The Pacific Exploration of Asian Continental Emission Phase A (PEACE-A) aircraft measurement campaign was conducted over the western Pacific in January 2002. Correlations of carbon monoxide (CO) with carbon dioxide (CO2) and back trajectories are used to identify plumes strongly affected by Asian continental emissions. ΔCO/ΔCO2 ratios (i.e., linear regression slopes of CO-CO2) in the plumes generally fall within the variability range of the CO/CO2 emission ratios estimated from an emission inventory for east Asia, demonstrating the consistency between the aircraft measurements and the emission characterization. Removal rates of reactive nitrogen (NOx and NOy) for the study region (altitude <4 km, 124°–140°E, 25°–45°N) are estimated using the correlation with CO2, the photochemical age of the plumes, and the NOx/CO2 emission ratio derived from the emission inventory. The plume age is estimated from the rates of hydrocarbon decay and hydroxyl radical (OH) concentration calculated using a constrained photochemical box model. The average lifetime of NOx is estimated to be 1.2 ± 0.4 days. Possible processes controlling the NOx lifetime are discussed in conjunction with results from earlier studies. The average lifetime of NOy is estimated to be 1.7 ± 0.5 days, which is comparable to the NOy lifetime of 1.7–1.8 days that has been previously reported for outflow from the United States. This similarity suggests the importance of chemical processing near the source regions in determining the NOy abundance.
 Fossil fuel combustion and biomass burning in east Asia are large sources of trace gases such as carbon monoxide (CO), carbon dioxide (CO2), nitrogen oxides (NOx), sulfur dioxide (SO2), and volatile organic compounds (VOCs) [Streets et al., 2003]. These emissions can significantly affect the distribution of ozone (O3) over the western Pacific, and even the United States by long-range intercontinental transport [Jaffe et al., 2003]. To assess the impact of Asian emissions on O3 and its precursors on regional/hemispheric scales, it is important to evaluate the loss rate of reactive nitrogen species (NOx and NOy) during transport processes, because the concentrations of these species often control the rate of photochemical O3 production.
 A number of aircraft/ground-based measurements have been conducted in east Asia to investigate the chemical characteristics of Asian outflow. The NASA Pacific Exploratory Mission-West campaigns (PEM-West A: September–October 1991; PEM-West B: February–March 1994) were the first systematic studies focusing on the impact of Asian outflow over the western Pacific [Hoell et al., 1996, 1997]. The NASA Transport and Chemical Evolution over the Pacific (TRACE-P) mission was conducted in the same region in February–April 2001, deploying improved instrumentation and using three-dimensional chemical transport models (3D-CTMs) [Jacob et al., 2003].
 The Pacific Exploration of Asian Continental Emission (PEACE) campaign of the Japan Aerospace Exploration Agency (JAXA) was designed to investigate the chemical characteristics of Asian outflow over the western Pacific in winter and spring. These seasons have not been covered by PEM-West-B and TRACE-P [Kondo et al., 2004b]. The PEACE campaign consists of two phases (phase A: January 2002; phase B: April–May 2002). In this analysis we focus on the data obtained during the PEACE-A campaign. The major purpose of this paper is to estimate the removal rates of NOx and NOy. A brief description of the aircraft measurements during PEACE-A is given in section 2. An emission inventory for east Asia in January 2002 is described in section 3. We use CO and CO2 as tracers of Asian outflow, and the method of our analysis is described in section 4. A comparison of the observations with the emission inventory and estimates of the removal rates of NOx and NOy are presented in section 5.
2. Aircraft Measurements
2.1. Airborne Instruments
 In situ measurements of gases and aerosols were made onboard the Gulfstream-II (G-II) aircraft of the Mitsubishi Diamond Air Service (DAS). Figure 1 shows the flight tracks of the G-II aircraft during the PEACE-A campaign. Thirteen flights were conducted, mainly over the Sea of Japan (35°–45°N) and the East China Sea (20°–35°N) during 6–23 January 2002. The instruments used for the PEACE-A campaign were similar to those used for the Biomass Burning and Lightning Experiment Phase B (BIBLE-B) campaign [Kondo et al., 2002]. A brief description of the key measurements in this analysis is given below. CO mixing ratios were measured using a vacuum ultraviolet (VUV) resonance fluorescence technique with a time resolution of 1 s [Takegawa et al., 2001]. CO2 was measured using a nondispersive infrared absorption (NDIR) technique with a time resolution of 1 s (LI-COR) [Machida et al., 2002]. Measurements of nonmethane hydrocarbons (NMHCs) were made using a whole air sampling technique followed by laboratory analysis [Blake et al., 2003]. The integration time of the whole air sampling was typically 40–60 s at lower altitudes (<4 km). O3 was measured using a dual-beam UV photometer [Kita et al., 2002].
 Nitric oxide (NO) and total reactive nitrogen (NOy) were measured using a NO-O3 chemiluminescence detector with a time resolution of 1 s [Kondo et al., 1997b, 2003, 2004a]. Photostationary NO2 mixing ratios were calculated using a photochemical box model. The box model used in this analysis is the same as that used by Kondo et al. [2004b]. The reactions of NO with O3 and peroxy radicals (HO2, CH3O2, C2H5O2, …) were taken into consideration. The peroxy radical concentrations were calculated assuming photochemical steady state. The sum of observed NO and calculated NO2 is given as NOx. The precision and accuracy of the 1-s NOx data were estimated to be 13 pptv and 20%, respectively.
