Mechanisms controlling surface ozone over East Asia : a multiscale study coupling regional and global chemical transport models

Mechanisms controlling surface ozone over East Asia: a multiscale study coupling regional and global chemical transport models M. Lin, T. Holloway, T. Oki, D. G. Streets, and A. Richter Center for Sustainability and the Global Environment (SAGE), University of Wisconsin-Madison, Madison, WI, USA Institute of Industrial Science, University of Tokyo, Tokyo, Japan Argonne National Laboratory, Argonne, IL, USA Institute of Environmental Physics, University of Bremen, Bremen, Germany Received: 18 August 2008 – Accepted: 3 October 2008 – Published: 3 December 2008 Correspondence to: M. Lin (mlin26@wisc.edu) Published by Copernicus Publications on behalf of the European Geosciences Union.


Introduction
Ozone (O 3 ) is a secondary pollutant produced in the troposphere by photochemical oxidation of non-methane volatile organic compounds (NMVOCs), carbon monoxide (CO), and methane (Fiore et al., 2002) in the presence of nitrogen oxides (NO x =NO+NO 2 ) radicals.Ground-level O 3 is a major ingredient of urban smog that poses a significant risk for public health, and O 3 throughout the troposphere is an important greenhouse gas (Fishman et al., 1979).Ozonesonde observations in Japan (Naja and Akimoto, 2004) and Hong Kong (Liu et al., 2002) show that the seasonal cycle of O 3 in the boundary layer has a broad summer minimum at low latitudes of the Asian Pacific Rim, in contrast to summer maximum observed at regionally polluted sites in North America and Europe (Jacob, 1999).Recent studies over eastern China report that surface O 3 exhibits a narrow peak in early summer (May or June) and a sharp drop in July and August, based on measurements taken in Beijing (Ding et al., 2008;Lin et al., 2008c), at three mountaintop sites (Li et al., 2007), and at the rural site Lin'an (Xu et al., 2008).
However, a distinctly different seasonal pattern of surface O 3 with a broad summertime maximum during May-August is observed at Mt. Waliguan, which is located in the northeastern edge of the Tibetan Plateau.Meteorological simulations (Ding and Wang, 2006) and regional model analysis by tagging emission sources (Ma et al., 2002(Ma et al., , 2005) ) suggest that the elevated surface O 3 concentrations at Waliguan are mostly caused by the downward transport of stratospheric air, as opposed to transport of anthropogenic pollutions from eastern China (Zhu et al., 2004).Other sources affecting Asia include European emissions, which exert the strongest influence on lower troposphere O 3 in East Asia in the springtime, and North American emissions which exhibit the greatest influence in the upper troposphere, in the fall (Wild et al., 2004;Liu et al., 2002;Figures Back Close

Full Screen / Esc
Printer-friendly Version Interactive Discussion transport models (CTMs) due to the complex physical and chemical processes occurring from global to local scales and by their strong coupling across scales.A number of regional CTMs have been applied to study episodic chemical transport and transformation of Asian pollutants in springtime (e.g.Carmichael et al., 2003b;Zhang et al., 2003;Wang et al., 2006b), and the seasonal cycle of surface O 3 in eastern China (Li et al., 2007) and Japan (Yamaji et al., 2006).Results from MICS-Asia regional model intercomparison found that O 3 predictions for July over central east China differ by ∼20 ppb among seven regional CTMs (Han et al., 2007).This divergence in model estimates suggests that important questions remain on the key mechanisms controlling the O 3 budget over East Asia.The chemical mechanisms have been found to substantially impact model predictions and sensitivity to NMVOCs and NO x emission controls over North America (e.g., Sarwar et al., 2008;Luecken et al., 2007;Arnold and Dennis, 2006).However, the response of local and regional O 3 in Asia to choice and implementation of chemical mechanisms has not yet been evaluated.This study employs the regional Community Multiscale Air Quality (CMAQ v. 4.5.1)model (Byun and Ching, 1999;Byun and Schere, 2006) and the global Model for Ozone and Related Tracers (MOZART v. 2.4) (Horowitz et al., 2003) to examine mechanisms controlling surface O 3 over East Asia, including the impacts of boundary conditions (BCs), photochemical schemes, long-range transport, agricultural burning, and meteorological conditions.Section 2 gives an overview of observational data of ground-based stations and satellite remote sensing, regional and global models used, BCs treatment, and chemical mechanisms.Discussion on O 3 seasonality is presented in Sect.3. We first examine the impacts of both global model BCs and photochemistry on seasonal predictions of surface O 3 at available ground-based measurements in Siberia, Japan, China, and Southeast Asia.Then with a focus on the summer season, O 3 production efficiency and spatial variability are analyzed in detail as they respond to different chemical schemes.Satellite data of tropospheric NO 2 columns and fire count are used to discuss the impacts of agricultural burning over central east China.Influence of long-range transport are also evaluated.Section 4 examines diurnal variability of sum- Satellite data of tropospheric NO 2 columns and fire count are used to help interpret the discrepancies between modeled and observed seasonal variations of surface O 3 .Tropospheric NO 2 column (monthly mean, 0.25×0.25 • ) is based on the GOME (Global Ozone Monitoring Experiment) retrieval by Bremen University (Richter et al., 2005).CMAQ results are sampled at the GOME overpass time at China to calculate tropospheric NO 2 column (Lin et al., 2008b).To examine the intensity of biomass burning and its impact on O 3 concentrations, we use the total fire count maps from MODIS (Moderate Resolution Imaging Spectroradiometer) on NASA's Aqua satellite.The MODIS data is obtained from Louisa Emmons at NCAR (National Center for Atmospheric Research).Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version Interactive Discussion

