On the export of reactive nitrogen from Asia : NO x partitioning and effects on ozone

The partitioning of reactive nitrogen (NO y) was measured over the remote North Pacific during spring 2006. Aircraft observations of NO, NO2, total peroxy nitrates (6PNs), total alkyl and multi-functional nitrates ( 6ANs) and nitric acid (HNO3), made between 25 ◦ and 55 N, confirm a controlling role for peroxyacyl nitrates in NO x production in aged Asian outflow.6PNs account for more than 60 % of NOy above 5 km, while thermal dissociation limits their contribution to less than 10 % in the lower troposphere. Using simultaneous observations of NO x, 6PNs,6ANs, HNO3 and average wind speed, we calculate the flux of reactive nitrogen through the meridional plane of 150 ◦ W (between 20 ◦ and 55 N) to be 0.007± 0.002 Tg N day−1, which provides an upper limit of 23± 6.5 % on the transport efficiency of NO y from East Asia. Observations of NO x, and HOx are used to constrain a 0-D photochemical box model for the calculation of net photochemical ozone production or tendency ( 1O3) as a function of aircraft altitude and NO x concentrations. The model analysis indicates that the photochemical environment of the lower troposphere (altitude < 6 km) over the north Pacific is one of net O3 destruction, with an experimentally determined crossover point between net O 3 destruction and net O3 production of 60 pptv NOx. Qualitative indicators of integrated net O3 production derived from simultaneous measurements of O3 and light alkanes (Parrish et al., 1992), also indicate that the north Pacific is, on average, a region of net O3 destruction.