 NOy compounds were catalytically converted to NO on the surface of a gold tube heated at 300°C. The effect of particulate nitrate (NO3−) on the NOy measurements is estimated here. During PEACE-A, two converter systems were placed in the aircraft pod and were connected to two independent inlets: one directed forward and the other rearward. The forward and rearward NOy systems used during PEACE-A are similar to those used during the SAGE III Ozone Loss and Validation Experiment (SOLVE) campaign [Kondo et al., 2003; Irie et al., 2004]. NOy measurements made using the forward facing inlet are referred to as NOy(f) and those using the rearward facing inlet as NOy(r). The particle collection efficiency for the rearward facing inlet during PEACE-A was estimated to be nearly 100% for particle diameters (Dp) < 0.4 μm and almost zero for Dp > 3 μm, on the basis of simple aerodynamic calculations [Hinds, 1998]. Although the conversion efficiency for nitrate aerosols was not fully evaluated in the laboratory, our recent TRACE-P analysis has demonstrated that the NOy(r) represents the sum of gas-phase NOy and submicron nonrefractory NO3− aerosols [Kondo et al., 2004a]. Note that nonrefractory means volatile at the temperature of the NOy converter (300°C) (e.g., NH4NO3). The amount of nonrefractory NO3− on larger particles (>0.4 μm) that are not efficiently detected by the rearward NOy system can be estimated using the difference between the forward and rearward NOy.
where EF (>10 for Dp > 0.4 μm) represents the enhancement factor for the forward facing inlet due to subisokinetic sampling. Strictly speaking, equation (1) gives an upper limit of NO3− on larger particles because NOy(f) also includes NO3− on smaller particles with an enhancement factor larger than unity. Figure 2 shows the correlation of NOy(f) and NOy(r) for the entire data set used in this analysis. Ninety-two percent of the data points lie below the 2:1 correspondence line and >99% of the data points lie below the 5:1 line. Given NOy(f) < 2NOy(r) and EF > 10, the ratio of nonrefractory NO3− to NOy(r) is estimated to be <10%. Therefore the amount of nonrefractory NO3− on larger particles was generally small as compared to NOy(r) during PEACE-A. Hereafter NOy(r) is expressed simply as NOy and is used for the present analysis. Thus the term “removal of NOy” in this analysis represents either dry/wet deposition of NOy compounds or conversion of NOy compounds to refractory NO3− aerosols (e.g., Ca(NO3)2, NaNO3). Because such conversion is generally an irreversible process, it can be regarded as an ultimate sink of NOy in the atmosphere.
2.2. Vertical Profiles
Figure 3 shows an example of vertical profiles of CO, CO2, NOx, NOy, O3, and relative humidity (RH) observed over the Sea of Japan (135°E, 36°N) on 11 January 2002 (flight 04). Large enhancements of CO mixing ratios (200–400 parts per billion by volume (ppbv)) were observed below 2.5 km. The CO mixing ratios between 2.5 and 4 km were lower (i.e., 100–200 ppbv), and those above 4 km were fairly constant at ∼100 ppbv. Correspondingly, the RH profile showed a slight decrease at 2.5 km and a sharp gradient at 4 km. The former is interpreted as the top of the convective boundary layer where the air masses were well mixed, and the latter as the top of the planetary boundary layer where the air masses had been influenced by surface emissions but not well mixed. The CO2 and NOy profiles exhibit almost the same vertical structures as the CO profile. There is no clear structure in the O3 profile. A detailed analysis of the O3 production and loss rates during PEACE-A is presented by Kondo et al. [2004b].
 Profiles similar to Figure 3 were frequently observed during PEACE-A, and we rarely found significant enhancements of primary emissions such as CO, CO2, and NOy above 4 km. Reanalysis data from the National Centers for Environmental Prediction (NCEP) (http://www.ncep.noaa.gov/) indicate that airflow patterns in the boundary layer over east Asia were strongly influenced by the Siberian High in January 2002. It is likely that the vertical transport of trace gases was limited during PEACE-A because of the strong influence of the Siberian High during this period.
3. Emission Inventory
3.1. Distribution of CO/CO2 Emission Ratio
 A comprehensive emission inventory of gaseous and particulate emissions in east Asia for the year 2000 has been developed by Streets et al.  and evaluated by several investigators using 3-D CTMs [e.g., Palmer et al., 2003; Woo et al., 2003]. Streets et al.  showed that the emissions of CO, CO2, and other species from the Asian continent tend to have an annual maximum in January mainly because of an increase in domestic heating. An emission inventory for January 2002 was produced for this study taking into account the seasonal trends of gaseous and particulate emissions [Streets et al., 2003] and assuming that the annually averaged anthropogenic emissions in 2002 were the same as those in 2000. Figure 4 shows a 1° × 1° grid distribution of CO/CO2 emission ratios for January 2002. Although the emission inventory includes both anthropogenic and biomass burning sources, major primary emissions such as CO, CO2, and NOx were likely dominated by anthropogenic sources because the biomass burning activities in east Asia were very weak in January 2002 (e.g., fire images from the Along Track Scanning Radiometer (ATSR) (http://shark1.esrin.esa.it/ionia/FIRE/AF/ATSR/)).