Model configurations
To distinguish mechanisms controlling surface O 3 over East Asia from regional to local scale, the CMAQ model is employed at two horizontal scales, an 81×81 km 2 primary domain over East Asia and a 27×27 km 2 nested domain over Northeast Asia (Fig. 2a).
The fine domain covers eastern China, the Korean Peninsula and Japan, and is designed to avoid crossing strong local sources at the boundaries.The meteorological fields for both domains are generated from the MM5 meteorological model (Grell et al., 1994) driven with NCEP/NCAR 2.5×2.5 • global reanalysis data every six hours for 2001.The MM5 simulations apply the three-dimensional grid nudging technique and include 23 vertical layers with the depth of first layer setting to 73 m.Vertical layer collapsing is performed to generate eight-layer meteorological fields for CMAQ.We have tested CMAQ with a full range of 23 vertical layers and the results show that the vertical layer collapsing does not significantly undermine CMAQ performance.The cloud scheme of Grell et al. (1994) was chosen for the physical parameterization, and the MRF (Medium Range Forecast) scheme of Hong and Pan (1996) was employed for PBL (planetary boundary layer) parameterization.
We have employed the same models to study long-range transport and fate of acidifying substances over East Asia, and the results are presented in Lin et al. (2008b) for model evaluation and in Lin et al. (2008a) for estimating source-receptor relationships.The reader is referred to Lin et al. (2008b) for a detail description of MM5 simulations, emission data and their processing.With mostly same model configurations as described in Lin et al. (2008b), which focuses on acidifying substances, the present study focuses on O 3 and associated chemistry and physics.Additional model treatments are described in Sect.2.3 for gas-phase chemistry and in Sect.

Gas phase chemistry
In the interest of evaluating model sensitivities to photochemical schemes, we have tested two widely used chemical mechanisms, the Carbon Bond IV (CBIV) mechanism (Gery et al., 1989) and the Statewide Air Pollution Research Center (SAPRC99) mechanism (Carter, 2000).The speciation of NMVOCs to the mechanism-dependent species of CBIV and SAPRC99 is based on a new emission processing model as described in Lin et al. (2008b).CBIV is a lumped-structure condensed mechanism in which organic species are categorized according to the reactions of similar carbon bonds (C-C, C=C, C-CHO etc.).The CBIV mechanism in CMAQ contains 36 species and 93 reactions including 11 photolytic reactions.Compared with CBIV, SAPRC99 includes more detailed organic chemistry, explicit organic peroxy radicals, and more complete organic intermediates, and provides better representation of peroxides for low NO x conditions.The SAPRC99 mechanism has assignments for 400 types of VOCs, and can be used to estimate reactivities for 550 VOC categories (Carter, 2000).A total of 24 model species are used to represent the reactive organic product species: 11 are explicit, and 13 represent groups of similar oxidation reactivity and emission magnitudes using the lumped molecule approach.The SAPRC99 mechanism in CMAQ includes 72 species and 214 reactions including 30 photolytic reactions.

Boundary conditions
In addition to chemistry, treatment of lateral and top boundary conditions (BCs) is another major factor affecting regional CTM predictions, especially for long-lived species such as O 3 , PAN (peroxyacetyl nitrate), and CO etc.To examine the impacts of BCs, Compared with lateral-fixed BC derived from in-situ measurements, the global model BC provides more realistic spatial and temporal variability.Global CTMs, however, are themselves uncertain, so using global model BCs might introduce additional uncertainty to regional model predictions (Tang et al., 2007).In this study, the CMAQ 81 km predictions used monthly mean BCs derived from the MOZART global model.The simulation of the global MOZART model is driven with NCEP reanalysis meteorology for 2000-2002 (hereafter referred to as MOZART-NCEP).The evaluation by Holloway et al. (2007) showed that MOZART-NCEP tends to overpredict monthly averaged EANET surface O 3 over Japan except in springtime.The overprediction is particularly significant, up to 40%, in summertime.In order to reduce the uncertainties introduced in the import of MOZART-derived BCs, we adjusted the concentration of O 3 at the domain boundaries to correct for the MOZART bias over Japan as evaluated in Holloway et al. (2007).The percentage adjustments, which vary on monthly basis from 0 to 40%, are plotted in Fig. 2a.Results of the CMAQ predictions with original and adjusted MOZART BCs are presented in Sect.3.1.
Monthly mean BCs provide a seasonal perspective, but do not resolve possible episodic signals.Thus, we drive the CMAQ 27 km prediction with hourly varying BCs extracted from the CMAQ 81 km simulation to see how temporal variation of BCs influences predicted O 3 .The results are presented in Sect. 4.