Introduction
The partitioning of reactive nitrogen (NO y ), among its various oxidation products, determines the spatial scales by which NO x (NO x ≡NO + NO 2 ), or its temporary reservoirs, are transported.As a result, NO y partitioning impacts the production rates of both ozone (O 3 ) and secondary organic and inorganic aerosol on local, regional and global scales.Analysis of trends in both O 3 concentration and reactive nitrogen on the west coast of North America indicate that background O 3 concentrations over the North Pacific have steadily risen over the past three decades (e.g., Jaffe et al., 2003;Parrish et al., 2009), with speculation that the growth in NO x emission rates in eastern Asia is largely responsible (Richter et al., 2005;Zhang et al., 2007;Zhang et al., 2008;Zhang et al., 2009;Walker et al., 2010).The increase in background O 3 concentrations may limit the efficacy of future local and regional O 3 control strategies, particularly in California (Jacob et al., 1999;Cooper et al., 2010) although analysis of surface observations suggest most current violations in California are entirely under local control (Pusede and Cohen, 2012).Quantitative assessment of potential changes in the photochemistry of the North Pacific has been limited by a paucity of observation-based constraints.Using observations made primarily along the western coast of North America, Parrish et al. (2004a) concluded not only that background O 3 concentrations had increased over the past 2 decades by T. H. Bertram et al.: On the export of reactive nitrogen from Asia concentrations observed on the west coast of North America tracked the increase in NO x emissions reported by Streets et al. (2003), and the net ozone production rate, or O 3 photochemical tendency ( (O 3 )) had slowed.Aircraft observations of O 3 and PAN concentrations and (O 3 ) made in the eastern Pacific (east of 135 • W, e.g., ITCT 2K2 and CITE-1C) have been compared with those made in the western Pacific (west of 155 • E, e.g., PEM-WEST B and TRACE-P).However, there have been few tropospheric observations (outside of the transit flights for PEM-WEST B and TRACE-P that primarily sampled at the aircraft ceiling altitudes) that have sampled the remote Northern Pacific (between 155 • E and 135 • W).Here we discuss measurements made during the INTEX-B campaign aboard the NASA DC8 aircraft during April and May of 2006.
Prior to discussing the INTEX-B reactive nitrogen observations, we first briefly review NO x oxidation mechanisms and highlight the aspects of the reaction mechanism that are of particular relevance in the North Pacific.During the daytime, NO x can be sequestered from the catalytic O 3 production cycle following the three-body reaction of NO 2 with the hydroxyl radical (OH) to produce HNO 3 (Reaction R1), reaction with peroxy radicals (RO 2 ) to form a peroxy nitrate (RO 2 NO 2 ) (Reaction R2), the most abundant being peroxy acetyl nitrate (or PAN) a derivative of acetaldehyde (Singh et al., 1985(Singh et al., , 1986)), or through the formation of alkyl or multi-functional nitrates (RONO 2 ) following the reaction of NO with RO 2 (Reaction R3) (Calvert and Madronich, 1987;Trainer et al., 1991).
NO 2 also reacts directly with O 3 , producing the nitrate radical (NO 3 ), which quickly reaches thermodynamic equilibrium with dinitrogen pentoxide (N 2 O 5 ) (Reactions R4-R5) (Noxon et al., 1978;Platt et al., 1980).NO 3 and N 2 O 5 concentrations are limited during the day due to strong visible light absorption and subsequent dissociation of NO 3 as well as rapid reaction with NO (which is significantly reduced at night).Together these reactions limit the steady-state lifetime of NO 3 to seconds in sunlight.
Additionally, nitrous acid (HONO), formed through the hydrolysis of NO 2 has been shown to be a significant component of NO y at night near the surface (Finlayson-Pitts et al., 2003).
The partitioning of reactive nitrogen between the various NO x oxidation products is of great importance as each reservoir (e.g., RO 2 NO 2 , RONO 2 , HNO 3 and N 2 O 5 ) has a drastically different lifetime in the atmosphere.PNs are largely insoluble (e.g., the Henry's Law Constant for PAN is 2-5 M atm −1 at 273 • K) (Sander, 1999), have low accommodation coefficients (γ = 0.0001) for heterogeneous uptake (Kirchner et al., 1990), and measurements have shown them to have small deposition velocities relative to other constituents of NO y (Farmer et al., 2006;Turnipseed et al., 2006;Wolfe et al., 2009) leading to long atmospheric lifetimes that are limited by photolysis in the upper troposphere and thermal dissociation at warm temperatures characteristic of the lower troposphere.The PAN lifetime against thermal decomposition increases from hours to days in the BL to months in the UT where lower temperatures drive the equilibrium shown in Reaction (R2) to the right, toward RO 2 NO 2 and photolysis limits the atmospheric lifetime to tens of days (Talukdar et al., 1995;Kirchner et al., 1999).As a result, PNs that are formed in colder regions can be transported long distances in the free troposphere and serve as a net source of NO x in warmer environments, far away from their source region (Lamarque et al., 1996;Moxim et al., 1996;Horowitz and Jacob, 1999;Heald et al., 2003;Hudman et al., 2004).
Nitric acid is largely soluble (e.g., the Henry's Law Constant for HNO 3 is 2-8 × 10 5 M atm −1 at 273 • K) (Sander, 1999), and has a significant accommodation coefficient for heterogeneous removal (Choi and Leu, 1998;Arora et al., 1999;Tolocka et al., 2002) and a large deposition velocity (Munger et al., 1996(Munger et al., , 1998)).Alkyl nitrates ( ANs) are removed following photolysis or reaction with OH or O 3 .In addition, hydroxyl-and multi-functional nitrates, which comprise a large fraction of ANs, especially in regions of strong biogenic influence (Day et al., 2003), are thought to be removed effectively via deposition and heterogeneous removal processes (Farmer et al., 2006).In the presence of high surface area loadings N 2 O 5 can be hydrolyzed forming HNO 3 or ClNO 2 on chloride containing particles (Bertram and Thornton, 2009) and can be accommodated to the surface or dust particles (Tang et al., 2010), where the lifetime of ClNO 2 to photolysis is typically less than three hours.These chemical lifetimes and the associated partitioning among different NO y species determine the extent to which NO x is present in the atmosphere far from its source and, thus, have the potential to affect the rate of ozone production (e.g., Hudman, et al. 2004) and nitrogen deposition (e.g., Munger et al., 1996;Holland et al., 2005), downwind of the source region.
Previous aircraft observations of NO y over the North Pacific indicate that NO y is primarily comprised of PAN and HNO 3 (Koike et al., 1996;Singh et al., 1996;Talbot et al., 2003;Roberts et al., 2004).These measurements were confined closely to either the Asian or North American continents, with only a select number of transpacific flights that sampled the remote Pacific.The scientific objective of earlier flight campaigns was the characterisation of Asian outflow plumes near the source region (e.g., PEM West A, PEM West B and TRACE-P, Hoell et al., 1996Hoell et al., , 1997;;Jacob et al., 2003).Transit flights from the United States to the sampling region proved instructive in assessing the extent of transport and transformation of the Asian plumes (Heald et al., 2003).Recently, the Intercontinental Transport and Chemical Transformation 2002 (ITCT 2K2) Experiment made observations of Asian plumes transported to North America during spring (Parrish et al., 2004b).
Observations of Fresh Asian Emissions: Measurements taken during the PEM West B campaign, indicated a major role for PAN (ranging between 20 and 70 % of NO y ) in Asian Outflow, and a NO y budget that was closed to within the uncertainties of the measurements that comprise it (Singh et al., 1998).In these studies, Singh et al. described a strong latitudinal dependence in PAN, with NO y sampled in the northern pacific having larger PAN fractions than in southern air.This was attributed to a strong gradient in temperature and, hence, PAN thermal dissociation.Talbot et al. (2003) showed from observations taken during the TRACE-P experiment that the PAN/NO y ratio exhibited a strong vertical structure, again associated with the temperature dependence in the PAN thermal decomposition rate (Talbot et al., 2003).
Observations of Aged Asian Emissions: The aforementioned experiments were instructive in describing the partitioning of NO y in the outflow region; however, they did little to address the question of long range transport of NO y across the Pacific.To better address these questions, ITCT 2K2 was conducted in the Spring of 2002 to probe the chemical composition of Asian plumes that reach the North American Continent (Nowak et al., 2004).During these studies Roberts et al. measured a PAN to NO y ratio, in transported Asian plumes, ranging between 0.5 and 0.7 in air-masses above 2 km and as large as 0.8 in episodic Asian plume events.The observed ratio dropped rapidly (to less than 0.2) at the surface, again consistent with the calculated PAN lifetime to thermal dissociation (Roberts et al., 2004).The PPN to PAN ratio in air transported from Asia, as measured on ITCT 2K2, was on average 0.12.
The dominance of PANs in aged Asian outflow was first discussed by Singh et al. (1986), and can be understood given the context for the export of pollution from Asia.The export of pollution from Asia is typically lofted via frontal lifting in warm conveyor belts or injected directly into the free troposphere via deep convection (Bey et al., 2001;Stohl, 2001;Liu et al., 2003).The exported pollution can then be transported across the Pacific on timescales of 5-10 days (Yienger et al., 2000;Jaffe et al., 2001), with peak outflow and transport occurring in the late spring (Yienger et al., 2000).Episodic events have been sampled in the continental United States (Jaffe et al., 1999(Jaffe et al., , 2001) ) and their effect on air quality in the US has been the subject of many recent studies using large scale models (e.g., Fiore et al., 2002).The export process plays a critical role in the partitioning of NO y at the point of injection into the free troposphere (Miyazaki et al., 2003).Due to wet removal of HNO 3 associated with both warm conveyor belt (WCB) lifting and deep convection and the short chemical lifetime of NO x in the lower troposphere, NO y is expected to be largely dominated by PAN in Asian outflow plumes sampled in the free troposphere.Since PAN is only thermally stable at altitudes greater than 2-6 km (depending on latitude), PAN is expected to be a large source of NO x , and consequentially an O 3 precursor, in subsiding air-masses over the remote pacific (e.g., Jaffe et al., 1999;Kotchenruther et al., 2001a, b;Heald et al., 2003;Hudman et al., 2004;Zhang et al., 2008;Walker et al., 2010).
In what follows, we present observations of NO 2 , PNs, ANs, and HNO 3 made aboard the NASA DC-8 during the Intercontinental Chemical Transport Experiment -Phase B [INTEX-B] in April-May 2006 in the North Pacific (20-60 • N, 175 • E-135 • W).We use these observations to: (1) provide constraint on the mass flux of reactive nitrogen through the North Pacific during spring, and (2) calculate both a quantitative instantaneous and a qualitative integrated net O 3 production rate (Parrish et al., 1992) that can be compared with previous observations made in both the eastern and western Pacific.