 Emission ratios for CO/CO2 have been used as a good proxy for evaluating the combustion efficiency of emission sources, especially for biomass burning studies [e.g., Yokelson et al., 2003]. Lower CO/CO2 emission ratios correspond to higher combustion efficiency. Because combustion efficiency strongly depends on fuel type (petroleum, coal, etc.) and devices (engine, furnace, etc.), CO/CO2 emission ratios can be used for diagnosing the types of emission sources. For example, low CO/CO2 emission ratios (<20 ppbv/ppmv) typically originate from combustion of well-processed gas/liquid fuels (vehicular engines, natural gas stoves, etc.) [e.g., Zhang et al., 2000]. By contrast, high CO/CO2 emission ratios (30–100 ppbv/ppmv) typically come from combustion of poorly processed solid fuels (coal/biofuel stoves, biomass burning, etc.) [e.g., Yokelson et al., 2003; Zhang et al., 2000]. Fossil fuel power plants, ship engines, and aircraft engines generally have extremely high combustion efficiencies, producing much less CO relative to CO2 as compared to other sources [e.g., Nicks et al., 2003]. Figure 4 clearly illustrates that the emissions from Japan and South Korea are dominated by high-combustion-efficiency sources, while those from China are dominated by relatively low-combustion-efficiency sources.
3.2. Definition of Source Regions
 We have classified the source regions into four categories as follows: Japan (mainly Nagoya) (136°–138°E, 34°–36°N), Korea (mainly Pusan) (126°–130°E, 34°–36°N), Northern China (110°–125°E, 35°–44°N), and Central China (105°–123°E, 26°–35°N) (Figure 4). The emission ratios of Y to X (X, Y = CO, CO2, NOx, etc.) for the individual source regions were calculated as follows and are summarized in Table 1.
EY-X, EY–XMax, EY–XMin are the average, maximum, and minimum values of the Y to X emission ratio, respectively. FXi and FYi are the emission rates (mol month−1 grid−1) of species X and Y for the ith grid box in the individual source region, respectively. Only data with FCOi > 0.2 Gmol month−1 grid−1 were used to determine the values of EY–XMax and EY–XMin because the variations of FYi/FXi for small emission sources may not be representative. It should be noted that the uncertainties in the emission rates are not taken into account in determining the maximum and minimum values. The uncertainties in the emission rates are as follows: CO (156% for China, 34% for Japan, and 84% for other east Asia), CO2 (16% for China, 7% for Japan, and 13% for other east Asia), NOx (23% for China, 19% for Japan, and 24% for other east Asia) [Streets et al., 2003].
Table 1. Emission Ratios of Trace Gases for Individual Source Regionsa
Average of emission ratios derived from the emission inventory for January 2002. The values in parentheses represent the minimum and maximum values of the emission ratios.
 As described in section 3.1, emissions from Japan (Nagoya) and Korea (Pusan) are dominated by high-combustion-efficiency sources (low CO/CO2), while those from Northern/Central China are dominated by low-combustion-efficiency sources (high CO/CO2). Details about the interpretation of NOx/CO and NOx/CO2 emission ratios are given in section 5.1.2. Note that the ethene (C2H4) to ethyne (C2H2) emission ratios are specifically listed in Table 1 because these species are used to estimate the photochemical age of air masses (section 5.2).
4. Method of Analysis
4.1. Definition of Plumes
 The major purpose of this paper is to estimate the removal rates of NOx and NOy in Asian outflow during PEACE-A. To achieve this goal, we first need to identify air masses that were strongly affected by Asian continental emissions. The Y to X enhancement ratio relative to regional background air (ΔY/ΔX ratio), defined by the linear regression slope of the Y-X correlation, is used for the air mass identification. The uncertainty in the slope is determined using 68% confidence interval. The regional background air represents relatively clean air that dilutes polluted air masses emitted from anthropogenic sources. Ten-day back trajectories were calculated for the sampled air masses using meteorological data with a horizontal resolution of 1° in longitude and 1° in latitude. The data were provided by the European Centre for Medium-Range Weather Forecasts (ECMWF) (http://www.ecmwf.int). In this analysis, “plume” is defined as air masses that meet the following three criteria (criteria 1–3). (1) The air masses were sampled below 4 km, and the back trajectories for the air masses had passed over one of the source regions defined in section 3.2. (2) The enhancement of CO (ΔCO) in the air masses is larger than 50 ppbv, and that of CO2 (ΔCO2) is larger than 5 parts per million by volume (ppmv). (3) The square of the linear correlation coefficient (r2) for the CO-CO2 correlation is larger than 0.7.
 The threshold values (ΔCO > 50 ppbv, ΔCO2 > 5 ppmv, r2 > 0.7) were determined as follows. We assume that air masses observed at 5–7 km (i.e., just above the boundary layer) represent the regional background air. The median mixing ratio (± central 68% range) at 5–7 km was 111 (+15/−15) ppbv for CO and 373 (+0.5/0.6) ppmv for CO2. In the case of CO, the threshold value of 50 ppbv was determined so that ΔCO values in the plumes are more than 3 times larger than the variability in the regional background air (i.e., 15 ppbv). For CO2, the threshold value of 5 ppmv was determined so that ΔCO2 values in the plumes are more than 10 times larger than the variability (0.5–0.6 ppmv). The definition of CO2 threshold is more conservative than that for CO because vertical gradients of CO2 in the boundary layer may be larger than 0.5–0.6 ppmv because of biogenic sources/sinks at the surface. For example, relatively steep gradients of CO2 (1–2 ppmv between 0 and 3 km) were observed over remote extratropical ocean in the western Pacific region [Machida et al., 2002; Vay et al., 2003]. Using criteria 1 and 2, a total of 21 plumes were identified. The r2 values were greater than 0.7 for most of the plumes (18 plumes), while the r2 values were less than 0.6 for the other three plumes. The threshold value of r2 = 0.7 was determined so as to exclude those statistically less significant plumes.
