Interactive comment on “ Investigation of NO x emissions and NO x-related chemistry in East Asia using CMAQ-predicted and GOME-derived NO 2 columns ” by K . M . Han et al .

In this study, NO2 columns from the US EPA Models-3/CMAQ model simulations carried out using the 2001 ACE-ASIA (Asia Pacific Regional Aerosol Characterization Experiment) emission inventory over East Asia were compared with the GOME-derived NO 2 columns. There were large discrepancies between the CMAQ-predicted and GOME-derived NO2 columns in the fall and winter seasons. In particular, while the CMAQ-predicted NO 2 columns produced larger values than the GOME-derived NO 2 columns over South Korea for all four seasons, the CMAQ-predicted NO2 columns produced smaller values than the GOMEderived NO2 columns over North China for all seasons with the exception of summer (summer anomaly). It is believed that there might be some error in the NO x emission estimates as well as uncertainty in the NO x chemical loss rates over North China and South Korea. Regarding the latter, this study further focused on the biogenic VOC (BVOC) emissions that were strongly coupled with NO x chemistry during summer in East Asia. This study also investigated whether the CMAQ-modeled NO2/NOx ratios with the possibly overestimated isoprene emissions were higher than those with reduced isoprene emissions. Although changes in both the NOx chemical loss rates and NO 2/NOx ratios from CMAQmodeling with the different isoprene emissions affected the CMAQ-modeled NO2 levels, the effects were found to be limited, mainly due to the low absolute levels of NO 2 in Correspondence to: C. H. Song (chsong@gist.ac.kr) summer. Seasonal variations of the NO x emission fluxes over East Asia were further investigated by a set of sensitivity runs of the CMAQ model. Although the results still exhibited the summer anomaly possibly due to the uncertainties in both NOx-related chemistry in the CMAQ model and the GOME measurements, it is believed that consideration of both the seasonal variations in NO x emissions and the correct BVOC emissions in East Asia are critical. Overall, it is estimated that the NO x emissions are underestimated by ∼57.3% in North China and overestimated by ∼46.1% in South Korea over an entire year. In order to confirm the uncertainty in NOx emissions, the NOx emissions over South Korea and China were further investigated using the ACEASIA, REAS (Regional Emission inventory in ASia), and CAPSS (Clean Air Policy Support System) emission inventories. The comparison between the CMAQ-calculated and GOME-derived NO2 columns indicated that both the ACEASIA and REAS inventories have some uncertainty in NO x emissions over North China and South Korea, which can also lead to some errors in modeling the formation of ozone and secondary aerosols in South Korea and North China.


Introduction
Nitrogen oxides (NO x ≡NO+NO 2 ) emitted from anthropogenic sources, such as fossil fuel combustion and biomass burning, as well as natural sources, such as lightning and microbiological processes in soil, play important roles in tropospheric ozone chemistry and secondary aerosol formation.Several studies have focused on NO x emissions from China to determine their influence on air quality and aerosol radiative forcing in East Asia (e.g., Uno et al., 2007;Wang et al., 2007;Zhang et al., 2007).Recent studies using satellite measurements reported that NO 2 columns (or NO 2 vertical column density, VCD) have increased significantly in East Asia since 2001 (Richter et al., 2005;van der A et al., 2006;He et al., 2007).Such increases in NO x emissions over China were confirmed partly by a bottom-up emission inventory study (Zhang et al., 2007).In order to test the accuracy of NO x emissions, several studies were carried out over East Asia comparing the 3-D model-predicted NO 2 columns with satellite-derived NO 2 columns (Kunhikrishnan et al., 2004;Ma et al., 2006;van Noije et al., 2006;Uno et al., 2007).The comparisons revealed large inconsistencies between the NO 2 columns from the 3-D CTM (Chemistry-Transport Model) simulations and the satellitederived NO 2 columns.For example, Uno et al. (2007) reported that the 3-D CTM-derived NO 2 columns with the REAS (Regional Emission inventory in ASia) emission inventory were lower by a factor of 2-4 over polluted Central East China, compared with the GOME (Global Ozone Monitoring Experiment)-retrieved NO 2 columns.In addition, Ma et al. (2006) also reported that the 3-D CTM-derived NO 2 columns with the ACE-ASIA (Asia Pacific Regional Aerosol Characterization Experiment) emission inventory underestimated the NO x emissions over China during the summer for the year 2000 by more than 50% compared with the GOMEderived NO 2 columns.However, there has been no detailed investigation carried out as to how and why the NO 2 columns over China were under-predicted by 3-D CTM simulation using the ACE-ASIA or REAS inventory.In addition, there are no reports on the possibly important seasonal (or monthly) variations in NO x emissions in East Asia and the relationship between the rates of NO x chemical loss and the fluxes of biogenic isoprene emissions in East Asia, even though they could be important factors for evaluating NO x emissions.The latter (biogenic emissions) could be important because they can control the levels of OH radicals, which can affect the NO x chemical loss rates.
On the other hand, an accurate estimation of NO x emissions in China is important because NO x emissions from North China tend to persistently affect the air quality of South Korea (e.g., Arndt et al., 1998).From 2003, the Korean government began to implement an ambitious pollution abatement policy that aimed at improving the air quality of Seoul Metropolitan area, called the "Total Air Pollution Load Management System", by reducing the levels of secondary pollutants, such as O 3 , PANs (Peroxy Acetyl Nitrates), and nitrate (Korean Ministry of Environment, 2006).This policy included a specific plan to reduce the total anthropogenic NO x emissions from the Seoul Metropolitan area by 53%, from 309 387 Ton yr −1 to 145 412 Ton yr −1 .The reference and target years for the Total Air Pollution Load Manage-ment System are 2001 and 2014, respectively.However, the critical and largest uncertainty in implementing this policy is to evaluate and quantify accurately the influences of the emissions outside the policy domain on the air quality of the Seoul Metropolitan area.Due to the strong and persistent source-receptor relationship between North China and South Korea, it is important to use accurate emission inventories for both the source ("North China") and the receptor regions ("Seoul Metropolitan area" or "South Korea") for the reference and target years, 2001 and 2014.This study examined the accuracy of NO x emissions from North China and South Korea using the CMAQ-simulated and GOME-derived NO 2 columns.In addition to the uncertainty in NO x emission itself, this paper also discussed the possibly important uncertainty factors that could cause inconsistencies between the CMAQ-derived and GOMEderived NO 2 columns.In particular, both seasonal variations in the NO x emissions and the HO x -NO x -isoprene photochemistry in East Asia were examined in detail on account of its possibly strong relationship with the inconsistency between the CMAQ-simulated and GOME-derived NO 2 columns in East Asia.