Intercontinental chemical transport experiment -Phase B [INTEX-B]
We use observations obtained during the INTEX-B campaign, conducted out of Honolulu, HI and Anchorage, AK during April and May of 2006 using the NASA DC-8 (20-60 • N, 175 • E-135 • W).Research flights were primarily conducted during daytime (88 % of the observations were made at SZA < 90 • ); the only nighttime flight was the transit between Honolulu and Anchorage.In the following analysis all observations were used.The principle objective of the INTEX-B campaign was to characterise the transport of Asian pollution, which is most frequent and rapid in spring, during periods of strong frontal activity (Yienger et al., 2000).Research flights were designed to sample pollution lofted from the Asian boundary layer (BL) by cold frontal activity and transported across the Pacific toward North America in the free troposphere.Observations highlighted in this study include in situ measurements of ozone, NO, NO 2 , total peroxy nitrates ( PNs), total alkyl and multi-functional nitrates ( ANs) and nitric acid (HNO 3 ) (Thornton et al., 2000;Day et al., 2002;Fairlie et al., 2007).Aircraft flight tracks are shown in Fig. 1a, where sampling legs north of 35  shown in Fig. 1b, highlighting an approximately 10 • C difference in temperature between the two regions from the surface through the mid-troposphere.The implications of the observed temperature difference on the reactive nitrogen budget are discussed in detail in Sect.3.2.

