Modeling study of ozone seasonal cycle in lower troposphere over east Asia


  • Jie Li,

    1. Nansen-Zhu International Research Center, State Key Laboratory of Atmospheric Boundary Layer Physics and Atmospheric Chemistry, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China
    2. Also at Graduate University of the Chinese Academy of Sciences, Beijing, China.
    Search for more papers by this author
  • Zifa Wang,

    1. Nansen-Zhu International Research Center, State Key Laboratory of Atmospheric Boundary Layer Physics and Atmospheric Chemistry, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China
    Search for more papers by this author
  • Hajime Akimoto,

    1. Frontier Research Center for Global Change, Japan Agency for Marine-Earth Science and Technology, Yokohama, Japan
    Search for more papers by this author
  • Chao Gao,

    1. Nansen-Zhu International Research Center, State Key Laboratory of Atmospheric Boundary Layer Physics and Atmospheric Chemistry, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China
    2. Also at Graduate University of the Chinese Academy of Sciences, Beijing, China.
    Search for more papers by this author
  • Pakpong Pochanart,

    1. Frontier Research Center for Global Change, Japan Agency for Marine-Earth Science and Technology, Yokohama, Japan
    Search for more papers by this author
  • Xiquan Wang

    1. Nansen-Zhu International Research Center, State Key Laboratory of Atmospheric Boundary Layer Physics and Atmospheric Chemistry, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China
    Search for more papers by this author


[1] On the basis of three mountain sites (Mount Tai, Hua and Huang) newly founded in east-central China and several other sites from the Acid Deposition Monitoring Network in east Asia (EANET) and WMO World Data Centre for Greenhouse Gases (WDCGG), we investigate seasonal cycle of ozone over east Asia and its budgets in east-central China by using a regional chemical transport model (NAQPMS). The observations show a striking ozone pattern of two sharp peaks in May-June and September–October at three mountain sites in east-central China which are higher than those observed at other mountain sites in Europe and North America. Ozone budgets analysis by the model confirms that maximum of net photochemical productions reaches 31.8, 15.1, and 11.4 ppbv/d at Mount Tai, Hua, and Huang, respectively. The net photochemical production dominates the formation of ozone maximums at Mount Tai and Hua in June, and the importing transport also plays a comparable importance at Mount Huang. In comparison with those in the western North Pacific, east-central China shows stronger net photochemical productions, which are comparable to anthropogenic sources regions in Europe and North America.

1. Introduction

[2] Tropospheric ozone is an important trace gas in the troposphere. Besides being a greenhouse gas [Intergovernmental Panel on Climate Change, 1996], it plays a key role in determining the oxidizing capacity of the atmospheric as a photochemical precursor of OH radicals [Brasseur et al., 1999]. High ozone in the atmosphere is also thought to be responsible for the crop damages [Cheung and Wang, 2001; Pochanart et al., 2002].

[3] It is well known that one of the major sources of tropospheric ozone is the photochemical production involving the pollutants emitted by various industrial and anthropogenic activities in the lower troposphere [Crutzen et al., 1999; Lal et al., 2000]. Over the past 3 decades, tropospheric background ozone levels over the midlatitudes in the Northern Hemisphere have been observed to rise by approximately 0.5–2%/a including a substantial anthropogenic component [Vingarzan, 2004].

