNOy ratios between the ground site and the airplane (see below and Figure 7) indicate that local pollution was probably not a significant proble•n at the ground site. Table 2 shows mixing ratios of C2H2, C2Ch, NOy, and 03 on the tower and from the Electra (PBL only) for selected flights. Acetylene is emitted by bimnass and fossil fuel burning and has a lifetime of in the troposphere of 2-3 weeks, while C2Ch is a purely in-EST) and midday (1000-1600 EST). Ozone was depleted at the dusthal product with a lifetime of 12-14 weeks. Mixing ratios of lowest altitudes due to surface uptake, which was strongest during NOy and 03 at the ground site were near or below their daytime the daytime (M93). Mixing ratios of NO were somewhat elevated background values of 247 pptv and 28 ppbv (as defined by mediat 0.05 m height compared to other altitudes, indicating that NO ans for days 178-196; see Table 1), respectively, on days 215, 219,    highly industrialized area. The enhancements of C2C14 and C2H2 at the tower on day 221 were much smaller, indicating that anthropogenic pollution, though present, was more dilute than during the episode of days 209-211. These results indicate that distant industrial sources had a pronounced influence on NOx, NO v, and 03 levels observed at the ground site. The higher mixing ratios of 03 in the polluted air masses most likely reflect net photochemical production (or suppressed destruction) in the presence of elevated

Enhancements of C2H2 (and other hydrocarbons) observed on days 217 (mission 13) and 220 were not associated with increased C2C14 or NO v and hence were most likely due to boreal biomass fires. Detailed investigations of biomass burning and industrial pollution plumes observed from the Electra over Alaska (ABLE 3A [Wofsy et al., 1992]) and northeastern Canada (ABLE 3B [Talbot et al., this issue; Wofsy et al., this issue]) show that NOv/CO and NOv/hydrocarbon emission ratios are low in biomass fires compared to industrial emissions. Jacob et al. [1993] have used a chemical tracer model (CTM), with transport fields from the Goddard Institute for Space Studies global circulation model [Hanson et al., 1983], and with parameterized photochemistry [Spivakovsky et al., 1990], to simulate mixing ratios of 03 and its precursors over North America. The model produces mixing ratios of 03, NO•, and CO in general agreement with observations taken at a number of sites throughout the United States, and successfully simulates episodes of high 03 in the eastern United States that occur during periods of stagnant meteorology in summer (see Jacob et al. [ 1993] for details).
Table 3 compares the frequency distributions for NO• generated from the model for 1430 EST during June-August in the (4 ø x 5 ø) grid cell containing the Schefferville tower, with those observed at the tower (1000-1600 EST). We compare the model results and observations for daytime only because at night the observed mixing ratios are somewhat depleted below the shallow inversion by surface deposition and chemistry (see Figure 2) (equation (1)), all of the parameters required to compute P were measured at this site. The value of P, computed for NOx and 03 measurements taken above the treetops at 6.2, 9.5, 18.2, and 30.8 m, 1 day). The NOx/NOy ratios appear to increase somewhat However, NOx levels at our site are much lower that at Niwot at the lowest NOy abundances, but this may be due in part to ran-Ridge, rarely exceeding 150 pptv, and we find no significant rela-dom measurement errors at low NO• and NOy. Data obtained durtionship between peroxy radicals and NO• at our site. The latter ing ABLE 3A over the tundra of SW Alaska also showed lower result is in agreement with the observations of Ridley et al. [ 1992], NO,JNOy ratios, 0.08 (+_0.02, also apparently uncorrelated with the LOCAL TIME (hours) mixing ratio of NOy [Bakwin et al., 1992], indicating important differences between processes that control NO• mixing ratios in these two regions.

DIscussioN
The low mixing ratios of NO•, NOv, and 03 and the low NOx/NOy ratios observed confirm that this woodland site is indeed remote from direct industrial influence. During periods when "background" (Arctic) air was sampled, NO v mixing ratios (generally 180-340 pptv) were among the lowest seen at any continental location, and were similar to levels we observed at a tundra site in southwest Alaska during ABLE 3A [Bakwin et al., 1992]. NO• mixing ratios were also low in "background" (Arctic) air, typically 2040 pptv, but were 2-3 times higher than at the ABLE 3A tundra site. Background NO• levels at the taiga site were near the expected crossover point for net production/destruction of 03 by photo- taiga (ABLE 3B) sites associated with pollution from industrial and/or biomass-burning sources. Pollution events are especially evident in the ABLE 3B data set (Figure 1). It is difficult or impossible to separate the effects of industrial processes from those of biomass fires on the basis of the NOx, NOy, and 03 data alone, however the hydrocarbon and halocarbon data ( Table 2) clearly indicate that industrial pollution is a source for trace gases measured at the taiga woodland site. Future studies should include al. [1993] show that essentially all of the transport of industrial pollution occurs in the PBL.