Instrument descriptions
Observations of NO 2 , PNs, ANs and HNO 3 were made using Thermal Dissociation -Laser Induced Fluorescence (TD-LIF) (Thornton et al., 2000;Day et al., 2002).Briefly, NO 2 fluorescence is detected following excitation of a specific jet-cooled rovibronic transition at 585 nm.The resulting fluorescence is collected by a PMT at 90 • to the laser axis, which is both optically and temporally filtered to remove laser scatter.The measured fluorescence is directly correlated to NO 2 following calibration to a NIST traceable NO 2 calibration standard (accuracy of ±5 %) (Bertram et al., 2005).The NO 2 calibration constant was determined and applied as a function of inlet pressure due to the nonlinear response of the system to pressure, a result of reduced jet-cooling of NO 2 at low ambient pressures.Higher order reactive nitrogen classes ( PNs, ANs and HNO 3 ) are detected by coupling a thermal dissociation inlet to the LIF sensor (Day et al., 2002).In this system, we heat the ambient air stream to the dissociation threshold for the class of NO y species of interest (200 • C for PNs, 350 • C for PNs + ANs, and 550 • C for PNs + ANs + HNO 3 ) and detect the NO 2 dissociation product using NO 2 LIF.As configured for INTEX-B N 2 O 5 would be detected in the PNs channel (Fuchs et al., 2012), and ClNO 2 would likely be detected in the ANs channel based on the analysis of Thaler et al. (2011), however, this would likely only impact the nighttime transit flight and morning flights conducted before complete ClNO 2 photolysis.With regard to detection of particulate nitrate, unpub-lished laboratory measurements have shown that TD-LIF is sensitive to volatile particulate nitrate.The resulting NO 2 , formed following the thermal dissociation of the precursor compound, would be detected in the corresponding TD-LIF channel (e.g., semi-volatile organic nitrate aerosols are detected as ANs).Specific discussion of alkylnitrates in the aerosol phase can be found in Rollins et al. (2012).
The TD-LIF instrument flown on INTEX-B consisted of two NO 2 LIF detection cells and four independent thermal dissociation ovens.The first detection cell alternated between sampling the ambient and 350 • C channel, and the second detection cell alternated between sampling the 200 • C and 550 • C channel.The sampling duty cycle was such that in each two minute sampling period, NO 2 , PNs, ANs, and HNO 3 were sampled for 90, 60, 15 and 15 s, respectively.The resulting system has an NO 2 detection limit of 8 pptv/10 s at 760 Torr (ground) and 25 pptv/10 s at 10 km at S/N = 2.The sensitivity of the TD-LIF technique toward PNs, ANs and HNO 3 is determined by the partitioning of the individual components of NO y as discussed in Day et al. (2002), Wooldridge et al. (2010) and Perring et al. (2010).
During the combined MILAGRO and INTEX-B flight campaigns, three wing-tip to wing-tip instrument comparison flights were conducted between the NASA DC-8 and NSF C-130 (Kleb et al., 2011).Direct comparisons for TD-LIF measurements of NO 2 , PNs, and HNO 3 were available.The details of the comparisons can be found in Kleb et al. (2011), alongside discussion of instrument precision, accuracy, and limit of detection.Comparison on chemiluminescence and LIF NO 2 measurements yielded a slope of 1.20 ± 0.01, intercept of −39.1 ± 1, and R 2 = 0.87, over a concentration range from the instruments limit of detection (LOD) to 796 pptv.Comparison on CIMS and TD-LIF PNs measurements yielded a slope of 1.35 ± 0.03, intercept of −83 ± 10 and R 2 = 0.94, over a concentration range from the instruments LOD to 2175 pptv.Comparison on TD-LIF and mist chamber (also on DC-8) HNO 3 measurements yielded a slope of 0.91 ± 0.01, intercept of −28 ± 4, and R 2 = 0.84, over a concentration range from the instruments LOD to 7530 pptv.For the comparisons above, the technique is listed as x-axis (C-130) and y-axis (DC-8).No comparison was available for the TD-LIF ANs measurement, however, Beaver et al. (2012), demonstrate agreement between independent AN measurements made using CIMS and TD-LIF ANs measurements at the surface (BEARPEX 2009), R 2 = 0.89, slope 0.91.
In what follows, we also use collocated aircraft measurements of nitric oxide (NO), O 3 , particulate nitrate (pNO − 3 ), OH and HO 2 , butane and ethane.Discussion of each measurement technique, along with its associated uncertainty can be found in Kleb et al. (2011).

Observations of reactive nitrogen during INTEX-B
The vertical distribution in the sum of the measured gasphase components of NO y (here defined as NO x + PNs + ANs + HNO 3 ) are shown in Fig. 2, alongside the vertical profile in ozone for observations made both North (Fig. 2a) and South (Fig. 2a) of 35 • N. Observations were separated into 1 km altitude bins, where the median in each bin is shown with a solid line, and the shaded region represents the interquartile range of the observations.The fraction of NO y carried by particulate NO − 3 is not shown in Fig. 2, due to sparse data coverage, and limited sampling during vertical profiling.The contribution of particulate NO − 3 , as measured using mist chamber -ion chromatography (Talbot et al., 1997;Dibb et al., 2003), to the NO y budget is shown in Fig. 3, and discussed below.The observed range in gasphase NO y mixing ratio (200-400 pptv) is broadly consistent with previous observations of NO y in the both the eastern and western Pacific (Koike et al., 1996;Talbot et al., 2003;Nowak et al., 2004).Further, the vertical distribution of O 3 and NO y are correlated in the troposphere reflecting their coupled source and sink mechanisms.
The partitioning of NO y , between NO x , PNs, ANs, HNO 3 , and aerosol nitrate is shown in Fig. 3 as a function of altitude.Here, the fraction of NO y in each altitude bin was calculated from the mean profile of each of the individual constituents.In the upper troposphere (above 10 km), NO y is largely composed of HNO 3 , due to transport and mixing of stratospheric air, rich in HNO 3 , to the upper troposphere and to the occasional sampling of purely stratospheric air in the northern pacific where the tropopause height (less than 10 km) is lower than the DC-8 aircraft ceiling (12.5 km).In the mid-troposphere (4-10 km) PNs comprise as much as 80 % of total NO y .In the lower troposphere (below 4 km), the PN fraction again decreases, a result of the strong temperature dependence in the PAN thermal dissociation rate constant, where the instantaneous lifetime of PAN (at 35 • N) goes from 20 days at 6 km to approximately two days at 4 km (Fig. 4).As a result, PAN thermal dissociation represents a significant source of NO x to the remote troposphere.However, the NO x lifetime with respect to reaction with OH is short, thus, NO x produced by PN decomposition is converted to HNO 3 on the time scale of days.The large fraction of HNO 3 in the lower troposphere is likely a result of the oxidation of NO x , formed from the thermal dissociation of PAN in subsiding air-masses.While we cannot rule out the possibility of direct HNO 3 transport from the Asian continent to the sampling region such transport would appear to occur exclusively at low altitudes.In the upper troposphere, in the presence of mineral dust aerosol, gas-phase HNO 3 readily reacts heterogeneously with CaCO 3 resulting in the sequestration of nitrate in the particle phase as shown in Fig. 3 (Jordan et al., 2003;McNaughton et al., 2009).