[4] East Asia is a one of the most rapidly developing regions in the world [Wang and Mauzerall, 2004]. Asian NOx emissions, which contributed only in a minor fraction to global emissions during the 1970s, have increased rapidly since then and surpassed emissions from North America and Europe in the mid-1990s [Akimoto, 2003]. Streets and Waldhoff [2000] warned that NOx emissions from 1990 to 2020 in east Asia would have a double increase due to population growth, economic expansion, and industrial development. As the biggest developing country in the world, China plays a dominative role in east Asian emissions [Streets and Waldhoff, 2000]. Recent studies indicated that east Asian emissions (especially China) had a fast boundary layer (BL) outflow in middle latitudes in spring [Zhang et al., 2004; Wild and Akimoto, 2001]. Liu et al. [2003] argued that Asian pollution outflow is strongest at 30–45°N in the boundary layer, while because of the frontal and orographic lifting, the outflow is also found at 20–35°N in the lower free troposphere in spring. Yamaji et al. [2006] revealed that China had a significant influence on the temporal and spatial variability of tropospheric ozone in east Asia. A global model indicated that the net annual O3 production in east Asia reached 117 Tg/a with strong variability with seasonal meteorology [Mauzerall et al., 2000]. As for the ozone production within China, Luo et al. [2000] reported that rural areas in the southern and northern China tended to be NOx-limited and VOC-limited, respectively.

[5] In order to validate the modeling study of ozone in China, observational data with reasonable regional representativeness are definitely needed. Although such measurements are very limited in China, some measurements at the surface level in nonurban sites have been reported. Luo et al. [2000] collected ozone data from August 1994 to August 1995 at five nonurban sites in China and reported that maximum O3 concentrations at eastern China appeared in the fall and early winter. However, recent observations in 1999–2001 at Lin'an in Yangtze Delta region indicated that the ozone averaged level showed a monthly peak in May, and the concentrations of CO and NOx were about 1–5 times of those in rural areas of North America and Europe [Cheung and Wang, 2001; T. Wang et al., 2001a]. Lam et al. [2001] indicated that the high O3 in winter at Hong Kong was caused by the long-range transport from the continent of China. T. Wang et al. [2001b] showed that pollutants could be transported to western Hong Kong from the inner Pearl Delta region. The observed O3 at Shangdianzi (a rural site around Beijing) showed that O3 had a summer maximum [Liu et al., 2006].

[6] In addition to these surface data near sea level, observational data at mountain sites may be advantageous for understanding the regional pollution since they are expected be more regionally representative being free from effects of local emissions. In this scope, several-month observation data of ozone at Mount Tai has been reported by Gao et al. [2005]. However, full seasonal cycle and even longer-term observational data should be more useful for the validation of models and the study of regional pollution in China including its impact to other countries. For this purpose, we have set up three mountain sites at 1500–2000 m above sea level over east-central China (100–120°E, 30–40°N) including Mount Tai to observe ozone and carbon monoxide (CO) since 2003 (CO has been measured mostly from March 2005).

[7] On the basis of the data observed at three newly founded mountain sites in east-central China and other sites from the Acid Deposition Monitoring Network in east Asia (EANET) ( and WMO World Data Centre for Greenhouse Gases (WDCGG) (, this paper attempts to analyze a full-year seasonal variation of tropospheric ozone over east Asia with a nested chemical transport model NAQPMS (Nested Air Quality Prediction Modeling System). Next, after discussing on seasonal variations in the distribution of ozone level over east Asia, we focus on the mechanisms of the seasonal variation O3 at three mountain sites in east-central China by evaluating the contributions of transport, net photochemical production and deposition.

2. Measurements of Ozone and Data Analysis

2.1. Monitoring Sites in China

[8] Three mountain stations, Mount Tai (117.10°E, 36.25°N, 1533 m a.s.l), Mount Huang (118.15°E, 30.13°N, 1836 m a.s.l) and Mount Hua (110.09°E, 34.49°N, 2064 m a.s.l), have been set up in east China with the cooperation of Frontier Research Center for Global Change (FRCGC), Japan Agency for Marine-Earth Science and Technology (JAMSTEC), and the Institute of Atmospheric Physics (IAP), Chinese Academy of Sciences (CAS), for continuous observation of ozone and CO. In this work, only ozone data have been analyzed. The locations of these stations are shown in Figure 1 together with those of other stations whose ozone and CO data were used for the study of seasonal variation and model verification.

Figure 1.