In the CTM, roughly 5% (0.9 x 109 g N d -1) of the NOx emitted by industrial sources in the United States and Canada is transported to the Arctic northward of 60øN during summer. This is approximately equivalent to each of the sources of NO• to the Arctic troposphere from biomass burning [Jacob et al., 1992;Wofsy et al., this issue] and from the stratosphere [Jacob et al., 1992]. North American sources supply approximately 35% of the global source of NO• from industrial processes, and a large portion of the remainder is emitted at midlatitudes in Europe and Asia. If a similar fraction (5%) of emissions from all industrial sources is exported to or emitted in the Arctic, the total industrial source of NOy to the Arctic in summer would be about 2.6 x 109 g N d -1. Examination of statistics for NO• emission from industrial sources compiled by Harneed and Dignon [1988] indicates that the 5% estimate may be reasonable for Europe and Asia, where industrial centers reside farther north than in North America.
Measured wet and dry deposition of NOy at our site totaled about 35 g N ha -] month -] (M93), very similar to deposition rates at the ABLE 3A tundra site at 62øN (36 g N ha -1 month -1 [Bakwin et al., 1992;Talbot et al., 1992]). If we assume that deposition rates measured at these two sites are representative of the world northward of 60øN (3.4 x 10 9 ha) in summer, we would calculate a deposition flux of 4.0 x 10 9 g N d -1, giving a rough budget for NOy for the Arctic troposphere in summer. Sources from industrial pollution (2.6 x 10 9 g N d q) would account for =50% of the total, with the balance from biomass burning (0.9) and the stratospheric input (0.9), approximately balanced by wet and dry deposition (4.0). Mixing ratios of NOy observed in the Arctic are much higher in winter than in summer due to increased transport from midlatitudes and reduced loss rates [Honrath and Jaffe, 1992 Table 1) were on average twice those observed during cooler periods of westerly and northwesterly flow (178-196 and 222-229). Higher NOx is expectduring mission 10 (as well as industrial pollution encountered in ed to lead to greater net production of 03; however, 03 levels were the marine boundary layer off northeast Canada during mission not significantly different between these two broad intervals. 16), and they focused largely on data above the PBL. The tower Southwesterly flow is generally associated with subsidence, with data clearly show that the episode on days 209-211 was not strongly capped PBLs [Fitzjarrald and Moore, this issue], so that unique, reflecting a pattern that repeated several times. The the supply of 03 from aloft may be reduced. Also, since the loss meteorological conditions leading to transport of industrial pollu-of 03 by surface deposition is about twice as fast as photochemical tion from the Great Lakes region (i.e., high pressure over much of production, our result is perhaps not surprising. Further, the northeast Canada) are associated with subsidence, and results froin periods of strongest southwesterly flow and highest levels of inthe three-dimensional chemical tracer model (CTM) of Jacob et dusthal pollution, such as the episodes of days 209-211 and day 221, we observed a clear enhancement of 03 over background lev-"missing" during downslope periods [Atlas et al., 1992], at our els (Figure 1). The overall impact of industrial pollution on re-ABLE 3A site about half of the observed NOy is "missing" and apgional 03 levels is moderated by the rapid loss of NOx within the pears to consist of fairly stable species that are resistant to deposi-PBL and by deposition of 03 to the surface.
tion [Bakwin et al., 1992]. At the Pennsylvania site [Buhr et al., We find that NO/NO2 ratios observed at the taiga site are not 1990] and our ABLE 3B taiga site [Sandholm et al., this issue], well described by a simple photochemical steady state involving nearly all of NOy is accounted for by measured species. This only NOx and 03, indicating that other oxidants are important in comparison may indicate that the "missing" NOy species do not converting NO to NO2. Peroxy radicals produced from isoprene play a major role in the budgets of NO• at these sites. oxidation are likely the major cause of low the NO/NO2 ratios (F93). Oxidation of NO by 03 alone is sufficient to explain Acknowledgments. This work was supportedby NASA grant NAG1-55 NO/NO2 ratios we observed at the tundra site during ABLE 3A to Harvard University and by the Alexander Host Foundation. We are [Bakwin, 1989], consistent with estimates of low peroxy radical grateful to M. Shipham for useful discussions concerning the meteorologimixing ratios and production rates from hydrocarbon precursors cal context for these measurements, to J. Barrick for loan of the calibrated [Jacob et al., 1992].
UV radiometer, and to S. Fan for helpful discussions concerning photo-Though NOy levels were similar in "background" air at the tun-chemistry at the ABLE 3B site. We also thank D. Barr