Latitudinal gradients in PNs
Latitudinal gradients in PAN have been observed previously in the lower troposphere (e.g., Singh et al., 1998;Roberts et al., 2004).This is due to the strong temperature dependence in the PAN thermal dissociation rate as shown in Fig. 4. At 2 km altitude, the instantaneous PAN lifetime increases from 0.2 days at 30 • N to over 4 days at 50 • N, at which point photolysis becomes the dominant loss process.The effect of PAN thermal dissociation is shown clearly in the vertical distribution of NO y partitioning as a function of latitude.As shown  in Fig. 3, PNs comprise over 40 % of NO y from the surface to the tropopause north of 35 • N. In contrast, southern samples (latitudes below 35 • N) show a strong shift from the NO y budget being controlled by PNs to being dominated by the sum of HNO 3 and particulate NO − 3 at low altitudes, consistent with the profile shape of the PAN thermal dissociation rate.

Intercontinental transport of reactive nitrogen
The extent to which the rapid increases in NO x emissions, observed over East Asia during the past decade (Richter et al., 2005;Zhang et al., 2007), impact ozone production rates in the remote North Pacific and set the western boundary condition for North American regional air quality models is dependent on the chemical transformations that occur post emission and the export efficiency of NO y from the source region to the free troposphere.NO x emissions estimates over East Asia have been calculated using both top-down (Richter et al., 2005;Zhang et al., 2008;Walker et al., 2010) and bottom-up (Streets et al., 2003;Zhang et al., 2009) techniques.Due to rapid increases in NO x emissions, we compare our observations with emission inventories that were calculated for the 2006 INTEX-B sampling period.Specifically, Zhang et al. (2009) estimated the total East Asian anthropogenic emissions of NO x to be 11.2Tg N yr −1 for 2006, where 6.3 Tg N yr −1 were attributed to anthropogenic emissions in China.Using a top-down approach, Zhang et al. (2008)  (6.9 Tg N yr −1 ) needed to be increased by nearly a factor of two to match 2006 OMI NO 2 observations.The fraction of reactive nitrogen emissions that leave the boundary layer is dependent on both the transport mechanism and the partitioning of reactive nitrogen between its soluble and insoluble forms.Measurements made during the TRACE-P field campaign in 2000 indicated that the time averaged export flux of NO y across the 130 • E meridonial plane between 30 • and 40 • N was 8 % between 0 and 2 km and 10 % between 2 and 7 km (Koike et al., 2003).The measurements of Koike et al. indicate that a total of 18 % of emitted NO y is transported out of the source region.
During INTEX-B, the DC-8 sampled between 20-60 • N in the region of 175 • E-135 • W (Fig. 1a).This sampling domain is characterised by zonal flow from Asia to North America with higher wind speeds recorded at higher altitudes and in the Northern section of the sampling domain (Hudman et al., 2004).To calculate the flux of nitrogen across the North Pacific between 20 • and 55 • N, we first construct curtain plots from the mean values of the westerly component of the observed wind speed and gas-phase NO y number density binned into 2 km altitude and 5 • latitude bins.For each of the sampling bins considered here, average westerly component of the wind speed and NO y number density were uncorrelated (R 2 < 0.2).We then calculate the flux as the product of these two observable properties (Fig. 5).The total daytime flux through this window for the INTEX-B sampling period is 0.007 ± 0.002 Tg N day −1 .The observed gas-phase NO y flux is 10 % NO x , 62 % total peroxyacyl nitrates, 5 % alkyl nitrates and approximately 23 % nitric acid.Particulate nitrate was not included in the above analysis due to sparse data coverage, based on the altitude and latitude range where the flux is largest and the percent of NO y in the form of pNO 3 , we expect that the contribution of pNO 3 (not accounted for here) to the daytime flux to be to be of order 10 %.
If we assume that all transpacific transport of Asian pollution crossing the 150 • W meridonial plane proceeds through the 20-55 • N sampling window, we can compare the magnitude of the observed gas-phase NO y flux with the NO x emissions inventories for East Asia to provide an estimate on the fraction of NO x emissions that pass through the North Pacific en route to North America.Using the Zhang et al. (2009) bottom up inventory for East Asia anthropogenic NO x emissions (11.2 Tg N yr −1 ), our observation represents an upper limit of 23 ± 6.5 % for the fraction of Asian NO x emissions that pass through the North Pacific (20-55 • N).
The sample calculation has three important caveats: (1) as discussed in Forster et al. ( 2004), a small fraction of transpacific transport is carried north of the sampling window discussed here.(2) the above calculation represents an upper limit as the observed NO y is also impacted by other sources such as biomass burning and stratospheric exchange, and (3) transpacific transport of Asian pollution is known to be episodic.Due to limited sample coverage, it was not possible to conduct this analysis with shorter time resolution.As discussed in Walker et al. (2010) and Zhang et al. (2008), a transport event was observed between 5-9 May 2006, which the DC-8 sampled on several occasions.