 The identified plumes are summarized in Table 2. The plume ID consists of a group name (e.g., Japan-1, Korea-1, N-China-1, and C-China-1), a flight number (F01, F02, etc.), and a leg number (L01, L02, etc.). First, the duration of each flight was divided into 5–14 flight legs on the basis of sampling altitudes and back trajectories. Second, the source region for each leg was investigated using the trajectories, and the group name was determined corresponding to Table 1. Note that different flight numbers and legs with similar sampling locations on the same day belong to the same group (e.g., both F02L08 and F02L12 belong to Korea-1) but are treated as different plumes. A comparison of the observed ΔCO/ΔCO2 ratios to the CO/CO2 emission ratios estimated from the emission inventory is discussed in section 5.1.1.
Each plume ID consists of a group name, flight number, and leg number. The durations of individual flights (F01, F02, etc.) were divided into 5–14 flight legs (L01, L02, etc.) on the basis of back trajectories. The name of each group represents the source region (defined in Table 1) estimated from the trajectories. Different flight legs with similar sampling locations on the same day were categorized as the same group.
ΔCO/ΔCO2 ratios were determined from linear regression analysis.
Northern China Plumes
Central China Plumes
 The majority of the plumes were sampled below 3 km because the vertical transport was weak during PEACE-A (section 2.2). The infrequent vertical mixing of air masses during the PEACE-A period should have caused the accumulation of pollutants at lower altitudes, leading to the formation of distinct plumes. Indeed, we rarely found such distinct plumes in the same altitude regime during the PEACE-B period (April–May 2002), when the upward lifting by convection was very active over central China [Oshima et al., 2004]. Therefore the plume analysis presented in this study is not applicable to PEACE-B. The differences in photochemical and transport processes between PEACE-A and PEACE-B conditions are discussed by Kondo et al. [2004b].
4.2. Interpretation of ΔY/ΔX Ratio
 Trajectory analysis shows that the transport time from source regions to sampling points was generally less than 4 days. Therefore chemical species with photochemical lifetimes longer than 1 month can be regarded as long-lived, and those with photochemical lifetimes shorter than ∼4 days as short-lived. Table 3 summarizes the photochemical lifetimes of CO and NMHCs against reactions with hydroxyl radical (OH) under the conditions experienced during PEACE-A. The diurnally averaged OH concentrations were calculated using a constrained photochemical box model. The data obtained below 3 km were used to calculate the median OH concentration because the majority of the plumes were transported below 3 km prior to the measurements. The median value (± central 68% range) of OH concentration was 5 (+6/−3) × 105 cm−3. The kinetic parameters listed by Atkinson et al.  and Calvert et al.  were used for calculating the rates of OH reactions. In Table 3, CO, ethane (C2H6), and C2H2 correspond to long-lived species, and C2H4 and propene (C3H6) correspond to short-lived species.
Table 3. Photochemical Lifetime of CO and NMHCs Under the Conditions Experienced During PEACE-A
Median value of the diurnally averaged OH concentration of 5 × 105 cm−3 (<3 km, 25°–45°N) was used for the calculation.
 If X and Y are both long-lived, the ΔY/ΔX ratio should be equal to the EY-X (or between EY-XMax and EY-XMin) of the source region estimated from the trajectories. If X is a long-lived species and Y is a short-lived species, the ΔY/ΔX ratio should be systematically smaller than EY-X because of the removal (including both photochemical loss and deposition) of species Y. In this case, we assume that the ΔY/ΔX ratio is approximated as
where t and τY are the elapsed time since emission and the removal lifetime of species Y, respectively.
 Strictly speaking, ΔY/ΔX ratios are affected by chemical and dilution processes, and these two effects are generally not distinguishable [Parrish et al., 1992; McKeen et al., 1996; Takegawa et al., 2003]. However, we can use the equation (5) under the conditions where the mixing ratio of Y in the background air is much smaller than that in the plume. This condition is applicable to the short-lived species considered in this analysis (i.e., Y = NOx, NOy, C2H4).
5. Results and Discussion
5.1. Comparison of Observations With Emission Inventory
5.1.1. CO/CO2 Emission Ratio
 The ΔCO/ΔCO2 ratios observed in the plumes can be directly compared to the CO/CO2 emission ratios from the source regions because the production and loss of CO during transport should be negligible on a timescale of ∼4 days. We consider that biogenic sources/sinks of CO2 such as vegetation did not significantly affect the ΔCO/ΔCO2 ratios for this analysis because of the following two reasons. First, gross flux of CO2 from/to vegetation is likely small during the PEACE-A period because biogenic activity is weak in winter. Second, we have chosen distinct plumes in which the enhancements of CO2 are large (>5 ppmv) and the correlation of CO versus CO2 is tight (r2 > 0.7). Vegetation is generally distributed over a wide area as compared to the anthropogenic sources. Thus it should affect the CO2 mixing ratios in regional background air (i.e., air masses that dilute the plumes) rather than in the formation of such distinct plumes.