Experimental methods
In this study, three dimensional Eulerian CTM simulations over East Asia were carried out in conjunction with the Meteorological fields generated from the PSU/NCAR MM5 (Pennsylvania state University/National Center for Atmospheric Research Meso-scale Model 5) model in order to compare the CTM-predicted NO 2 columns with the satellite (GOME)-derived NO 2 columns.

US EPA Models-3/CMAQ modeling
In this study, a 3-D Eulerian CTM, US EPA Models-3/CMAQ (Community Multi-scale Air Quality) model was used in conjunction with the MET fields generated from PSU/NCAR MM5 modeling over an approximately 3 week period for four seasons: Late Fall (9-27 November 2001), Spring (25 March 2002-13 April 2002), Late Summer (24 August 2002-13 September 2002), and Winter (11-28 February 2003) (Byun and Ching, 1999;Byun and Schere, 2006).The details of the modeling conditions were reported by Song et al. (2008).For the MET fields, the 2.5 • ×2.5 • resolved re-analyzed National Centers for Environmental Prediction (NCEP) data with automated data processing (ADP) of the global surface and upper air observations were employed using four-dimensional data assimilation (FDDA) techniques (Stauffer andSeaman, 1990, 1994).The MET fields were generated at "1-h intervals" during the four episode periods.The CMAQ modeling system then used the meteorological fields generated from the PSU/NCAR MM5 and emission fields.The schemes selected in CMAQ modeling are as follows: the piece-wise parabolic method (PPM) for advection (Collela and Woodward, 1984); 4th generation carbon bond mechanism (CBM 4) for gas phase chemistry (Gery et al., 1989); the Carnegie-Mellon University (CMU) aqueous chemistry mechanism for cloud chemistry (Pandis and Seinfeld, 1989;Fahey and Pandis, 2003); the AERO3 module for particulate dynamics and aerosol thermodynamics (Binkowski and Roselle, 2003); and the Wesley scheme for the dry deposition of both gaseous and particulate species (Wesley, 1989).The 4th generation carbon bond mechanism for gas-phase chemistry included explicit VOC species, such as ALD2 (higher aldehyde, C>2), ETH (ethane), FORM (formaldehyde), ISOP (isoprene), OLE (olefin), PAR (paraffin), TOL (toluene), and XYL (xylene).As indicated above, far more detailed atmospheric chemistry and physical processes, aerosol dynamics, and thermodynamic gas-aerosol processes were considered in these calculations than in other global and regional chemistry-transport modeling studies in order to better consider the atmospheric fate of NO x .For example, in this study the analysis was not restricted to "clear sky conditions".In other words, it fully considered cloud chemistry, wet scavenging, and the effects of clouds on the photolysis reaction rates.
The horizontal domain of the CMAQ modeling shown in Fig. 1 covered the region from approximately 100 • E to 150 • E and 20 • N to 50 • N, which included Korea, Japan, China, and parts of Mongolia and Russia with a 108 km×108 km grid resolution.For vertical resolution, 24 layers were used with σ -coordinates using the model-top at 180 hPa.For comparison, NO 2 vertical column loading was integrated from the surface to 250 hPa (approximately corresponding to ∼10 km a.s.l. in these calculations).The CMAQmodeled NO 2 columns were averaged between 10:00 LST and 12:00 LST because the GOME measurements were taken approximately at 10:30 LST over East Asia.The total number of the grid points in the CMAQ model calculations was 36 432. Figure 1 also shows the four main study regions used for the comparison studies: i) North China (Region A, 30 Here, the remote continental areas in China were excluded from our analysis, partly because they are remote areas, and the NO 2 columns showed a similar order of magnitude of the absolute errors to the GOME measurements (∼10 15 molecules cm −2 ).

Emissions
Emission is an important input parameter in a modeling study.Poor agreement between 3-D modeling studies and satellite measurements is expected if the emission inventories incorrectly reflect the seasonal and spatial emission fluxes from the various sources.In order to consider anthro-  Streets et al. (2003) provided detailed information on the emission inventory (hereafter, labeled the ACE-ASIA inventory) used in this study.The ACE-ASIA emission inventory included NO x emissions from fossil fuels and bio-fuel combustion as well as biomass (vegetation) burning in East Asia.However, the inventory did not consider the NO x emissions from lightning and microbial activity in soil.In general, emission from lightning is believed to make a small contribution to the total NO x budget (Martin et al., 2003).On the other hand, Wang et al. (2007) reported that soil NO x emissions might be important, accounting for up to ∼43% of the combustion source during summer in East Asia.In addition, the original ACE-ASIA emission inventory was built up for the year 2000.Therefore, the NO x emissions for East Asia were modified slightly by multiplying a factor of 1.05 in order to account for an annual increase in NO x emissions from China for the year 2001 (Zhang et al., 2007).Anthropogenic NMVOC emissions were assumed to be constant without any seasonal variation.In this study, chemical speciation (chemical species splitting) of the total NMVOC emissions in East Asia was performed using the SPECIATE database built up by the US EPA.The major biogenic 1 • ×1 • resolved emissions data of isoprene and monoterpene were obtained from the Global Emissions Inventory Activity (GEIA, http://geiacenter.org/presentData/Seasonal variations in the CMAQ-derived NO 2 columns (unit: ×10 15 molecules cm −2 ) in the first column and GOME-derived NO 2 columns in the second column (unit: ×10 15 molecules cm −2 ).The differences between the CMAQderived and GOME-derived NO 2 columns are shown in the third column.nvoc.html), which was created as an activity of the International Geosphere-Biosphere Program (IGBP).Total ground coverage is obtained within 3 days at the equator with a 960 km wide track swath (4.5 s forward scan and 1.5 s backward scan).