NO x production rates
To investigate the chemical and thermal repartitioning of NO y in the INTEX-B sampling region, we calculate the diurnally averaged, altitude dependent NO x production rates (molecules cm −3 s −1 ) from the thermal decomposition of PNs and the photolysis and reaction of hydroxyl radicals with HNO 3 using our ambient observations coupled with the aforementioned time-dependent chemical box-model.As shown in Fig. 6, the fraction of NO x produced from PNs is strongly altitude dependent, reflecting both the temperature dependence in the thermal decomposition rate and the concentration profile shown in Fig. 4. As a result, NO x production from HNO 3 becomes an increasing fraction of the total production rate with increasing altitude, accounting for nearly 30 % of in situ NO x production above 5 km.This further highlights the importance of accurate representation of HNO 3 in chemical transport models.

Photochemical ozone production
The production rate of O 3 in the troposphere is primarily controlled by the cycling of NO x in the presence of volatile organic carbon (VOC), oxidants and sunlight.In order to accurately model current O 3 abundances and assess the impact of future control strategies, it is critical to attain a mechanistic understanding of the chemical processes that drive O 3 production in the troposphere.In what follows, we investigate O 3 photochemical production in the North Pacific using two separate approaches: (1) quantitative calculation of the net instantaneous ozone production rate ( O 3 ), and (2) qualitative assessment of the integrated O 3 using observations of O 3 and ratios of hydrocarbons (Parrish et al., 1992).We calculate the instantaneous net O 3 production rate ( O 3 ) directly from measurements of NO (chemiluminescence), NO 2 (LIF; Thornton et al., 2000), OH and HO 2 (LIF; Faloona et al., 2004), H 2 O (Diode laser hygrometer; Diskin et al., 2002), and O 3 (chemiluminescence; Fairlie et al., 2007), and calculations of O( 1 D) and RO 2 made using a photochemical box model constrained by observations of C1-C5 straight chain hydrocarbons using Eqs.(1-3) (Thornton et al., 2002).
The uncertainty in O 3 can be large, especially at low NO x concentrations ([NO x ] < 100 pptv) observed over the North Pacific.Specifically, systematic error in the measurement of NO and NO 2 have the potential to bias the calculation significantly.Comparison of the observed NO 2 /NO ratio with that calculated from steady-state (Eq.4) is a useful test of the validity of the measurements at low NO x concentrations.
Figure 7b shows the ratio of the observed NO/NO 2 ratio to that calculated via PSS from measurements of O 3 , HO 2 and J (NO 2 ) and model calculations of RO 2 as a function of measured NO x .The results suggest that systematic error in the measurement of NO or NO 2 may be important at NO x concentrations below 20 ppt, where the ratio of the observations to model calculations is 0.46.This is likely either a systematic positive bias in NO or negative bias in NO 2 , both of which would result in a systematic over estimate of O 3 for NO x < 20 pptv (∼ 20 % of the data). NO The dependence of O 3 on NO x is shown in Fig. 7c, where O 3 is calculated from atmospheric measurements of the components defined in Eqs.(1-3), for all INTEX-B samples where the PAN lifetime was less than ten days.The frequency distribution of NO x mixing ratio is shown in Fig. 7a.While this analysis averages over a wide variety of chemical environments and VOC reactivity, it is instructive in describing the mean behaviour of the lower troposphere over the remote pacific and its sensitivity to increasing NO x loadings.
The altitude dependence of both the instantaneous and diel averaged O 3 are shown in Table 1, alongside previous model determinations of O 3 in both the eastern and western Pacific.During INTEX-B, 24 h averaged mean O 3 were calculated to be −0.12 ± 0.16, −0.05 ± 0.14, −0.02 ± 0.10 for 0-2 km, 2-4 km, and 4-6 km, respectively.Instantaneous mean O 3 were calculated to be −1.13 ± 1.53, −0.47 ± 1.29, −0.14 ± 1.04 for 0-2 km, 2-4 km, and 4-6 km, respectively.The stated uncertainty in O 3 is 1σ , and reflects a combination of atmospheric variability and instrument uncertainty.The increase in O 3 with altitude is driven primarily by a decrease in [H 2 O] resulting in a corresponding decrease in L(O 3 ).The diel averaged calculations indicate that on average, the photochemical environment below 6 km in the North Pacific is one of net O 3 destruction.
The average instantaneous O 3 is statistically lower for INTEX-B (2006) than the corresponding measurements of O 3 made during CITE-1C (1984) in the eastern Pacific,  potentially reflecting a shift in hydrocarbon concentrations and/or oxidant loadings (Chameides et al., 1989).However, more likely the differences in O 3 reflect the measured factor of two enhancement in NO between CITE-1C and INTEX-B (Table 1).The altitude dependent, diel averaged O 3 compares well with that determined during PHOBEA (−0.83, −0.34 and −0.24), at comparable NO concentrations (Kotchenruther et al., 2001c).For comparison, O 3 calculated for the PEM-WEST B and TRACE-P campaigns in the western Pacific reflects a net O 3 production regime (Crawford et al., 1997;Davis et al., 2003).This is presumed to be driven by statistically higher concentrations of NO x (as much as a factor of three).To investigate the dependence of O 3 production on NO x concentration, we calculate O 3 as a function of NO x where τ PAN is less than 10 days.As shown in Fig. 7c, O 3 increases linearly with increasing NO x , exhibiting NO x -limited behaviour over the entire sampling regime.In this low NO x regime, the crossover point between net O 3 destruction and net O 3 production, has been identified at around 60 pptv, consistent with the early work of Fishman et al. (1979) as well as the analyses presented above.This key diagnostic is critical for assessing how future increases in NO x emissions will affect global O 3 abundances and illustrates the sensitivity of the global O 3 budget to increasing NO x .As a result, quantifying the magnitude and spatiotemporal distribution of NO x and its transport and chemical evolution is crucial for modelling of tropospheric O 3 .
One limitation of the aforementioned model calculation of O 3 is that it provides a measure of the instantaneous ozone production rate under a specific set of conditions, making it difficult to determine the total ozone produced since emission without detailed knowledge of the time history of NO x .Here, we follow the approach of Parrish et al. (1992Parrish et al. ( , 2004a)), and examine the correlation between O 3 and the natural log of the propane to ethane ratio.As shown in Fig. 8, the positive slope indicates that the photochemical environment of the north Pacific is still one of net O 3 destruction (for air masses sampled between 0-1 and 1-2 km).The magnitude of the slope has been used previously to provide a qualitative assessment of changes in the photochemical environment over the north Pacific (Parrish et al., 2004a).Specifically, the less negative slope (0.19 ± 0.06) measured during ITCT 2K2 (2002) as compared with that determined in 1985 at Pt. Arena, CA (0.86 ± 0.1) was interpreted as a decrease in the efficiency of photochemical O 3 destruction (Table 2).Interpretation of the observations within the context of the Parrish et al. (2004a) analysis would indicate that the photochemical environment became more efficient at O 3 destruction between 2002 and 2006.This interpretation would be inconsistent with the measured trend in NO x emissions in eastern Asia between 2002and 2006(Zhang et al., 2009).Another interpretation would be that the difference between the slopes calculated for the ITCT 2K2 and INTEX-B campaigns reflects a difference in the altitude of the air mass trajectory www.atmos-chem-phys.net/13/4617/2013/Atmos.Chem.Phys., 13, 4617-4630, 2013   from the source region to the sampling region and or the degree to which background or stratospheric air is entrained into the sampled air mass.