Seasonal cycle of surface ozone
We performed three CMAQ regional simulations at 81 km horizontal scale with different chemical mechanisms and global BCs, and compared seasonal variations of surface O 3 predictions with ground-based measurements.Measurement sites are classified into three groups representing different seasonal patterns at higher latitudes (Fig. 2a), lower latitudes (Fig. 2b) and mid-latitudes (Figs.2c and 3) in the study domain.Corre-

Impacts of chemical schemes
Comparison of CMAQ simulations with two widely-used chemical schemes shows that the difference in surface O 3 predictions between CBIV and SAPRC99 is negligible in wintertime, but amounts for ∼10 ppb difference in summertime.The divergence of summertime O 3 predictions between two chemical schemes is particularly pronounced at regionally polluted or downwind sites in mid-latitudes (Figs.2c and 3).In contrast, photochemistry exerts a much smaller influence at coastal and island sites in lower latitudes of the Asian Pacific Rim (Fig. 2b).The high chemical sensitivity in the polluted regions and in downwind areas reflects the fact that the summertime O 3 budget is dominated by local processes of NO x -NMVOC-CO photochemistry.Our results over East Asia show that for O 3 , SAPRC99 produces higher values than CBIV, as found over the United States (Arnold and Dennis, 2006;Faraji et al., 2008;Luecken et al 2007 ;Byun, 2002;Yarwood et al., 2003), but while all of these prior studies have focused on O 3 yields at polluted urban sites.We show the same tendencies at remote coastal sites (Oki, Tappi, and Rishiri), reflecting the influence of northeasterly transport in the boundary layer in Northeast Asia summertime (see Fig. 4).We find that both CBIV and SAPRC99 tend to overpredict summertime O 3 at mid-latitudes sites, but the overprediction by SAPRC99 is more significant.CMAQ with the CBIV chemistry better captures both the seasonal variations and magnitudes of observed monthly mean O 3 at most sites shown in Figs. 2 and 3.The SAPRC99 chemistry, however, overpredicts summertime surface O 3 by 15-25 ppb at the mountainous sites (Mt.Tai and Mt.Hua) and the coastal sites (Oki and Pohang).
To further examine the spatial variations of chemical sensitivities, Fig. 4 shows the distributions of monthly mean surface O 3 and peroxyacetyl nitrate (PAN) in July from the SAPRC99 and CBIV predictions.Both simulations show a similar spatial pattern of surface O 3 with low concentrations at south of 30 • N and with a region of O 3 in the excess of 45 ppb stretching across Northern China.Central east China and downwind areas show typical concentrations of 60 ppb or higher.The difference between SAPRC99 and CBIV in predicting O 3 is most pronounced in the region between 30 • N and 50 • N of the study domain.The SAPRC99 mechanism shows a higher O 3 production efficiency over this polluted region where photochemical formation of O 3 tends to dominate the O 3 budget.
Several reasons may explain the higher O 3 predicted by SAPRC99.From the emissions point of view, SAPRC99 includes a more-detailed emission representation of hydrocarbon classification for individual VOC species.Formation of O 3 is thus sensitive to the aggregated emissions of highly reactive VOC species.Emission speciation of VOCs, however, may not take a full advantage of detailed VOC categories by SAPRC99 since most VOC speciation profiles for Asia are based on the source measurements in North America or Europe.In addition to the expected uncertainty in VOC speciation of anthropogenic sources, natural sources of reactive VOCs (especially isoprene and monoterpenes) are poorly known.Isoprene emission, primarily from oaks Introduction