NO
The NO 2 analysis for GOME is based on a Differential Optical Absorption Spectroscopy (DOAS) retrieval method (Richter and Burrows, 2002;Richter et al., 2005).The wavelength range of 425-450 nm was used for the NO 2 DOAS fit because the differential absorption is large and interference by other species is small.In addition to the NO 2 cross- section (Burrows et al., 1998), the cross-sections of O 3 (Burrows et al., 1999), O 4 (Greenblatt et al., 1990), H 2 O (Rothman et al., 1992), a synthetic Ring spectrum (Vountas et al., 1998), and an undersampling correction (Chance, 1998) were included in the fit.In order to calculate the tropospheric NO 2 slant column, the stratospheric contribution of NO 2 to the measured slant column was removed by subtracting the slant column taken on the same day at the same latitude in the 180 • -230 • longitude region from the total slant column using the reference sector method (Richter and Burrows, 2002).Cloud screening was applied to remove measurements with a cloud fraction >0.3, as determined from the GOME measurements using the FRESCO (Fast Retrieval Scheme for Clouds from the Oxygen A-band) algorithm (Koelemeijer et al., 2001).The tropospheric slant column was then converted to a vertical tropospheric column using the appropriate air mass factor (AMF).The AMF is defined as the ratio of the observed slant column to the vertical column and was calculated with using the radiative transfer model (SCIATRAN) (Rozanov et al., 1997).The monthly averaged AMF on a 2.5 • ×2.5 • grid was determined using the NO 2 vertical profiles (shape facor) from a global chemical transport model, MOZART-2 (Model for Ozone and Related Tracers).
The error budget of satellite measurements of tropospheric NO 2 columns from GOME has been discussed in detail (e.g.Richter and Burrows, 2002;Martin et al., 2003;Boersma et al., 2004;Richter et al., 2005;van Noije et al., 2006).The main contributions to the error are slant column fitting uncertainties, uncertainties related to the subtraction of the stratospheric contributions, uncertainties from residual clouds, and AMF.For an individual GOME pixel, the slant column fitting uncertainties were 2×10 14 −4×10 14 molecules cm −2 and the uncertainties in the subtraction of the stratospheric contribution were <1×10 15 molecules cm −2 (Richter and Burrows, 2002).These values can be reduced further by averaging the space or time.In remote regions, the retrieval errors were dominated by uncertainties in the slant column fitting and the subtraction of the stratospheric contributions.However, these uncertainties are relatively small in polluted regions.The total uncertainties in the retrieval of tropospheric NO 2 columns over continental polluted regions is largely determined by the AMF calculation due to surface reflectivity, clouds, aerosols, and the trace gas profile (Richter et al., 2005;van Noije et al., 2006).An overall assessment of errors leads to 5×10 14 −1×10 15 molecules cm −2 for additive error and 40-60% for relative error for the monthly averages over polluted areas (Richter and Burrows, 2002;Richter et al., 2005).More detailed error and uncertainty analysis of the GOME NO 2 columns can be found in the previous publications (Richter and Burrows, 2002;Martin et al., 2003;Boersma et al., 2004;Richter et al., 2005;van Noije et al., 2006).

Results and discussions
In order to properly determine the contributions from North China emissions to air quality of South Korea, it is important to evaluate the accuracy of the NO x emission inventories over both regions and understand NO x -related gas-phase chemistry.Initially, 3-day backward trajectory analysis was conducted (Sect.3.1) to confirm the strong source-receptor relationship between North China (A) and South Korea (C).Subsequently, the CMAQ-predicted NO 2 columns are then spatially and seasonally compared with the GOME-derived NO 2 columns (Sect.3.2).The ACE-ASIA NO x emissions in North China (A, source region) and South Korea (C, receptor region) are then compared with other recently-released inventories such as, REAS, "date-back" ANL (Argonne National Laboratory) inventory, and CAPSS (Sect.3.3).

Backward trajectory analysis
In this study, a 3-day backward trajectory analysis for each month of 2001 was carried out using the NOAA HYSPLIT model (Draxler, 1998) to confirm the persistent source (North China, A)-receptor (South Korea, C) relationship, as well as to determine how frequently the air masses travel from North China (A) to South Korea (C).Approximately 20 days per month were selected.The trajectory ends at a point (37.5 • N, 127.0 • E) at an altitude of 1 km under which Seoul is located.As shown in Fig. 2, the air masses traveled from North China (A) to South Korea (C) during almost the entire year.However, in July, the air masses appear to be affected by the emissions from South China (B).In August and September, the air masses arrive in Seoul from the North, East, South (August) and the Northeast (September).Although air masses do not always travel from North China (A) to Seoul during July, August, and September, it is clear that the South Korean air quality is most strongly and persistently affected by the emissions from North China (A) throughout almost the entire year.Therefore, in the framework of the source-receptor relationship, this study focused particularly on the emissions from two regions, North China (A) and South Korea (C).
3.2 CMAQ-predicted and GOME-derived NO 2 columns, NO x emissions, and NO x -related chemistry in East Asia

CMAQ-predicted vs. GOME-derived NO 2 columns
Figure 3 shows the spatial distributions of the CMAQpredicted NO 2 columns and the GOME-derived NO 2 columns for four episodes over East Asia.Figure 4 shows a close-up of the area of South Korea for better visualization.There were large discrepancies between the two quantities in the late fall and winter seasons (i.e.cold seasons), as well as strong seasonal variations, particularly in the GOME-derived NO 2 columns.Interestingly, the CMAQpredicted NO 2 columns were larger than the GOME-derived NO 2 columns for all seasons over South Korea (C) (Fig. 4).
On the other hand, the CMAQ-calculated NO 2 columns over North China (A) were smaller than the GOME-derived NO 2 columns for all seasons except for summer (Fig. 3).Note that the differences between the CMAQ-calculated and the GOME-derived NO 2 columns in the third column in Figs. 3  and 4 have negative values ("blue colors") over North China (A) and positive values ("red-orange colors") over South Korea (C).Scatter plots between the CMAQ-predicted and GOME-derived NO 2 columns were also made for North China (A), South China (B), South Korea (C), and Japan (D) for further confirmation.Figure 5 shows that there are no clear seasonal trends in South China (B) and Japan (D), whereas the CMAQ-predicted NO 2 columns over South Korea (C) are obviously larger than the GOME-derived NO 2 columns for all seasons.In addition, the CMAQ-predicted NO 2 columns over North China (A) were clearly smaller than the GOME-derived NO 2 columns for all seasons except for summer.These results are further confirmed through statistical analyses (see Sect. 3.2.2).These results suggest that there are some errors in the estimations of NO x emissions over North China (A) and South Korea (C) in the ACE-ASIA NO x emission inventory.Of course, this inference should be valid only when the following assumption is held: In CMAQ modeling, emission is the largest uncertainty, and the other atmospheric chemical and physical processes are reasonably accurate.In this study, we chose the GOME-derived NO 2 columns as reference values, although the GOME measurements also have uncertainties, as discussed in Sect.2.3, mainly from the assumptions made in the radiative transfer calculations (Richter et al., 2005).In this study, we focused on the highly polluted East Asian regions where the monthly averages of NO 2 columns reach ∼2×10 16 molecules cm −2 (see Figs. 3,4,and 5; also refer to van Noije et al., 2006 andUno et al., 2007).The NO 2 columns are larger than the magnitudes of overall errors of the GOME NO 2 column measurements, 0.5×10 15 −1×10 15 molecules cm −2 .An attempt was also made to determine why the difference between the CMAQ-predicted and GOME-derived NO 2 columns became minimal only during summer in Fig. 3.This issue is discussed in detail in Sect.3.2.3.