Conclusions
The observations presented here provide experimental measures of the partitioning of reactive nitrogen in the remote Pacific and provide an opportunity to test model representations of the transport and chemical evolution of NO y from the Asian continent.In agreement with previous studies, we find a dominant role for PNs throughout the Pacific region, displaying a strong latitudinal dependence, consistent with the known temperature dependence in the thermal dissociation of PAN.Using simultaneous observations of NO x , PNs, ANs, HNO 3 and average wind speed, we calculate the flux of reactive nitrogen through the meridional plane of 150 • W (between 20 • and 55 • N) to be 0.007 ± 0.002 Tg N day −1 , which provides an upper limit of 23 ± 6.5 % on the transport efficiency of NO y from East Asia.Observations of NO x , and HO x are used to constrain a 0-D photochemical box  model for the calculation of net photochemical ozone production or tendency ( O 3 ) as a function of aircraft altitude and NO x concentrations.The model analysis indicates that the photochemical environment of the lower troposphere (altitude < 6 km) over the north Pacific is one of net O 3 destruction.Qualitative indicators of integrated net O 3 production derived from simultaneous measurements of O 3 and light alkanes (Parrish et al., 1992), also indicate that the north Pacific is on average net O 3 destruction, however, comparison with previous analyses suggests that interpretation of the trend in the correlation between O 3 and the ratio of propane to ethane requires careful assessment of air mass trajectory and entrainment of background and/or stratospheric air.
Figure 1: left panel INTEX-B flight tracks made between 17 April 2006 and 15 May 2006 over the Northern Pacific Ocean.Sampling legs north of 35°N are shown in black, while legs south of 35°N are shown in grey.right panel Observed mean temperature within 1 km altitude bins between 0-12 km divided into Northern (black) and Southern (grey) sampling bins.The dashed lines represent one standard deviation of the mean.

Fig. 1 .
Fig. 1. (left panel) INTEX-B flight tracks made between 17 April 2006 and 15 May 2006 over the Northern Pacific Ocean.Sampling legs north of 35 • N are shown in black, while legs south of 35 • N are shown in grey.(right panel) Observed mean temperature within 1 km altitude bins between 0-12 km divided into Northern (black) and Southern (grey) sampling bins.The dashed lines represent one standard deviation of the mean.

Figure 2 :
Figure 2: Median vertical profile in ozone (black) and gas-phase NOy (grey) (NOy ≡ NOx + PNs + ANs + HNO3) as observed during the INTEX-B field campaign over the North Pacific during the Spring of 2006 (April-May), North of 35ºN left panel and South of 35ºN right panel.The solid line depicts the median value in 1 km altitude bins and the shaded regions represent the interquartile range.

Fig. 2 .
Fig. 2. Median vertical profile in ozone (black) and gas-phase NO y (grey) (NO y ≡ NO x + PNs + ANs + HNO 3 ) as observed during the INTEX-B field campaign over the North Pacific during the Spring of 2006 (April-May), North of 35 • N (left panel) and South of 35 • N (right panel).The solid line depicts the median value in 1 km altitude bins and the shaded regions represent the interquartile range.