Conclusions References
Tables Figures

Back Close
Full and other deciduous trees, exhibits strong seasonal and diurnal variability, depending on the driving variables which include temperature, solar radiation, Leaf Area Index and plant functional type.Biogenic isoprene plays an important role in contributing the summertime O 3 budget in many areas.A sensitivity test of CMAQ with SAPRC99 is carried out by removing biogenic isoprene emissions, and we find that surface O 3 concentrations over central east China and downwind areas decrease by 10-15 ppb in average.The SAPRC99's overprediction of summertime O 3 is likely related to the uncertainty in estimating natural isoprene flux.Online calculation of isoprene emission depending on meteorological conditions is recommended for future studies.
From the chemistry point of view, SAPRC99 is a more reactive mechanism that includes explicit organic peroxy radicals and isoprene chemistry.As shown in Fig. 4, SAPRC99 produces 50-70% more PAN relative to CBIV over polluted central east China.Gas-phase PAN is produced via the reversible reaction of the peroxyacetyl radical with NO 2 : PAN serves as a reservoir and carrier of NO x in upper troposphere, and it may be transported over long distances and lead to enhanced O 3 production away from the primary NO x sources by releasing NO 2 when the air masses warm up (Glatthor et al., 2007).Instead of producing PAN, CBIV produces 10-30% higher nitric acid (HNO 3 ) which is quickly dry deposited and thus permanently removes a significant fraction of NO x from the system which otherwise may have been recycled.In addition to the difference in O 3 , PAN, and HNO 3 predictions, the two mechanisms also show large differences in formaldehyde (HCHO) and hydrogen peroxide (H 2 O 2 ) concentrations.Photolysis of HCHO can be a significant radical source as well as an important sink of reactive free radicals, so it plays an important role in O 3 chemistry in the troposphere.CMAQ with the SAPRC99 mechanism predicts 10-50% higher HCHO and 20-50% lower H 2 O 2 over central east China and downwind areas than CMAQ with the CBIV chemistry.
Compared with monthly mean available O 3 measurements at sites across east Asia (shown as filled circles in Fig. 4), CMAQ shows correlations of R=0.7 (CBIV) and 0.6 (SAPRC99), and root mean square errors of RMSE=9.0 ppb (CBIV) and 14.6 ppb Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version Interactive Discussion (SAPRC99).Although SAPRC99 includes a more detailed organic chemistry, the results presented here reveal that CMAQ with the CBIV mechanism tends to improve model performance in reproducing observed monthly mean values of ground-level O 3 in China and Japan.The analysis of photochemical sensitivity has demonstrated that the uncertainty in gas-phase chemistry can lead to significant bias for summertime O 3 predictions in the industrial areas of East Asia.SAPRC99 and CBIV response differently to NO x emissions control.Thus choosing a mechanism that best represents regional atmospheric conditions is essential in regulatory applications,i.e., emissions control policy over central east China.

Impact of monsoonal intrusion on summertime O 3
The intrusion of low-O 3 marine air masses is the key mechanism contributing to the summer minimum of surface O 3 at the Asian Pacific Rim (Naja and Akimoto, 2004;Liu et al., 2002).As shown in Fig. 2b, the monsoonal summer minimum is more pronounced at lower latitude sites.Compared with the remote island sites, O 3 sources at Chinese sites show more complex interactions among local photochemical production as explained by high chemical sensitivity (Sect.3.1), transport of regional pollution, and monsoonal intrusion.2c and 3), consistent with results reported for seven regional CTMs (Han et al., 2007) and 21 global CTMs (Fiore et al., 2008) in comparing with EANET data taken in Japan.The systematic summertime overprediction found in both regional and global CTMs is likely due to model inability in simulating cloud activity and convective mixing, and inadequate representation of southwesterly inflow of marine air masses in the global meteorological analysis data.
It should be noted that model results are for 2001 while measurement data at three mountaintop sites (Mt.Tai, Mt.Hua, and Mt.Huang) obtained from Li et al. (2007) are 2 mm/day in average lower than that in August of 2004 and the climatological mean.
To examine the influence of long-range transport, 3-day backward trajectories arriving at 1000 m above ground level over Mt.Tai were computed every 6 h (at 00:00, 06:00, 12:00, and 18:00 UT) for August 2001 and August 2004 using the NOAA HYS-PLIT 4 model.Cluster analysis was applied to all trajectories in August and the results are shown in Fig. 6.We found that air masses originated from polluted North China Plain (NCP) exerts a stronger influence at Mt. Tai in August 2001 as contrast to August 2004.As shown in Fig. 4 of O 3 distribution map in July, summertime O 3 concentrations in NCP are much higher than that in southern China where frequent impacts of marine air masses occur.Thus more long-range transport of O 3 pollution from NCP also contributes to the discrepancies between model results for 2001 and the measurement at Mt. Tai for 2004.
In summary, the large discrepancy of summertime O 3 between model simulations and measurements for central east China are also attributed to the high uncertainty in photochemical production, the weakened EASM in 2001, and relatively more longrange transport from NCP.In addition, we find that CMAQ overpredicts nighttime O 3 mixing ratios at Beijing by 10-20 ppb as compared with the values reported by measurements (Wang et al., 2006a;Lin et al., 2008c;Ding et al., 2008).CMAQ was initially developed for regulatory purposes, for which daytime O 3 is the greatest concern.Thus the nighttime cooling and vertical mixing physics might not be well parameterized as suggested by previous CMAQ evaluations (e.g.Zhang et al., 2006;Arnold and Dennis, 2006).