Figure 5 shows correlations of CO versus CO2 for all of the plumes listed in Table 2. The most important feature is that the observed ΔCO/ΔCO2 ratios exhibit distinct values depending on the source region, indicating that they can be used as a good tracer for investigating the origins of anthropogenic plumes. In addition, the observed ΔCO/ΔCO2 ratio falls within the variability range of the CO/CO2 emission ratio for most of the plumes, demonstrating the overall consistency among the aircraft measurements, trajectories, and emission characterization. We note that the observed ΔCO/ΔCO2 ratios were higher by a factor of 1.5–2 than the average CO/CO2 emission ratio for the Japan plumes (Table 2). Considering that the uncertainty in the emission rate of CO is much larger than that of CO2 for Japan (and also for the other regions), it is suggested that the discrepancy is mainly due to the uncertainty in the CO emission rate in the emission inventory. However, difference between observed ΔCO/ΔCO2 ratios and emission inventory does not significantly affect the identification of plumes as long as it is smaller than or comparable to the variability range of the CO/CO2 emission ratio in the emission inventory.
 Examples of individual plumes are shown in Figures 6 and 7 (Korea-1 (F02L08) and N-China-1 (F04L02) plumes in Table 2, respectively). These figures include the sampling points color-coded by CO mixing ratio, back trajectory, and CO-CO2 correlation for the plumes. The ΔCO/ΔCO2 ratio observed in the Korea-1 (F02L08) plume was higher by ∼25% than the average CO/CO2 emission ratio from the Korea region. On the other hand, the ΔCO/ΔCO2 ratio observed in the N-China-1 (F04L02) plume was lower by ∼30% than the average CO/CO2 emission ratio from the Northern China region. A similar tendency is found for the other plumes (Table 2). However, these small discrepancies may not be significant when we take into account the variability range (i.e., minimum and maximum values) of the CO/CO2 emission ratios.
 Correlations of halocarbons with CO can provide additional information on the chemical characteristics of the plumes. Figures 8 and 9 depict the correlations of methyl bromide (CH3Br) and Halon-1211 (H-1211) with CO in the Korea-1 (F02L08) and N-China-1 (F04L02) plumes, respectively (same plumes as Figures 6 and 7). Blake et al.  found that CH3Br is a good tracer of Japan and Korea emissions and H-1211 is a good tracer of China emissions, on the basis of the whole air samples collected during the TRACE-P campaign. The Korea plume shows a higher ΔCH3Br/ΔCO ratio and lower Δ(H-1211)/ΔCO ratio than the Northern China plume, which is consistent with the findings of Blake et al. . Although the CH3Br-CO correlation in the Korea plume exhibits a large scatter (r2 = 0.39), the systematic difference between these two plumes supports the validity of plume identification applied in this study.
5.1.2. NOx/CO and NOx/CO2 Emission Ratios
 Two fresh plumes were sampled below 1.5 km over Nagoya, Japan on 7 and 10 January 2002 (Japan-2 (F02L14) and Japan-4 (F03L08) plumes in Table 2). Figures 10a and 10b show the correlations of NOy with CO and CO2 in the plumes. Assuming that loss of NOy in these fresh plumes was negligible, the ΔNOy/ΔCO (or ΔNOy/ΔCO2) ratios for these plumes can be interpreted as the NOx/CO (or NOx/CO2) emission ratios for this region. The observed ΔNOy/ΔCO ratios were lower by a factor of ∼2 than the average NOx/CO emission ratio estimated from the emission inventory (Figure 10a). By contrast, the observed ΔNOy/ΔCO2 ratios showed good agreement with the emission inventory (Figure 10b). As discussed in section 5.1.1, we suggest that the discrepancy in the NOx/CO emission ratio is mainly due to the uncertainty in the CO emission rate. The uncertainty in the NOx/CO emission ratio in the emission inventory has a significant influence on the estimates of removal rates of NOx and NOy, if we use CO as a tracer of NOx and NOy.
 To obtain a better understanding of NOx/CO2 emission ratios from various sources, we show data observed in fresh savanna fire plumes over northern Australia during the BIBLE-B campaign for comparison [Shirai et al., 2003; Takegawa et al., 2003]. The ΔCO/ΔCO2, ΔNOy/ΔCO, and ΔNOy/ΔCO2 ratios observed in fresh savanna fire plumes were 86 ppbv/ppmv, 0.018 ppbv/ppbv, and 1.5 ppbv/ppmv, respectively. There is a significant difference in the ΔNOy/ΔCO ratios between Nagoya (PEACE-A) and Australia (BIBLE-B), while the ΔNOy/ΔCO2 ratios are comparable. It has been pointed out that the NOx/CO2 emission ratios from biomass burning are relatively uniform (2 ± 1 ppbv/ppmv) for various types of fires, although they have a dependence on the nitrogen content of the fuels [Andreae et al., 1994; Yokelson et al., 2003; and references therein]. The results obtained in this study suggest that the uniformity of NOx/CO2 emission ratios holds even though the types of emission sources are significantly different from biomass burning.