Statistical analysis
Table 1 shows the seasonal and regional statistical analyses between the CMAQ-predicted and GOME-derived NO 2 columns.The following four statistical parameters were introduced for statistical analyses: i) Root Mean Square Error (RMSE, absolute error); ii) Mean Normalized Gross Error (MNGE, relative error); iii) Mean Bias (MB, absolute bias); and iv) Mean Normalized Bias (MNB, relative bias).The four statistical parameters are defined in Eqs.(1) to (4): In Table 1, RMSE analysis showed that the magnitudes of the "absolute" differences were much larger over North China (A) and South Korea (C) than over South China (B) and Japan (D) (These are denoted as "bold fonts" in Table 1).Large uncertainties are expected over North China (A) and South Korea (C).The MNGEs range from 39.6% to 66.4% over North China (A) and from 37.9% to 48.6% over South Korea (C).Bias analysis (MB and MNB) showed that the CMAQ-predicted NO 2 columns tend to have positive biases, compared with the GOME-derived NO 2 columns over South Korea (C).In contrast, over North China (A), the CMAQpredicted NO 2 columns tend to have negative biases.These statistical analyses are also in line with the results reported in Sect.3.2.1.

Summer anomaly in NO 2 columns over East Asia
The first and second columns in Fig. 3  Fig. 6.Seasonal variations in the anthropogenic NO x (red thick line), SO 2 (blue dot line), and BC (black dash line) emissions in China (Streets et al., 2003).
GOME-derived NO 2 columns show strong seasonal variations.The differences between the two NO 2 columns reduce to almost zero during summer, and are also small over North China (A) during spring (Fig. 3).There are two main factors that may be involved in these phenomena: i) seasonal variations in NO x emissions; and ii) seasonal variations in the NO x chemical loss rates.First, the distribution of NO 2 columns can be influenced by the seasonal variations in NO x emissions.Streets et al. (2003) reported almost no seasonal variations in anthropogenic NO x and SO 2 emissions, which is in contrast to black carbon (BC) emissions.The monthly fraction of the NO x emissions is almost constant, as shown in Fig. 6, even though the fractions of NO x emissions increase slightly in December and January.If this is true, then the seasonal changes in NO x emissions are not a likely cause of the seasonal variations in the NO 2 columns (Note that uniform NO x emission fluxes were also assumed in the CMAQ model runs based on the almost constant NO x emissions shown in Fig. 6.However, the issue of seasonal variations in NO x emissions in East Asia will be revisited, and explored further in Sect.3.2.4).Secondly, in order to explain this anomalous (or unexpected) phenomenon in NO 2 columns, this study examined the seasonal variations in NO x chemical loss rates in East Asia.The chemical mechanisms for NO x loss are HNO 3 , nitrate, and organic nitrate formation (i.e.N(V) formation) in the atmosphere.The N(V) forms from NO x by reactions (R1) through (R9): NO 2 +OH+M → HNO 3 (M : third body) (R1)

−→ 2H
Therefore, the NO x chemical loss rate (L NO X ) can be constructed by Eq. ( 5) (Song et al., 2003): (5) where, the first, second and third terms in the right hand side of Eq. ( 5) represent the NO x chemical loss rate due to HNO 3 formation via reactions (R1), (R2), and (R3), respectively.The fourth and fifth terms represent heterogeneous nitrate formation by reactions (R4) and (R5), respectively.In Eq. ( 5), the heterogeneous mass transfer coefficients (s −1 ) of k 4,h and k 5,h for NO 3 and N 2 O 5 radicals were calculated using the Schwartz formula (k i =γ i S i v i /4) (Schwartz, 1986).
In the Schwartz formula, γ i , S i , and v i represent the reaction probability, aerosol surface density (µm 2 cm −3 ), and molecular mean velocity (cm s −1 ) for species i, respectively.Since PAN is a temporary reservoir of NO x , both the production and decomposition terms of PAN (i.e., R6 and R7) were considered in Eq. ( 5).In addition, in summer organic nitrate (denoted as RONO 2 in the text and NTR in the CBM4 mechanism of the CMAQ model) formation is important (R8 and R9).Therefore, both reactions were taken into account.
Here, R represents the organic functional group.Among the eight pathways for removing NO x , L NO x is influenced mainly by (R1), (R5), and (R9) in Eq. ( 5) (particularly, (R1) and (R5) during winter and (R1) and (R9) during summer).The third column in Fig. 7 shows the modelpredicted, spatial distributions of the NO x chemical loss rates at the surface.As expected, the rates of NO x loss are much faster during summer than during the other seasons.However, Fig. 7 shows an unexpected phenomenon.The OH levels in spring are, on average, higher than those in summer over China.It is believed that this may be caused by active biogenic VOC (BVOC) emissions in summer (the anthropogenic NMVOC emissions were kept constant in the CMAQ model runs for the four episodes without seasonal variations).The first column in Fig. 7 shows the biogenic isoprene concentrations at the surface for the four episodes.The isoprene concentrations were highest during summer.The biogenic isoprene emissions influence the formation of ozone, and affect rates of NO x removal by controlling the levels of hydroxyl radicals.
Hydroxyl radicals (OH) are produced by O 1D +H 2 O reaction and HONO photo-dissociation.Hydroxyl radicals (OH) are converted to perhydroxyl radicals (HO 2 ) or organic peroxyl radicals (RO 2 ) through reactions with CO, CH 4 , and VOCs.Formaldehydes (HCHO) are also an important source of hydroxyl radicals in that perhydroxyl radicals (HO 2 ) are produced by HCHO photo-dissociation and HCHO+OH reaction.Perhydroxyl radicals (HO 2 ) or organic peroxyl radicals (RO 2 ) convert NO to NO 2 , and are converted back to hydroxyl radicals (OH).Figure 8 gives an illustration of these relationships.Most importantly, the isoprene emissions can create a shift in the HO x cycle.Excessive amounts of isoprene (C 5 H 8 ) can deplete OH radicals, producing HO 2 or RO 2 through reaction with excessive isoprene.Subsequently, under the isoprene-abundant environment, HO 2 and RO 2 radicals are removed from the atmosphere producing two products, hydrogen peroxides (H 2 O 2 ) and organic hydroperoxides (ROOH).Therefore, in summer, the modeled OH concentrations at the surface are quite low in China, where the isoprene emissions are strong (check the anti-correlation between the isoprene and OH radical concentrations during summer in Fig. 7).Therefore, if excessive biogenic isoprene emissions were used in the CMAQ modeling, it is possible that the modeled (or virtual) NO x levels could be higher than the actual NO x levels due to the underestimated L NO x .In addition to isoprene, monoterpenes also convert OH to RO 2 .However, the CMAQ "v4.3" model only considered the gasphase isoprene chemistry.