Impacts of agricultural burning in central east China
Comparison of observed and simulated seasonal cycle of surface O 3 at Mt. Tai and Mt.Hua (Fig. 2c) illustrates that CMAQ with the CBIV chemistry underestimates the peak values in June.To examine if there are additional emissions sources in June missing in the model, we compared the CMAQ-derived tropospheric NO 2 columns with the retrieval values from GOME. Figure 7  The present study uses biomass burning emission data for 2000 from GEIA/RETRO database (http://www.aero.jussieu.fr/projet/ACCENT/RETRO.php) because the data for 2001 with seasonal variability was not available when our study was initialized in 2005.In recent years, a more advanced database called GFED (Global Fire Emission Database) with an inter-annual variability during 1997-2006(van der Werf et al., 2006) ) has been developed.We find that GFED data for June 2001 reports even more intensive biomass burning emissions as compared with both GEIA/RETRO 2000 (used here) and GFED 2000 (highlighting inter-annual variability) estimates.However, GFED 2001 shows similar magnitude to GFED emissions for June 2004 when the measurements at Mt. Tai and Mt.Hua took place.In addition to the uncertainty in O 3 production rate in the CBIV chemistry, the lack of agricultural burning emissions in June in the model would partially explain why the magnitude of O 3 peak values in June is underestimated by CMAQ with CBIV.Introduction

Conclusions References
Tables Figures

Back Close
Full Previous studies examining the budget of tropospheric O 3 along the Asian Pacific Rim (e.g.Liu et al., 2002;Yamaji et al., 2006) generally used monthly or daily mean data.
Here, we use hourly data to examine the role of vertical mixing and chemistry evolution on the diurnal features of ground-level O 3 .Four sets of O 3 predictions with different chemical schemes of CBIV and SAPRC99, horizontal resolutions of 81 km and 27 km, and boundary conditions of monthly mean and hour varying values are compared with the measurement data at the EANET sites located in Japan (Figs. 9∼10).The observed and predicted NO x mixing ratios are also show for the sites where relatively complete NO x observation data are available.

Resolution dependence of chemical processes
The difference between CBIV and SAPRC99 is further examined here by comparing both 81 km and 27 km predictions with hourly measurements data of EANET.The comparison of hourly O 3 predictions between CBIV and SAPRC99 shows an intriguing feature that with 81 km resolution CBIV performs better than SAPRC99, in reproducing the magnitude of observed O 3 , but with 27 km resolution SAPRC99 shows better ability in capturing the observed 1-h maximum of O 3 mixing ratios, for example, 9-11 July at Sado-seki and some days at Ijira.In general the fine-grid predictions give better agreement against observations than the coarse-grid predictions for the minimum and O 3 mixing ratios at most sites.In addition to better resolved precursors emissions and improved meteorological predictions with the high resolution, using hourly varying BCs instead of monthly mean values, might be another key reasons why the fine-grid simulation improves CMAQ performance in predicting O 3 variations.Increased spatial resolution improves results regardless of chemical mechanism.Our results going from hemispheric-scale (81 km) to regional scale (27 km) modeling grids, are consistent with Arnold and Dennis (2006) who show the similar improvement from regional (32 km) to urban (8 km and 2 km) scales.Since we are using global model BCs and then downscaling the 81 km results to 27 km, the results indicate the importance of the resolution-dependence of chemical processes in global-to-regional model downscaling.
The results suggest that the SAPRC99 mechanism may be used to predict high O 3 over urban areas but it requires high quality emission data for O 3 precursors, such as NO x and the aggregated emissions for individual VOC species.The CBIV mechanism is more suitable for O 3 prediction at regional scales with a coarse grid spacing, i.e., 81 km in this study where emissions of O 3 precursors may have higher uncertainties.

Boundary layer evolution and NO x titration
In addition to the chemical processes, the characterization of the near surface meteorology is needed to help interpret the differences between predicted and observed pollutant levels at ground-based measurement sites.
The temporal evolution of O 3 concentrations at the ground level is controlled strongly by the diurnal variation of the atmospheric boundary layer.We examine the correlations between temporal evolution of ground-level O 3 and boundary layer height at a rural site (Ijira-Fig.10a) and two remote sites (Happo-Fig.10b, Yusuhara-Fig.10c).Observations at urban sites are directly influenced by on-site emissions which might not be well represented by 27 km model resolution, thus are not discussed in the present study. In