 The preceding discussion provides some insights into the merits and disadvantages of using CO or CO2 as a tracer of NOx from combustion sources. CO2 is a good tracer for NOx because these two species are major products of high-combustion-efficiency processes, while biogenic interactions of CO2 sometimes make it difficult to define the signals from combustion sources, especially for aged air masses. CO does not have interactions with biogenic sources/sinks, while NOx/CO emission ratios show a large variability depending on the combustion efficiency, especially for the cases where the emission sources are heterogeneous. In addition, the uncertainty in the emission rate of CO is generally much larger than that of CO2. In the present analysis we have chosen only distinct plumes where the enhancements of CO2 are large (>5 ppmv) and the correlation of CO versus CO2 is tight (r2 > 0.7). Therefore we choose CO2 as a tracer of NOx rather than CO to estimate the removal rates of NOx and NOy.
5.2. Plume Age Derived From Hydrocarbons
 The elapsed time since emission for each plume (plume age) can be estimated using equation (5) and a pair of NMHCs with known photochemical lifetimes (Table 3). We define plume age (t) as follows:
where (=2.7 days) is the average lifetime of C2H4 during PEACE-A (Table 3). We have chosen C2H2 as a long-lived species and C2H4 as a short-lived species because of the following two reasons. First, the lifetime of C2H4 is 2.7 days, so that the photochemical history of the plumes (typically <4 days old) can be clearly resolved. Second, the emission ratio of C2H4 to C2H2 (EC2H4-C2H2) does not significantly depend on the region of emission (Table 1). The uniformity of the C2H4/C2H2 emission ratio is understandable considering the similarity in the emission patterns of these two species, which are dominated by biofuel burning and transportation for east Asia [Streets et al., 2003, Figure 6].
 The plume age estimated for each plume is shown in Table 4. The number of plumes is smaller than that in Table 2 because it is limited by the availability of NMHC data. The uncertainty in the plume age was estimated from the fitting error in determining the ΔC2H4/ΔC2H2 ratio and the maximum and minimum values of the emission ratio given by equations (3) and (4). The variability range of the OH concentration (see section 4.2) is not included in the error estimate. In addition, injection of fresh emissions of hydrocarbons into the plumes during transport is another source of uncertainty [Parrish et al., 1992] that is not included in the error estimate.
Table 4. Plume Age and Lifetimes of NOx and NOy for Individual Plumes
Plume age was derived from ΔC2H4/ΔC2H2 ratios ([OH] = 5 × 105 cm−3).
Transport time was roughly estimated on the basis of trajectories.
 For comparison, transport time was roughly estimated on the basis of the trajectories and is listed in Table 4. The transport time tends to be shorter than the plume age for each plume. There are two possible explanations for this discrepancy. First, the transport time does not include the residence time (stagnant time) near the source region. Second, the plume age might be overestimated if OH concentrations in the plumes were actually higher than the median value of the model-predicted OH concentration (5 × 105 cm−3), especially near the source regions. Indeed, earlier studies have shown that OH concentrations in fresh biomass burning or ship plumes were significantly higher than those in background air [Hobbs et al., 2003; Song et al., 2003]. Without detailed information on the OH chemistry in the plumes for this analysis, however, we use the median value of the model-predicted OH concentration to estimate the plume age.
5.3. Removal Rates of NOx and NOy
Figures 11 and 12 depict the correlations of NOx and NOy with CO2 in the Korea-1 (F02L08) and N-China-1 (F04L02) plumes, respectively (same plumes as Figures 6 and 7). For both plumes, the ΔNOx/ΔCO2 and ΔNOy/ΔCO2 ratios are significantly lower than the NOx/CO2 emission ratios, indicating that removal of NOx and NOy occurred within the source region or during the transport from it. From equation (5) we obtain
The left-hand sides of these equations give the fractions of NOx and NOy remaining in the observed plumes. Figures 13a and 13b depict the (ΔNOx/ΔCO2)/ENOx-CO2 and (ΔNOy/ΔCO2)/ENOx-CO2 as a function of the plume age (t) defined by equation (6), respectively. Each data point corresponds to the linear regression slope for each plume and is colored on the basis of average sampling altitude. The photochemical age is assumed to be zero for the plumes sampled in the boundary layer over Nagoya, Japan because these plumes are considered to be very fresh (section 5.1.2). The uncertainty was estimated from fitting errors for the ΔNOx/ΔCO2 (or ΔNOy/ΔCO2) ratio and the maximum and minimum values of the NOx to CO2 emission ratios (Table 1). It is important to note that these plots do not represent a Lagrangian history for a particular event because the plumes were sampled at various locations on various days during PEACE-A. In spite of this variability, there is a clear tendency that both (ΔNOx/ΔCO2)/ENOx-CO2 and (ΔNOy/ΔCO2)/ENOx-CO2 show systematic decrease with plume aging. There is a slight dependence of plume age on altitude. The plumes sampled below 1.2 km (warm colors in Figure 13) have plume ages shorter than 2.5 days, while those sampled above 1.2 km (cold colors) have plume ages longer than 2.5 days. However, there is no significant altitude dependence of the removal rates of NOx and NOy.