Figure 3 :Fig. 3 .
Fig. 3. Vertical distribution of the partitioning of reactive nitrogen (NO y ) between NO x (brown), PNs (blue), ANs (red), HNO 3 (green) and particulate nitrate (grey) as observed during the INTEX-B field campaign over the North Pacific during the Spring of 2006 (April-May), North of 35 • N (A) and South of 35 • N (B).The fraction of NO y in each altitude bin was calculated from the median profile in each of the individual constituents.

Figure 4 :
Figure 4: Median vertical profile in ΣPNs (grey, -o-) and PAN instantaneous lifetime (τPAN) with respect to thermal dissociation, photolysis, and reaction with OH (black, -□-) as observed during the INTEX-B field campaign over the North Pacific during the Spring of 2006 (April-May), North of 35ºN (left panel) and South of 35ºN (right panel).The solid line depicts the median value in 1 km altitude bins and the shaded regions represent the interquartile range.

Fig. 4 .
Fig. 4. Median vertical profile in PNs (grey, -o-) and PAN instantaneous lifetime (τ PAN ) with respect to thermal dissociation, photolysis, and reaction with OH (black, --) as observed during the INTEX-B field campaign over the North Pacific during the Spring of 2006 (April-May), North of 35 • N (left panel) and South of 35 • N (right panel).The solid line depicts the median value in 1 km altitude bins and the shaded regions represent the interquartile range.

Figure 5 :Fig. 5 .
Fig. 5. Daytime flux of reactive nitrogen [gN m 2 day −1 ] calculated using all available observations of gas-phase NO y and wind-speed gridded into 5 latitude × 2 km altitude bins.Observations were filtered to remove strong stratospheric influence (O 3 /CO > 1.25).
Forster et al. (2004) calculated a 15 yr climatology for the average transport of Asian CO for the months of March, April, and May.As shown in Fig. 2b of Forster et al. (2004), the Asian CO tracer is concentrated between 20-50 • N (but extends all the way to 80 • N) and 3-12 km at 125 • W. For the INTEX-B sampling period, Walker et al. (2010) and Zhang et al. (2008), using satellite observations of CO and O 3 in combination with kinematic back trajectory analysis, show that transpacific transport of Asian pollution during this INTEX-B sampling period was found to be North of 20 • N at 150 • W.

Figure 6 :Fig. 6 .
Fig. 6.Diurnally averaged NO x production rates (molecules cm −3 s −1 ) from nitric acid (left) and PNs (centre) as a function of altitude.The fraction of NO x production from each channel is shown in the right panel.Observations were filtered to remove strong stratospheric influence (O 3 /CO > 1.25) and are for the northern pacific (Latitude > 35 • N).
(A) Observed frequency distribution of NO x below 6 km.(B) Ratio of the 597 NO 2 /NO, compared with that calculated from photo-stationary steady state (Eq.598 Calculated instantaneous net ozone production rate as a function of observ 599 concentrations.Observations were filtered to strong remove stratospheric 600 (O 3 /CO>1.25)and are for the northern pacific sampling region (Latitude > 35ºN).601

Fig. 7 .
Fig. 7. (A) Observed frequency distribution of NO x below 6 km.(B) Ratio of the observed NO 2 /NO, compared with that calculated from photo-stationary steady state (Eq.4).(C) Calculated instantaneous net ozone production rate as a function of observed NO x concentrations.Observations were filtered to strongly remove stratospheric influence (O 3 /CO > 1.25) and are for the northern pacific sampling region (Latitude > 35 • N).
study * Ratio reported is the slope of the linear least squares fit to the correlation between the natural log in O 3 with the natural log of the propane-to-ethane ratio (as shown in Fig.6for the INTEX-B data).The error represents the 95 % confidence limit.

Figure 8 :
Correlation of the measured O3 concentrations with the natural log of the propane to 604 ethane ratio in the marine boundary layer during INTEX-B.Linear least squares fits determined 605 using data from ITCT 2K2 and Pt.Arena, CA are included for comparison Parrish et al. [2004].606 607 608 609

Fig. 8 .
Fig. 8. Correlation of the measured O 3 concentrations with the natural log of the propane to ethane ratio in the marine boundary layer during INTEX-B.Linear least squares fits determined using data from ITCT 2K2 and Pt.Arena, CA are included for comparison Parrish et al. (2004).

Table 1 .
Model calculated net ozone production rates ( O 3 ), constrained by aircraft observations made in the troposphere over the Eastern, Central and Western North Pacific.NO x concentrations for PHOBEA include NO, NO 2 , NO 3 , HNO 4 , HONO, 2 N 2 O 5 .
1 NR = Not Reported. 2its for 24 h average O 3 are ppbv d −1 . 3its for instantaneous O 3 are ppbv h −1 . 4 measurement for CITE-2 is taken from the Wallops NO chemiluminescence instrument.5 6[NO] for PHOBEA is estimated as NO x -NO 2 -HNO 4 .7NOand O 3 measurements for PEM-WEST B and TRACE-P are binned medians.8Reporteduncertainty for INTEX-B measurements is 1σ .

Table 2 .
Qualitative determinations of integrated net ozone production rates as determined from correlations between ozone and the ratio of propane to ethane.