OH
As mentioned in Sect.2.2, the GEIA emission inventory was used in this study to consider the BVOC emissions in the CMAQ modeling.In this inventory, an isoprene flux of 20.0 Tg yr −1 over East Asia was estimated.However, Steiner et al. (2002) estimated an isoprene flux of only 13.6 Tg yr −1 over East Asia, using the land-cover conditions derived from the AVHRR satellite.Furthermore, Fu et al. (2007) recently estimated an even lower isoprene flux of 10.8 Tg yr −1 over East Asia from an inversion analysis of the GOME-retrieved HCHO columns.Overall, the isoprene flux used in this study might be approximately 1.5 to 2 times larger than those reported by Steiner et al. (2002) and Fu et al. (2007).The CMAQ-simulated HCHO (mostly, isoprene-derived in summer) columns were also compared with the GOME-retrieved HCHO columns for the summer episode (not shown).It was also found that the former is 2.23 times larger than the latter.
As indicated by previous discussions, the rates of NO x chemical loss during summer in the CMAQ model simulations might be lower than expected, due to the low virtual OH concentrations from the use of possibly overestimated isoprene emissions in the modeling study.The rates of NO x chemical loss were expected to be faster if the recent isoprene emissions in East Asia estimated by Fu et al. (2007) were used.Therefore, a set of sensitivity simulations of the CMAQ model were carried out with 100%, 50%, 30%, and 0% of the GEIA isoprene emissions in East Asia.The results from the sensitivity simulations are discussed at the end of this section.
In addition, the use of overestimated biogenic isoprene emissions in the CMAQ modeling can affect the NO 2 /NO x ratios.It is generally believed that there is some confidence in the ability of CTMs to simulate the actual NO 2 /NO x ratios (Martin et al., 2003).The current analysis was also based on the assumption that the actual NO 2 /NO x ratios could be successfully simulated by the CMAQ modeling (Note that the GOME platform only measures the NO 2 columns, not the NO x columns).Therefore, the correct NO 2 fractions are critical in this analysis (Leue et al., 2001).The NO 2 /NO ratios at a pseudo-steady state can be estimated by Eq. ( 6): where J 1 is the NO 2 photolysis reaction constant (s −1 ); and k, k , k , and k (cm 3 molecules −1 s −1 ) are the atmospheric reaction constants for the NO-to-NO 2 conversion reactions through NO+O 3 , NO+HO 2 , NO+CH 3 O 2 , and NO+RO 2 , respectively.The concentrations of HO 2 and RO 2 might also be overestimated if the biogenic isoprene levels are overpredicted using the overvalued isoprene emissions in the CMAQ modeling.This can lead to high NO 2 /NO ratios in Eq. ( 6), which may result in incorrectly high NO 2 /NO x ratios in the CMAQ modeling.
Figure 9 shows the changes in the NO x chemical loss rates and NO 2 /NO x ratios when 100%, 50%, 30%, and 0% of GEIA isoprene emissions were used in CMAQ modeling.(In Fig. 9, Cases I, II, III, and IV represent the CMAQ model runs that use 100%, 50%, 30%, and 0% of the GEIA isoprene emissions, respectively).The first row shows the CMAQcalculated NO 2 levels at the surface.However, unexpectedly, the NO 2 levels at the surface increased with decreasing isoprene emissions (from Case I to IV).The second, third and fourth rows of Fig. 9 show the isoprene levels, OH radical mixing ratios, and NO 2 /NO x ratios at the surface, respectively.As shown in Fig. 9, the isoprene levels decrease continuously from Case I to IV.The OH radical concentrations increase drastically and the NO 2 /NO x ratios decrease, as expected.However, the NO 2 levels increase, as shown in Fig. 9.This is due to the formation of organic nitrates.The fifth and sixth rows show the CMAQ-estimated L NO x from Eq. ( 5), and the sum of the PAN and organic nitrate concentrations.As shown, the sum of PAN and organic nitrate concentrations (mostly, organic nitrates) decrease drastically with decreasing isoprene emissions, which contribute greatly to the decreases in L NO x in the fifth row of Fig. 9.In other words, the increases in L NO x due to the large increase in OH radical concentrations are offset and even partly surpassed by the decreases in L NO x due to the drastically reduced formation of organic nitrates from Case I to IV.This results in the decreasing L NO x (the fifth row) and thus the increasing NO 2 levels (the first row) from Case I to IV.
Here, it should be noted that the rates of organic nitrate formation in summer (particularly, "isoprene nitrate" formation) are highly uncertain.Many field, laboratory, and modeling studies were carried out about the formation rates, yields, and recycling rates of isoprene nitrates in summer (cf. von Kuhlmann et al., 2004;Horowitz et al., 2007;more references therein).Although the CBM4 mechanism in this study has limitation to deal with the organic nitrate formation and recycling (e.g., CBM4 does not consider the recy-cling reactions of organic nitrates to NO 2 ), the above discussion has possibly important implications in comparison studies between the CTM-calculated and satellite-derived NO 2 columns (e.g., Kunhikrishnan et al., 2004;Ma et al., 2006;van Noije et al., 2006;Uno et al., 2007).For example, van Noije et al. (2006) carried out 17 global CTM ensemble simulations of tropospheric NO 2 columns, and compared the results with the GOME-retrieved NO 2 columns for the year 2000.They found similar levels and trends of the ensemble NO 2 columns over East Asia to those of this study but without considering the BVOC emissions over East Asia (at least, the coupled BVOC-NO x chemistry was not fully discussed).Although the NO 2 levels do not vary considerably with decreasing BVOC emissions, as shown in the first row in Fig. 9, these results might be "completely coincidental" due to the effect of a factor (e.g., organic nitrate formation) offsetting that of another (e.g., OH radical concentrations).In the present sensitivity runs, van Noije et al. 's work (2006) corresponds to the fourth column in Fig. 9 (Case IV), whereas in reality, the most likely case would be the second column in Fig. 9 (Case II) according to Fu et al.'s suggestions (2007).Therefore, it should be noted that although the CTM-calculated NO 2 columns without BVOC emissions (e.g., Kunhikrishnan et al., 2004;Ma et al., 2006;van Noije et al., 2006;Uno et al., 2007) are similar to those with BVOC emissions (e.g., this study), the detailed chemistry behind the similar CTM-derived NO 2 levels would be completely different.