 The lifetimes of NOx (τNOx) and NOy (τNOy) for the individual plumes were estimated using equations (6), (7), and (8) (Table 4). The average lifetimes of NOx () and NOy () were 1.2 ± 0.4 and 1.7 ± 0.5 days, respectively. The errors were determined from the standard deviation (1-σ) of the average, and the uncertainty in the OH concentration (a factor of 2) was not taken into account for the error estimate. An alternative way for estimating and is to calculate linear regression slopes for the data points of Figures 13a and 13b. They were estimated as = 1.1 ± 0.3 days and = 1.6 ± 0.3 days, which are close to the values shown above. The errors were determined from 68% confidence intervals for the least squares fit. It should be noted that there is no significant difference in the NOx and NOy lifetimes between the plumes, suggesting that differences in the emission sources were not an important factor determining the lifetimes of NOx and NOy during PEACE-A.
 The (ΔNOy/ΔCO2)/ENOx-CO2 values shown in Table 4 indicate that 80–90% of NOy was removed within 4 days following emission. The removal processes should be either dry/wet deposition of NOy compounds or conversion of HNO3 to refractory particles, although quantitative estimates of the contributions from dry and wet deposition are difficult to achieve in this study. Because dry/wet depositions of NOx and peroxyacetyl nitrate (PAN) are generally slow, this leads to a hypothesis that the majority (>80–90%) of NOx emitted from surface sources was oxidized to HNO3 or nitrate and a rapid removal of these species occurred in the plumes. PAN could appear as a major form of NOy in the plumes, as was observed in Asian outflow during PEM-West B and TRACE-P [Kondo et al., 1997a; Koike et al., 2003], if a majority of HNO3 and nitrate were removed from the plumes.
 To obtain a better understanding of the behavior of NOx oxidation products, the average NOx/NOy ratios and (ΔNOz/ΔCO2)/ENOx-CO2 values in the plumes are plotted as a function of plume age (Figures 14a and 14b), where NOz(=NOy–NOx) represents the NOx oxidation products. The (ΔNOz/ΔCO2)/ENOx-CO2 values represent the fraction of NOz produced per NOx emitted (i.e., NOz yield). The NOz yield would eventually approach 100% if there were no removal of NOz. However, it exhibited a broad maximum (10–15%) at plume ages of 2–3 days and then decreased with air mass aging after 3 days. Although this tendency does not represent a Lagrangian history for a particular event as mentioned earlier, it may be interpreted as follows. The NOz yield showed a net increase at plume ages of <2 days because the oxidation rate of NOx was faster than the removal rate of NOz. After 3 days when the NOx/NOy ratio decreased to as low as 10%, the oxidation of NOx was not an effective source of NOz any more, and the NOz yield showed a net decrease with air mass aging.
5.3.1. Interpretation of NOx Lifetime
 Here we investigate whether the estimated NOx loss rates in the plumes can be explained by the conversion of NOx to HNO3. Detailed investigation is, however, somewhat speculative because of the lack of in situ measurements of NOx reservoirs such as HNO3, nitrate, and PAN during PEACE-A. It has been well documented that PAN can be an efficient reservoir of NOx. Near source regions, formation of PAN may be a net sink of NOx. As air masses age, thermal decomposition of PAN can become a net source of NOx. The role of PAN depends on whether its concentration level is above or below the equilibrium point. Thus the effects of PAN formation/destruction on NOx lifetime cannot be assessed in a meaningful way in the absence of PAN observations.
 NOx is oxidized to HNO3 by reaction of NO2 with OH during the daytime,
During the nighttime, NO3 formation followed by hydrolysis of N2O5 on sulfate aerosols could be important for the formation of HNO3.
First, the possible contribution from reaction (R1) is considered. For PEACE-A conditions, the photochemical lifetime of NOx against reaction (R1), , is estimated to be 2.5 ± 0.5 days. The diurnally averaged OH concentration (5 × 105 cm−3), the model-predicted NO2/NOx ratio (0.7–1.0), and the kinetic parameters recommended by Sander et al.  were used for the calculation. The variability range of the OH concentration (see section 4.2) was not taken into consideration for the error estimate. Considering the estimated NOx lifetime of ∼1.2 days, reaction (R1) accounts for about half of the NOx loss rate in the plumes, which suggests that processes other than the reaction of NO2 + OH were also important for determining the NOx lifetime during PEACE-A. It should be noted that is inversely proportional to the assumed OH concentration as is because it is proportional to (equations (6) and (7)). Therefore the above result (i.e., / ∼ 0.5) does not explicitly depend on the uncertainty in the assumed OH concentration.
 Second, possible contribution from reactions (R2)–(R4) is considered using the same box model as descried in section 2.1. Aerosol chemical composition was not measured during PEACE-A. Here we assume that aerosol surface area (A) measured by the Multiple-Angle Aerosol Spectrometer Probe (MASP) [Liley et al., 2002] represents the surface area of sulfate aerosols. The surface area in the plumes ranged from 30 to 100 μm2/cm3 (median of 65 μm2/cm3) under cloud-free conditions. Earlier studies showed that the uptake coefficient (γ) for reaction (R4) exhibits large variability, ranging from 0.01 to 0.1 (central value of 0.04) [Tie et al., 2003, and references therein]. Using the above numbers as input parameters for the box model, the diurnally averaged NOx lifetime due to reactions (R2)–(R4) was estimated to be ∼1 day for the upper limit case (A = 100 μm2/cm3, γ = 0.1), ∼4 days for the middle case (A = 65 μm2/cm3, γ = 0.04), and ∼30 days for the lower limit case (A = 30 μm2/cm3, γ = 0.01). These estimates are subject to large uncertainties because of the wide range of γ and the assumption of the sulfate surface area. On the other hand, if we assume that the conditions in the plumes were between the upper limit and middle case, the NOx loss rate by reactions (R2)–(R4) would be comparable to reaction (R1). Therefore the NOx oxidation via reactions (R1)–(R4) can potentially explain most of the NOx loss processes in the plumes.