Rural sites
Ijira is a rural EANET site located in the NO x -rich central Japan.Intensified diurnal variations of NO x and O 3 mixing ratios were observed at the Ijira site (Fig. 10a).All four simulations generally capture the observed obvious diurnal feature of O 3 mixing ratios at Ijira, yet tend to overpredict the observed lower than 20 ppb nighttime mixing ratios.The fine 27 km grid simulations of both CBIV and SAPRC99 significantly improve nighttime O 3 predictions as compared with measurements.Ozone at Ijira is formed and accumulated in the daytime boundary layer with observed concentration levels that reach the daily 1-h maximum of 120 ppb.The radiative cooling at night leads to the formation of the stable surface layer near the ground.Above the stable surface layer and under the upper-level inversion, the characteristics of the atmosphere are relatively uniform; this layer is called the nocturnal residual layer.If elevated NO emission sources such as tall industrial smokestacks are not present, O 3 concentrations in the residual layer remain high, since deposition and other removal processes that occur near the surface are absent aloft.This can be observed at the Mountain site Happo (Fig. 10b).At the rural site Ijira, however, relatively high concentration levels of NO x were observed.Ozone concentrations at ground level at Ijira decrease after sunset to very low values at nighttime due to NO titration (NO+O 3 →NO 2 +O 2 ) and deposition processes and increase dramatically in the morning, reaching a maximum value in the afternoon (1-2 p.m. JST).The negative correlation between O 3 and NO x suggests that O 3 formation at rural Ijira is in the VOC-limited regime.
The difference in nighttime O 3 concentration levels between fine and coarse simulations amounts to 40 ppb during 2-3 July and 21-23 July.During these periods the difference in nighttime boundary layer height is found to be ∼500 m between the coarse and grid simulations of MM5 and explains why the find grid simulations of CMAQ predict lower nighttime O 3 concentrations.At the same time, however, NO x mixing ratios during these days are overpredicted by the fine grid simulations.This is different from 20257 Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version Interactive Discussion 9-11 July when the observed nighttime NO x concentrations reach to 30 ppb and are well reproduced by the fine grid simulations.The artificial dilution of NO x emissions in the coarse grid would partly explain overpredictions of nighttime O 3 during 9-11 July as a result of reduced NO titration.We find that even with the 27-km fine resolution, CMAQ still largely overpredicts nighttime O 3 on several days and it is not influenced by choice of chemical mechanism.
The overpredictions for nighttime O 3 are likely caused by a coarse vertical resolution for the first MM5 layer and vertical mixing physics.The 73-m depth of first MM5 layer may not be fine enough in simulating land surface interactions.We have tested CMAQ with a full range of 23 vertical layers from MM5 and the results do not improve nighttime O 3 overpredictions, suggesting that the vertical layer collapsing is not a primary reason for nighttime O 3 overprediction.In addition, nighttime boundary layer feature and associated vertical mixing play a key role on nighttime concentration level as discussed earlier.Furthermore, the minimum value of vertical eddy diffusivity (K z ) also has an important impact on nighttime concentrations (Zhang et al., 2006).Although CMAQ version 4.5.1 introduces a new option for minimum K z allowing a larger value of minimum K z for a larger fraction of urban land cover, the results presented here using the new K z option suggest that more studies are needed for the vertical diffusion physics applied in the current version of CMAQ.relate (R=0.18) with changes of surface wind speed.High O 3 episodes during 7-10 July and 19-22 are correlated with low winds and stagnant conditions (slow moving high pressure system).The evolution of boundary layer at Yusuhara is quite different from that at Ijira and Happo.The boundary layer heights presented in Fig. 10c suggest that Yusuhara is in the relatively stable surface layer except during 9-11 July.During 9-11 July, a positive correlation of ground-level O 3 to boundary layer height is found at Yusuhara, indicating the influence of vertical transport of aloft pollutants into the surface layer.During 25-27 July, however, the dominance of photochemical production as indicated by high NO x concentrations explains high concentration levels of surface O 3 in this period.