Monthly variations of NO x emissions in East Asia
The variations in the NO x chemical loss rates were examined.Despite considering the coupled BVOC-NO x chemistry, the largest and smallest differences between the CMAQsimulated and GOME-retrieved NO 2 columns were found in winter and summer, respectively.This suggests seasonal variations in NO x emissions in East Asia.
Several studies reported that NO x emissions from fossil fuel combustion in the United States and in East Asia are almost aseasonal (e.g., Streets et al., 2003;Jaeglé et al., 2005).Therefore, several 3-D global and regional CTM studies were carried out with constant (aseasonal) NO x emission fluxes, and were then compared with the GOME-and/or SCIAMACHY-derived NO 2 columns (e.g., van Noije et al., 2005;Uno et al., 2007).On the other hand, several researchers recently reported that there should be monthly variations in NO x emissions in East Asia (e.g., Kannari et al., 2007;Shin et al., 2008).In this study, the emission analysis data in East Asia was collected, and the monthly (or seasonal) variations in NO x emissions from four major emission sectors (industry, power generation, transport, and residential sectors) in China, South Korea, and Japan were investigated.Table 2 summarizes the methodologies used for estimating the monthly-varying NO x emissions in East Asia.In particular, in this study, the monthly variation factors from

A C
Case II Case I Case IV Case III Fig. 9. Results from the sensitivity runs of the CMAQ model with 100%, 50%, 30%, and 0% of GEIA isoprene emissions (Case I, II, III, and IV, respectively): NO 2 concentrations at the surface (the first row); isoprene concentrations at the surface (the second row); OH radical concentrations (the third row); NO 2 -to-NO x ratios (the fourth row); NO x chemical loss rates (L NO x ) (the fifth row); and the sum of PAN and organic nitrate concentration (the sixth row).All concentrations were averaged between 10:00 LST and 12:00 LST.The two white boxes on the panels represent North China (A) and South Korea (C). the power generation sector were estimated from the multiyear NO x mixing ratios measured through the Tele-Metering System (TMS: an automatic pollutant emission monitoring system) installed in the power plant (and large-scale point source) stacks (Shin, 2008).Figure 10 shows the estimated monthly variations in NO x emissions from China, South Korea, and Japan.As shown in Fig. 10, NO x emissions from China exhibit strong monthly variations unlike the previous estimations shown in Fig. 6.Based on this, the monthly variation factors of the NO x emissions were applied to the Models-3/CMAQ model runs for the four seasonal episodes.The first column of Fig. 11 shows the NO 2 columns from the CMAQ model runs with the monthly variations in NO x emissions for the four seasonal episodes (in summer 50% of GEIA isoprene emissions were applied).The GOMEretrieved NO 2 columns and the difference between the two NO 2 columns are shown in the second and third columns in Fig. 11, respectively.As anticipated, the CMAQ-derived NO 2 columns in Fig. 11 show clearer seasonal variations than those in Fig. 3. Accordingly, the differences between the two NO 2 columns decrease.However, the general trend remains similar compared with those in Fig. 3.In Table 3, more quantitative (statistical) analysis was carried out.For example, a comparison of the values in Table 3 with those in Table 1 shows that the RMSEs are reduced by 13.9%−19.5% in North China (except for summer), biases (MBs and MNBs) still exhibiting negative.
In spite of the clearer seasonal variations in the CMAQsimulated NO 2 columns, a summer anomaly can be still found.Unlike the cold seasons, the absolute levels of the GOME-derived NO 2 columns were reported to be low over North China (A) in summer (2−6×10 15 molecules cm −2 ), which is the actual and basic limitation in efforts to determine the same magnitudes of the differences between the two NO 2 columns in both the cold and warm season episodes.Since the OMI-or SCIAMACHY-derived "summer" NO 2 columns in more recent years (2007 and 2008) showed much higher NO 2 columns than those in the year 2001 (<2×10 16 molecules cm −2 ; not shown in this study) due to the rapid increases in NO x emissions over East Asia, the impacts of seasonal variations in NO x emissions on the NO 2 columns would be larger and the effects of the reduced isoprene emissions on the NO 2 columns could be different.Hence, they should be re-evaluated for recent years in future studies.

Other factors affecting NO 2 columns
Two factors (NO x chemical loss and monthly variations of NO x emissions) that affect the CMAQ-modeled NO 2 columns were investigated.However, the GOME-derived NO 2 columns also have potential problems in the retrieval procedures, particularly in winter due to the low sun, stable boundary layer, and large aerosol concentrations (van Noije  , 2006).Also, in the ACE-ASIA NO x emission inventory, the NO x emissions from microbiological activity in soil were not considered.On the other hand, according to a recent satellite observation-constrained top-down NO x inventory study reported by Wang et al. (2007), the soil NO x emissions can sometimes account for up to 43% of combustion sources during summer in China, depending on the application of fertilizers as well as seasonally variable temperatures and precipitation.A consideration of the soil NO x emissions over North China during summer could increase the CMAQ (or CTM)-derived NO 2 levels.Another important factor that can influence CMAQ (or CTM)-modeled NO 2 levels is the magnitude of the reaction probability of atmospheric N 2 O 5 radicals (γ N2O5 ).In CMAQ modeling, γ N2O5 was determined between 0.02 and 0.002 according to Riemer et al.'s parameterizations (2003).However, the magnitude of γ N2O5 is currently a "hotly-debated issue", ranging from 0.1 to 0.001 (Jacob, 2000;Riemer et al., 2003;Thornton et al., 2003;Evans and Jacob, 2005;Brown et al., 2006;Davis et al., 2008;more references therein).The magnitude of γ N2O5 is particularly important for winterepisode simulations.This is because in winter, the partitioning between NO 3 and N 2 O 5 radicals shifts toward the N 2 O 5 formation during the nighttime.Depending on the magnitudes of γ N2O5 and nighttime partitioning between NO 3 and N 2 O 5 radicals, the CTM-derived NO 2 levels can vary by 7%∼50% (e.g.Dentener and Crutzen, 1993;Evans and Jacob, 2005;van Noije et al., 2006).Therefore, in order to predict the NO x levels more precisely, the issue of the magnitude of γ N2O5 requires further investigation.