Distribution of stations in the study and modeling domain of NAQPMS. The numerical characters mean the locations and name in the study. Also shown is the monthly mean NOx emission rate (shaded, μg/m2/s) in August. The regions labeled I, II, III and IV represent four subregions.

[9] Mount Tai is an isolated mountain in the center of east-central China representing a highly industrial region. Mount Huang, located in the south edge of east-central China, is about 280 km southwest of the Yangtze Delta region. Mount Hua is located to the west of east-central China. Briefly, the locations of these three stations have been selected in favor of understanding the character of “regional pollution” in north China avoiding the influence of local pollution. More detailed description of the stations and instruments will be described separately (P. Pochanart et al., Seasonal variations and high mixing ratios of regional ozone pollution in east China, submitted to Geophysical Research Letters, 2007).

2.2. Data From Other Stations

[10] In the present work, data from several other stations over east Asia have been used to study ozone seasonal cycle and to validate the model performances shown in Figure 1. Five regional stations were taken from EANET: Rishiri (141.24°E, 45.13°N, 40 m a.s.l), Sadoseki (138.40°E, 38.25°N, 110 m a.s.l), Oki (133.18°E, 36.28°N, 90 m a.s.l), Yusuhara (132.98°E, 32.73°N, 225 m a.s.l) and Mondy (101.00°E, 51.67°N, 2006 m a.s.l). The detailed descriptions of these sites can be found at The other seven sites were from WDCGG, including Issyk-kul (76.98°E, 42.62°N, 1640 m a.s.l), Plateau-Assy (72.87°E, 43.25°N, 2519 m a.s.l), Ryori (141.8°E, 39.03°N, 260 m a.s.l), Sary-Taukum (75.57°E, 44.45°N, 412 m a.s.l), Ulaan-Uul (111.1°E, 44.45°N, 914 m a.s.l), Yonagunijima (123.02°E, 24.47°N, 30 m a.s.l), and Waliguan (100.9°E, 36.28°N, 3810 m a.s.l) with detailed descriptions given at

2.3. Overview of Observed Data

[11] Figure 2 shows the monthly cycle of ozone at several sites over east Asia (Oki, Yusuhara, Yonagunijima, Mount Tai, Hua and Huang, Mondy) from March 2004 to February 2005. Ozone values in January and February at Oki and Yusuhara were replaced by 2004 for the unavailability of year 2005 data at the moment this manuscript was prepared.

Figure 2.

Apparent season cycles of ozone at several sites (Mondy, Oki, Yusuhara, Yonagunijima, Mount Tai, Mount Huang, and Mount Hua). The symbols are the monthly mean, the characters are the station name, and the whiskers are 1 standard deviation. The locations of sites are shown in Figure 1.

[12] Generally, monthly mean ozone mixing ratios show a late spring or early summer maximum at these sites, which has been reported by previous studies [Pochanart et al., 1999, 2002, 2004; Monks, 2000; Tanimoto et al., 2005]. However, the ozone variations in other seasons were different among these stations. Thus we defined three regions in east Asia (as shown in Figure 1) according to the pattern of seasonal cycle.

[13] Region I mainly covers outflow region of mainland China east of 125°E including East China Sea and Japan. Oki, Yusuhara, and Yonagunijima are chosen as representative stations in this region. At these stations, the maximum occurs in spring (March–May), and reaches 60–65 ppbv in monthly average. After May, O3 starts to decrease till August with a minimum at 20–35 ppbv in July–August. In fall, there is a weak second peak in October–November which is slightly higher than winter. In addition, the month in which the spring maximum occurs is earlier at lower latitudes than the ones at higher latitudes. The latitudinal difference has been reported by Tanimoto et al. [2005].