Conclusions
We have employed the CMAQ regional CTM at two horizontal scales (81 km and 27 km) to examine the mechanisms controlling surface ozone (O 3 ) in China and Japan, including the impacts of boundary conditions (BCs), photochemical production, regional transport, burning of agricultural residues over central east China, and meteorological values.MODIS fire count maps suggest that the lack of NO x sources are likely due to the underestimated emissions from agricultural burning in central east China.Improvements of regional CTMs will require the efforts in reducing the uncertainty in emissions, meteorology, photochemistry, and global pollution inflows, and in assimilating large-scale observational data.The characterization of meteorological conditions is needed to get a complete understanding of model performance, which poses a demand for establishing systematic measurements of both meteorological parameters and chemical tracers.Full 2.4 for boundary conditions.Monthly mean NCEP/NCAR winds and geopotential heights during 1980-2005 are used to demonstrate the climatological features of East Asian summer monsoon circulation.In addition, to evaluate the influence of long-range transport on summertime O 3 budget, we calculated three-day back-trajectories using HYSPLIT (HYbrid Single-Par- Surface O 3 at Mt. Sto.Tomos stays as low as 20-30 ppb all year long with weak seasonal variability, indicating seasonal dominance of low emissions sources and marine air masses.At Hedo and Ogasawara, slightly to north, surface O 3 exhibits strong seasonal variations with a maximum in winter and spring, transported from the continental Asia, and a minimum in summer due to southwesterly monsoonal intrusion.The impacts of monsoonal intrusion on O 3 concentrations at three island sites on lower latitudes (Mt.Sto.Tomos, Hedo, and Ogasawara) are reasonably well simulated by the model.All simulations successfully reproduce the observed seasonal cycle of surface O 3 at Hedo with correlation of R≥0.95.The models also show a correlation of R=0.8 at Ogasawara although there is overprediction during January to February likely attributed to the error of MOZART global BCs at the east side.
for March 2004 -February 2005, and aircraft observation over Beijing from Ding et al. (2008) gives the climatological mean during 1995-2005.Wang et al. (2006a) reported approximately 60 ppb of surface O 3 observed in a mountaineous area (40.35 • N, 116.30 • E) 50 km north of the center of Beijing during 20 June -31 July 2005.As compared in Fig. 3, the reported measurement for July 2005 is much higher than the mean value in July duing 1995-2005, suggesting large inter-annual variability of summertime O 3 concentrations over Beijing.To examine if there are large inter-annual variations of meteorological conditions which might affect the increased O 3 concentrations in August 2001 over central east China as simulated by the model, we also analyze NCEP/heights at 850 hPa.To provide a climatological perspective of East Asian Summer Monsoon (EASM), we averaged NCEP/NCAR winds and geopotential heights at 850 hPa in June and August over 1980-2005.Figure 5 compares the climatological patterns with those for 2001.Winds and geopotential heights for 2004 show similar patterns to the climatological mean, thus the maps are not shown here.TheWestern North Pacific (WNP) anticyclone plays an important role in the connection between the EASM and the tropical SST anomalies (e.g.Lee et al., 2008), and is characterized by southwesterly airflow along the east Asian coast.The monsoonal flow transports Asian O 3 pollution along the Pacific Rim towards the northeast, and the pollution episodes are then subject to frequent lifting into the upper troposphere by convection(Liu et al., 2002).As seen from the distribution of geopotential heights for 1980-2005, the monsoon band shifts to northwest in August.Central east China is thus subject to more frequent cloudy and rainy weather in August as contrast to June(Ding et al., 2008), whereas central and south Japan is under weakened monsoon conditions in August.The more frequent impacts of monsoonal air masses from western Pacific to central east China could partially explain the low O 3 observed in August at the eastern China sites, including Mt. Tai and Hua measurements for 2004 (Fig.2c), Beijing aircraft observatory during 1995-2005 (Fig.3)(Ding et al., 2008), and ground-based measurement at Shangdianzi nearBeijing during 2004-2006(Lin et al., 2008c).In 2001, however, we found that the Western North Pacific (WNP) anticyclone moved to the northeast in August as compared with the climatological pattern.The northeastward shift of WNP anticyclone in August 2001 results in relatively less southwesterly monsoonal airflow to central east China and partially contributes the increased O 3 concentration simulated for August 2001.The geopotential heights in Fig. 5 show that there is a localized high pressure system over central east China in August 2001.The relatively high geopotential heights over central east China suggests the weakened EASM in the same region.Supporting this conclusion, comparison of two NCEP precipitation fields shows that the precipitation amount over central east China in August 2001 is maximum O 3 mixing ratios on observed low O 3 days.The average difference between the 81 km and 27 km simulations amounts to 40 ppb during these days.CMAQ 81 km calculations with both mechanisms tend to overpredict the maximum and minimum O 3 mixing ratios for observed low O 3 days at Sado-seki (2-8 July and 17-26 July) and at Oki (3-5 July).The overprediction of O 3 with 81 km resolution explains why the coarse grid simulations significantly overpredict monthly mean O 3 at Oki as presented in Fig. 2c.Overall the fine-grid predictions give better agreement for the diurnal variations of addition to O 3 and NO x , Figs. 10a∼10c also give the predicted boundary layer height (right/bottom axis) and wind speed at 10 m (left/top axis) from the 81 km and 20256 4.2.2Remote sitesDifferent from the rural site Ijira, an obvious diurnal pattern of surface O 3 is not observed at Mt. Happo and Yusuhara, which are located in relatively clean remote areas.In remote areas outside the impact of urban pollutant plumes, the vertical mixing process might be the dominant mechanism for the buildup of ground-level O 3 .All four simulations well capture the day-to-day variability in the observed data at Happo and Yusuhara.Although a similar diurnal pattern of boundary layer heights is shown at Ijira and Happo, nighttime O 3 at Mt. Happo remains high because elevated NO x sources are not present.Daily variability of ground-level O 3 at Happo is shown to better cor- photochemical production involving anthropogenic and natural emissions, and East Asian summer monsoon circulation.For example, the O 3 budget in mid-latitudes early summer is dominated by the photochemical production involving precursor emissions of NMVOC and NO x , whereas monsoonal intrusion of low-O 3 marine air masses from tropical Pacific plays a key role in controlling the O 3 budget in the boundary layer at lower latitudes of the Asian Pacific Rim.We find that the summer minimum of surface O 3 at two remote island sites (Hedo and Ogasawara) of south Japan and at two rural sites(Lin'an and Mt.Huang)  of eastern China are well reproduced by the model.Over central east China, however, the simulated O 3 for 2001 does not exhibit the same sharp drop in July and August concentrations that is observed at two mountaintop sites (Tai and Hua) for 2004 and at Beijing for 1995-2005.To help interpret the pronounced discrepancies of summertime O 3 between model simulations and measurements in central east China, we have examined the results of CMAQ with two widely used chemical schemes: CBIV and SAPRC99.The results suggest that there are high uncertainties in the photochemical production of summertime O 3 over the same region.The SAPRC99 mechanism produces higher values of mean summertime O 3 than the CBIV mechanism, amounting to a difference of 10 ppb, and explains why CMAQ with SAPRC99 tends to overpredict the observed O 3 over central east China and central Japan up to 25 ppb.CMAQ with the CBIV chemistry shows better performance in reproducing the monthly mean summertime minimum of observed O 3 .Analysis of NCEP winds and geopotential heights demonstrates that the southwesterly monsoonal intrusion to central east China is weakened in August 2001 compared with the climatologically mean for 1980-2005.Further analysis of the diurnal cycle of O 3 mixing ratios at the Japanese sites shows that vertical mixing associated with boundary layer evolution has an important effect on the magnitude and temporal evolution of ground-level O 3 .CMAQ with 81-km resolution tends to overpredict nighttime O 3 in urban and rural areas of East Asia.In terms of the uncertainty of precursors emissions, relatively strong densities of tropospheric NO 2 columns appear in June over central east China from the GOME instrument, while are not seen in the model-derived Fig. 1.Model domains (outer region for 81×81 km 2 and inner region for 27×27 km 2 ) and ground-based measurement sites Fig. 2a.Seasonal variations of observed and predicted surface ozone at the measurement site at (a) high latitudes, (b) low latitudes, and (c) middle latitudes.Summer months are marked as gray.The dotted line shows the percentage adjustments of O 3 concentrations imported from the MOZART global model at the domain boundaries.The adjusted percentage is correspondent to the bias between MOZART predictions and measurements over Japan.The whiskers are 1 standard deviation of measurement data.