NO x emission inventories for South Korea
As discussed previously, it was found that the CMAQpredicted NO 2 columns over South Korea (C) may be overestimated by ∼46.1% (based on MNGEs in Table 3) compared with the GOME-derived NO 2 columns.This finding can be partly confirmed with the emission fluxes recently released from the NIER for South Korea: CAPSS inventory.The CAPSS emission inventory for South Korea has been built up since 1999 as a part of the "Total Air Pollution Load Management System".The CAPSS was established following the SNAP 97 (Selected Nomenclature for Air Pollution), which was used as the CORI-NAIR (CORe INventory of AIR emission) emission inventory system of the EEA (European Environment Agency).The CAPSS is an 1 km×1 km-resolved, very detailed emission inventory that employs a hybrid approach (a combination of bottom-up and top-down approaches), including intensive surveys on large-scale point sources (such as power plants, smeltering facilities, and chemical & petrochemical plants), mobile sources with different automobile categories and classes, area sources with regional fuel-type consumption statistics, and non-road mobile sources such as vessels, aviation, and construction equipment.In detail, the CAPSS has the following major 11 classification codes: i) electric generating utility (EGU) combustion, ii) non-electric generating utility (NEGU) combustion, iii) industrial combustion, iv) industrial processes, v) storage and transport, vi) solvent utilization, vii) on-road mobile, viii) non-road mobile, ix) waste treatment, x) biogenic, and xi) agriculture (Heo et al., 2002; also, refer to http://airemiss.nier.go.kr/ nape/introduction/methodology/classification.jsp#).
The annual NO x emission fluxes of the CAPSS inventory was compared with those of two other inventories available for South Korea for 2001: ACE-ASIA and REAS.The annual NO x emission fluxes of the REAS inventory were obtained from the official REAS emission web site (http: //www.jamstec.go.jp/frcgc/research/p3/emission.htm).Figure 12 shows the distribution of the annual NO x emissions of the ACE-ASIA, REAS, and CAPSS inventory for the year 2001, showing high emission fluxes in metro-city areas such as Seoul, Incheon and Busan (refer to Fig. 1, regarding the locations of these cities).Table 4 shows the NO x emission fluxes of the CAPSS, ACE-ASIA and REAS emission inventories.The comparison shows that the annual NO x emission fluxes of the ACE-ASIA and REAS inventories were approximately double that of the CAPSS inventory over Seoul and Incheon, and were approximately 3 times larger over the Busan and Ulsan areas.This is in line with the conclusions drawn from a previous comparison study between the CMAQ-predicted and the GOME-derived NO 2 columns, i.e. the NO x emission fluxes of the ACE-ASIA inventory over South Korea were overestimated.Therefore, in order to correctly consider the emissions from South Korea (C), the NO x emissions from the ACE-ASIA and REAS inventory should be replaced by the NO x emissions of the CAPSS inventory, and the monthly (or seasonal) variations in NO x emissions discussed in Sect.3.2.4should be considered.
Figure 13 shows the CMAQ-predicted and GOME-derived NO 2 columns (the first and the second columns of Fig. 13) along with the differences between the two NO 2 columns (the third column of Fig. 13) over South Korea (C).Here, 50% of the GEIA isoprene emissions and the seasonal variations in NO x emissions with the "CAPSS inventory" were applied to the CMAQ modeling.Based upon the comparisons between the results in Fig. 13 and Fig. 4, the degree of inconsistency become much smaller, particularly over Seoul.However, the CMAQ-simulated NO 2 columns are still larger than the GOME-derived NO 2 columns over South Korea (C) during the entire year, which suggests that the CAPSS NO x emission fluxes can still be overestimated, particularly over the Busan areas (the second largest city in South Korea, refer to Fig. 1, regarding the location of Busan).

NO x emission inventories for China
There are several NO x emission inventories in China available for 2001 including i) ACE-ASIA/TRACE-P inventory (Streets et al., 2003), ii) REAS (Ohara et al., 2007), iii) EDGAR (Emission Database for Global Atmospheric Research) (Olivier et al., 1999(Olivier et al., , 2002)), and iv) GEIA.Here, the ACE-ASIA and REAS inventories were used for a comparison study of the NO x emission fluxes from China.
As discussed previously, when the ACE-ASIA inventory was used, the CMAQ-predicted NO 2 columns over North China (A) were underestimated by ∼57.3% (based on MNGEs in Table 3), compared with the GOME-derived NO 2 columns.In order to confirm this, the three emission inventories for China were inter-compared: ACE-ASIA, REAS, and "date-back" ANL inventory.Here, the "date-back" ANL inventory was estimated based on an emission inventory de-  veloped recently by Zhang et al. (2007) for 2006.The ANL inventory for 2006 is an "upgraded" and "updated" version of the ACE-ASIA emission inventory.The former ("upgraded") indicates that the ANL inventory was improved and methodologically evolved (particularly by the considerations of small-scale combustion sources in China).The latter ("updated") means that the ANL inventory reflects the rapidlygrowing NO x emissions from China (Zhang et al., 2007).The ANL inventory for 2006 also accounted for new emission factors, technology renewal, and bottom-up approaches for various emission sources.Since the ANL inventory is only available for "2006", an attempt was made to "dateback" the ANL inventory to "2001" by retaining the "upgraded" components of the ANL inventory but dating-back the "updated" parts of the ANL inventory to 2001.For this work, the NO x emission shapes of the ANL inventory were retained but the NO x emissions over China were reduced using China's statistical data as well as the increase in energy and fossil fuel consumption (Zhang et al., 2007).Figure 14 shows the annual distribution of the NO x emission fluxes of ACE-ASIA, "date-back" ANL, and REAS inventories in the upper panels.The differences in the NO x emission fluxes are shown in the bottom panels of Fig. 14.While the differences between the REAS and ACE-ASIA inventories (Fig. 14e) were relatively small, the differences between "date-back" ANL and ACE-ASIA inventories (Fig. 14d) and between the "date-back" ANL and REAS inventories (Fig. 14f) were relatively large.The annual NO x emission fluxes of the ACE-ASIA inventory are smaller than those of the "date-back" ANL inventory over North China (A).This was confirmed by analyzing the NO x emission fluxes of the ACE-ASIA, "date-back" ANL and REAS emission inventory in Table 5.The comparison shows that the NO x emission fluxes of the "date-back" ANL inventory were ∼10% larger than those of the ACE-ASIA and REAS in-ventory over North China (A), and were approximately 30% larger than that of the ACE-AISA inventory over South China (B).The total amount of NO x emission fluxes over North China (A) were largest in the "date-back" ANL inventory and smallest in the REAS inventory.Overall, the NO x emission fluxes of the ACE-ASIA and REAS inventories were probably underestimated over North China (A), as was reported by Ma et al. (2006) and Uno et al. (2007).It is believed that although the annual NO x emissions of the "dateback" ANL inventory may be the closest to the real situations for 2001, the NO x emission fluxes of the "date-back" ANL inventory were still low based upon comparisons of the CMAQ-predicted and GOME-derived NO 2 columns, in which there was ∼57.3% underestimation in NO x emissions over North China.Again, the correct emission inventory is a critical input for examining the source-receptor relationships.
The importance of using the correct annual NO x emission fluxes from North China for 2001 ("reference year" of the new Korean environmental policy of the "Total Air Pollution Load Management System") to examine the impact of Chinese emissions (source) on South Korean (receptor) air quality cannot be overemphasized.