[14] Region II covers east-central China, where the emissions of anthropogenic pollutants are expected to be the greatest in China (Figure 1) [Streets and Waldhoff, 2000]. In this region, a limited number of seasonal cycle measurement focused on ozone are available at a rural site, Lin'an [T. Wang et al., 2001a; Cheung and Wang, 2001] and several suburban sites [Zhou et al., 2004]. Mount Tai, Huang and Hua are located in this region. At these three sites, O3 showed a well-defined monthly pattern with two sharp peaks in May–June (60–80 ppbv) and September–October (50–70 ppbv), although the second peak in fall at Mount Hua was weak in this year (Figure 2). Thus the primary maximum in spring occurs typically 2 months later, and secondary maximum in fall occurs 1 month earlier in region II than in region I. The pattern is consistent with the observations at Lin'an, where elevated ozone levels were frequently observed in the late spring and fall [Cheung and Wang, 2001; T. Wang et al., 2001a]. Furthermore, it can be seen that summer O3 in region II presents higher level than winter, which is different from the situation in region I. For example, the ozone mixing ratios in July are 10 ppbv higher than and nearly the same level as in February at Mount Tai and Huang, respectively, in contrast to the situation that summertime ozone is lower than 15 ppbv than in winter in region I. These unique patterns suggest that different mechanisms exist in region I and II.

[15] The two-peaked pattern of ozone in this region is distinctly distinguished from that obtained at other rural mountain-top sites over high-polluting regions in Europe and North America, where O3 showed a single broad peak in May–August (figure not shown). For example, at Mount Cimone (10.42°E, 44.11°N, 2165 m a.s.l) [Campana et al., 2005] and Niwot Ridge (Saddle) (105.59°W, 40.05°N, 3538 m a.s.l) (WDCGG), the transport of photochemically processed boundary layer air masses from the Po basin (a polluting source) for Mount Cimone and Denver metropolitan for Niwot Ridge (Saddle) to the mountain summits resulted in the broad peak in summer [Campana et al., 2005; Poulida et al., 1991].

[16] Region III covers Mongolia and Siberia (Figure 1) with few anthropogenic emissions, which was defined as the natural level by previous studies [Pochanart et al., 2003]. Mondy in southern Siberia is defined as the reprehensive station for this region. Figure 2 shows a maximum (61 ppbv) in May at Mondy. From fall to winter, the ozone mixing ratios remain stable, different from other regions.

[17] Generally, various patterns of ozone over east Asia, especially the distinct pattern and elevated ozone in region II, imply the complexity of chemical and physical factors influence on the transformation of ozone over the east Asia. In the following section, we employ a nested chemical transport model NAQPMS for further investigation of the pattern mechanisms.

3. Model Description and Validation

3.1. Model Description and Setup

[18] Long-term simulation was conducted utilizing the Nested Air Quality Prediction Modeling System (NAQPMS), a fully modularized, three-dimensional chemical transport model. The NAQPMS reproduces the various physical and chemical processes over regional- and urban-scale atmospheric pollution and has been applied to study issues related to the mechanisms of dust events in east Asia [Wang et al., 2000a], the transport and chemical processes of pollutants (Sulfur, O3) [Wang et al., 2000b; Z. Wang et al., 2001; Zhu et al., 2004] and also the interaction between mineral aerosols and acid rain over east Asia successfully [Wang et al., 2002]. Since 2002, it has been applied to Beijing, Shanghai and Shenzhen Environmental Protection Bureau as a routine air quality forecast model [Wang et al., 2006b].

[19] The NAQPMS is implemented in two parts: (1) the meteorological model and (2) the nested chemical transport module developed by the Institute of Atmospheric Physics (IAP), Chinese Academy of Science [Z. Wang et al., 2001]. The meteorological fields used in NAQPMS are provided by the fifth generation mesoscale model (MM5) developed by the Penn State University (PSU) and National Center for Atmospheric Research (NCAR) [Grell et al., 1994].