Fig. 7 .
Fig. 7. Satellite measurements of tropospheric NO 2 column from GOME (left panel) and the calculated values from CMAQ (right panel) at the ERS-2 overpass time for June, July and August of 2001.The crosses indicate the locations of measurement sites shown in Figs. 2 and 3.
Discussion includes the sensitivity of model resolution and time-varying BCs, and the correlation with NO x concentrations, boundary layer height, and surface wind speed.
3 in China.To help validate the simulated seasonal cycle of surface O 3 in China and Southeast Asia, we collected measurement data from recently published research papers (Table1).Locations of measurement sites are shown in Fig.1.It should be noted that measurement data collected from different research papers might not be in the same year as the 2001 model simulations.In Sect.3.3, we have discussed inter-annual variability of meteorological conditions and associated impact on O 3 concentrations.
/www.arl.noaa.gov/ready/hysplit4.html).The HYSPLIT model was driven with the formated Final Analysis data (FNL) which has a horizontal resolution of approximately 190 km, 13 vertical layers and 6-h temporal resolution.
., Introduction Lin'an and at three mountaintop sites Huang, Tai and Hua, the observed surface O 3 exhibits a late spring/early summer maximum and drops to low values in July and August.Compared with the observed seasonal trend of surface O 3 in eastern China, the model shows correlations of R=0.54 at Lin'an, R=0.45 at Mt. Huang, R=0.68 at Mt. Tai, and R=0.68 at Mt. Hua.The observed peak in late spring and sharp drop in July of surface O 3 at Lin'an and Mt.Huang are well reproduced by the model, but surface O 3 in August at all eastern China sites are overpredicted.Especially at Mt. Tai, Mt.Hua and Beijing, model simulations with both gas phase chemistry of CBIV and SAPRC99 do not capture the decreasing trend of observed O 3 during summer months of June, July and August.Overall, CMAQ tends to overpredict surface O 3 in July and August at the midlatitudes sites in central Japan and central east China (Figs. Streets et al. (2003)ude and distribution 20253 Introduction July and August are reasonably reproduced in the model using anthropogenic emissions data fromStreets et al. (2003), suggesting the overprediction of O 3 in central east China in July and August is not due to the errors of NO x emissions.In June, however, we find that strong NO 2 sources over central east China from the GOME retrieval are missing in the CMAQ derived values.We have also examined NO 2 satellite data from more recent instruments (SCIAMACHY, GOME-2, and OMI) and found similar pattern in other years with greater June NO 2 over central east China.The stronger signal of June NO 2 columns from satellites could be attributed to longer lifetime of NO x compared with rainy/cloudy July and August, and also to additional sources from agricultural burning after harvest in central east China.The MODIS total fire count maps depict that intensive biomass burning activities occurred in June 2001 (Fig.8), which partially explains the additional NO 2 sources in June from GOME.Distinct differences between June and August MODIS fire count maps over central east China are consistent with ATSR-2 (Along Track Scanning Radiometer 2) total fire count maps (http://dup.esrin.esa.int/ionia/aboutionia.asp).

Table 1 .
Uno, I., Carmichael, G. R., et al.:Large-scale structure of trace gas and aerosol distributions over the western Pacific Ocean during the Transport and Chemical Evolution Over the Pacific (TRACE-P) experiment, J. Geophys.Res., 108, doi:10.1029/2002JD002946,2003.20242Introduction Measurement data used in this study.