Summary and conclusions
This study reports on comprehensive comparisons between the CMAQ-predicted and GOME-derived NO 2 columns in order to determine the accuracy of the NO x emission inventory over North China (A) and South Korea (C).Since both regions have a strong source-receptor relationship, an accurate knowledge of the emissions over both the regions is vital for understanding the contributions of North China emissions to South Korean air quality.When the ACE-ASIA emission inventory for 2001 was used, the CMAQ-predicted NO 2 columns were low by ∼57.3% over North China (A) and high by ∼46.1% over South Korea (C) compared with the GOMEderived NO 2 columns.This was further confirmed partly by comparing several emission inventories.The ACE-ASIA and REAS emission inventories showed large uncertainties over North China (A) and South Korea (C).The NO x emission fluxes of the ACE-ASIA inventory over South Korea and North China were overestimated by ∼50% and underestimated by ∼10%, respectively, compared with the CAPSS and "date-back" ANL inventories.Based on these analyses, the "date-back" ANL and CAPSS inventories appear to provide a better estimation of the real situation over North China (A) and South Korea (C), respectively, even though the NO x emissions of the "date-back" ANL inventory is still low.
This study investigated two main issues: (i) HO x -NO xisoprene photo-chemistry in East Asia and (ii) seasonal variations of NO x emissions in East Asia.The former was examined because isoprene chemistry is strongly coupled with the NO x chemical loss rates in North China (A) and South Korea (C).In particular, the biogenic emissions of isoprene and monoterpenes are the key parameter to control OH radicals, whose concentrations can sequentially control the NO x concentrations through nitric acid and particulate nitrate formation.Since recent studies reported lower BVOC emissions in East Asia, this study examined the impact of the BVOC emissions on the NO x chemical loss rates (L NO x ) in East Asia by relaxing the isoprene emission strengths.As expected, the OH levels increased drastically with decreasing isoprene emissions.Nevertheless, L NO x decreased slightly and the NO 2 levels increased accordingly, even with the reduced isoprene emissions.L NO x was retarded due to the less active organic nitrate formation with decreasing isoprene emissions.In addition, monthly variations in the NO x emission fluxes over East Asia were also examined based on two facts: (i) the recent reports of monthly variations in NO x emissions in East Asia; and (ii) the differences between CAMQ-modeled and GOME-retrieved NO 2 columns over North China (A) are the largest in winter and smallest in summer.Although consideration of the monthly variations in NO x emission yielded more consistent results, many issues in NO x -related chemistry and NO x emissions remain unclear, requiring further studies.
The correct NO x (and BVOC) emission inventory and NO x -related chemistry are critical for examining sourcereceptor relationships.Using the corrected emission inventories including the monthly variations of the NO x emissions for North China (A) and South Korea (C), a one year-long Models-3/CMAQ modeling over East Asia is currently underway to examine and quantify more accurately the influences of Chinese (source) emissions on the South Korean (receptor) air quality.

Fig. 1 .
Fig. 1.Modeling domain of the study.Four regions were defined: i) A: North China, ii) B: South China, iii) C: South Korea, and iv) D: Japan.

Fig
Fig. 3.Seasonal variations in the CMAQ-derived NO 2 columns (unit: ×10 15 molecules cm −2 ) in the first column and GOME-derived NO 2 columns in the second column (unit:

Fig. 10 .
Fig. 10.Monthly variations in anthropogenic NO x emissions in China, South Korea, and Japan.

Fig. 11 .
Fig. 11.As in Fig. 3, except for the CMAQ-model runs with 50% of GEIA isoprene emissions and monthly variations in anthropogenic NO x emissions in East Asia.

Fig. 13 .
Fig. 13.As in Fig. 4, except for the CMAQ-model runs with 50% of GEIA isoprene emissions and monthly variations in anthropogenic NO x emissions with the CAPSS inventory in South Korea.

Table 1 .
Statistical analysis for the comparisons between the CMAQ-predicted and GOME-derived NO 2 columns over East Asia.

Month Aug Sept Oct Nov Dec Jan Feb Mar April May June July Fraction of Annual Emissions 0.00 0.05 0.10 0.15 0.20 NOx SO2 BC Average
show seasonal variations of the NO 2 columns over East Asia.The CMAQ-derived NO 2 columns show relatively weak seasonal variations in North China (Region A), whereas the www.atmos-chem-phys.net/9/1017/2009/

Table 2 .
Seasonal allocation method for anthropogenic NO x emission fluxes in East Asia.

Table 3 .
Statistical analysis comparing the CMAQ-simulated and GOME-derived NO 2 columns over East Asia.The CMAQsimulated NO 2 columns were obtained from the model runs with 50% of GEIA isoprene emissions and the monthly-variations in NO x emissions over East Asia.

Table 4 .
Comparison of NO x emissions among the CAPSS, REAS, and ACE-ASIA inventories over South Korea for 2001.

Table 5 .
Comparisons of NO x emission among ACE-ASIA, "date-back" ANL, and REAS inventories over China for 2001.