Tie et al.  examined the effect of reactions (R2)–(R4) on the NOx budget, on the basis of a global CTM and the aircraft measurement data obtained during the Tropospheric Ozone Production about the Spring Equinox (TOPSE) mission. They found that this heterogeneous process plays an important role in controlling NOx concentrations in the Northern Hemispheric middle-to-high latitudes in wintertime. According to their simulations, the NOx loss rate against reaction (R2), which was the rate-limiting step for reactions (R2)–(R4), was comparable to or even faster than the one against reaction (R1) under similar conditions (altitudes <3 km, latitudes 25°–40°N, December). Thus the findings of Tie et al.  strongly support the hypothesis in this study.
5.3.2. Interpretation of NOy Lifetime
 As mentioned earlier, it was found that only 10–20% of the NOy compounds emitted from surface sources in east Asia remained in the plumes after 4 days. This fraction is smaller than the NOy export efficiencies of 20–40% in the boundary layer over the western Pacific during TRACE-P [Koike et al., 2003; Miyazaki et al., 2003], although the relationship between the export efficiencies and air mass aging was not discussed in those studies. Stohl et al.  estimated an effective mean NOy lifetime of 1.7–1.8 days in air masses exported from the United States to the North Atlantic (mainly at 35°–50°N), on the basis of aircraft data obtained during the North Atlantic Regional Experiment in spring 1996 and fall 1997 (NARE 96, 97). The NOy lifetime estimated in this study (1.7 ± 0.5) is close to the one estimated by Stohl et al. It is important to note that both this study and that of Stohl et al. estimated NOy lifetimes of 1–2 days, despite the significant differences in the type of emission sources and meteorological conditions. A common feature found in these two experiments is that a large fraction of NOy compounds emitted from surface sources were removed within 1–2 days. The transport time of 1–2 days roughly corresponds to the horizontal travel distance of 1000–2000 km in the boundary layer under PEACE-A conditions. Therefore this finding suggests that the processing near the source regions is critical in determining the NOy abundance.
6. Summary and Conclusions
 Removal of NOx and NOy in Asian outflow plumes below 4 km was investigated using aircraft data obtained during the PEACE-A campaign conducted over the western Pacific in January 2002. Correlations of CO with CO2 and back trajectories were used to identify the plumes transported from source regions in the Asian continent. The plumes originating from Japan (Nagoya) or Korea (Pusan) exhibited relatively low ΔCO/ΔCO2 ratios (14–24 ppbv/ppmv), while those from China showed high ΔCO/ΔCO2 ratios (28–48 ppbv/ppmv). The observed ΔCO/ΔCO2 ratios generally fall within the variability range of the CO/CO2 emission ratios estimated from the trajectories and the emission inventory on the basis of the work of Streets et al. , demonstrating the overall consistency among the aircraft measurements and the emission characterization. The present analysis indicates that ΔCO/ΔCO2 ratios can be used as a good tracer for investigating the origins of anthropogenic plumes when the enhancements of CO and CO2 are large enough and the correlation is tight.
 The photochemical age of the plumes was estimated using the observed ΔC2H4/ΔC2H2 ratios and the OH concentration calculated by a constrained photochemical box model. Removal of NOx was investigated using ΔNOx/ΔCO2 ratios and plume age. The photochemical lifetime of NOx was estimated to be 1.2 ± 0.4 days, which was ∼2 times shorter than the lifetime against reaction of NO2 + OH. This result suggests that other processes were also important in determining the NOx lifetime during the PEACE-A period. It is likely that NOx oxidation to NO3 followed by N2O5 formation and hydrolysis was another important factor controlling the NOx lifetime under PEACE-A conditions.
 The lifetime of NOy estimated using ΔNOy/ΔCO2 and plume age was found to be 1.7 ± 0.5 days, indicating the rapid removal of NOx oxidation products (NOz = NOy–NOx) in the plumes. The maximum NOz yield (i.e., fraction of NOz produced per NOx emitted) during the transport was only 10–15%. This maximum appeared at plume ages of 2–3 days, when the NOx oxidation and NOz removal rates were likely balanced. The NOy lifetime obtained in this study was comparable to the NOy lifetime of 1.7–1.8 days that has previously been reported for outflow from the United States, although there are significant differences in the types of emission sources and meteorological conditions between these two studies. A common feature found in these experiments is that a large fraction of NOy compounds emitted from surface sources was removed within 1–2 days. This finding suggests the importance of chemical/physical processing (i.e., NOx oxidation and dry/wet deposition of NOy) near the source regions in determining the NOy abundance on a regional or even global scale.
 The authors thank the staff of Diamond Air Service (DAS) for their support and efforts during the aircraft measurements. The PEACE-A mission was partially funded by the Japanese Ministry of Education, Culture, Sports, Science and Technology (MEXT). The ITCT 2K2/PEACE campaigns were conducted under the framework of the International Global Atmospheric Chemistry (IGAC) project (http://www.igac.noaa.gov/).