[20] The chemical transport module reproduces the physical and chemical evolution of reactive pollutants by solving the mass balance equation in the terrain-following coordinates. It includes advection and diffusion processes, gas/aqueous chemistry and parameterization of dry/wet deposition [Wang et al., 2002; Zhu et al., 2004]. In the module, the advection scheme employs a simplified but very accurate mass conservative peak-preserving, mixing ratio bounded advection algorithm [Wang et al., 2000b]. Dry deposition module is the Padro's scheme for gaseous species [Padro et al., 1991]. The vertical eddy diffusivity is parameterized by the scheme of Byun and Dennis [1995]. The detailed description can be found elsewhere [Wang et al., 2006a; Zhu et al., 2004].

[21] In this study the gas chemistry has been updated from Carbon-Bond IV to CBM-Z [Zaveri and Peters, 1999], which is composed of 133 reactions for 53 species and employs the lumped structure approach for condensing organic species and reactions. Compared with Carbon-Bond IV, CBM-Z extends the framework to function properly at larger spatial and longer timescales, including revised inorganic chemistry; explicit treatment of lesser reactive paraffins such as methane and ethane; revised parameterizations of the reactive paraffins, olefins, and aromatic reactions; alkyl and acyl peroxy radical interactions and their reactions with the NO3 radical; longer-lived organic nitrates and hydroperoxides; and revised isoprene chemistry based on the condensed one-product mechanism of Carter [1996]. Our modeled dry deposition velocity (0.3–0.4 cm/s) is also close to the value (0.25–0.3 cm/s) in the study of Zhang et al. [2004].

[22] Here the NAQPMS is applied for a full year (March 2004 to February 2005) with a horizontal grid dimension of 81 km × 81 km over the east Asian region shown in Figure 1. The vertical grid structure consists of 20 layers from the surface to the tropopause (about 15 km), with the lowest 7 layers below 1 km. The meteorological fields are simulated with grid assimilation options and well validated with observations before used to drive the chemical transport model.

[23] The anthropogenic emission inventory except CO is obtained from D. G. Streets with 0.1° × 0.1° resolution validated by Trace-P Modeling and Emission Support System (ACESS) [Streets et al., 2003; D. G. Streets, personal communication, 2004]. The anthropogenic CO emissions and NOx from soils were provided by FRCGC with 0.5° × 0.5° resolution ( Natural hydrocarbon emissions (1° × 1°) were from the Global Inventory Activity (GEIA) [Guenther et al., 1995]. The updated power plant emission database of China from China State Environmental Protection Agency (SEPA) is also used in the NAQPMS.

[24] The initial and boundary conditions were provided by the simulated results of a global chemistry/transport model (GEOS-CHEM) at a resolution of 4° × 5° with permission of D. Jacob at Harvard University. The meteorological information for GEOS–CHEM model (v7–02–04) [Bey et al., 2001] is supplied from the Goddard Earth Observing System 4 (GEOS-4) of the NASA Global Modeling Assimilation Office (GMAO). For the present work, the GEOS-CHEM was run for the same period and the results have been outputted at 0000 UTC every day as the boundary conditions of NAQPMS. The previous observations showed no significant diurnal variation of ozone at remote sites close to the boundary of modeling domain [Pochanart et al., 2003; Yan et al., 1997], so the 24-h resolution data could approximately represent the boundary situation.

[25] To focus on the seasonal cycle of ozone at the surface, our attention is first paid to how well the NAQPMS model describes the main features over east Asia, especially in east-central China. The ability of the model to simulate ozone in the lower troposphere (below 2 km) was evaluated by the comparison between the global model (GEOS-CHEM), observations at several sites shown in Figure 1 and the regional model (NAQPMS).

3.2. Comparison Between Global and Regional Model

[26] The boundary condition largely affects simulated results for the regional model. Figure 3 shows the evaluation of the global and regional model results by comparing with observations at Waliguan and Mondy, two remote mountain sites close to the boundary of the modeling domain. As shown in Figure 3, in general, the global and regional models capture O3 seasonal features caused by large-scale atmospheric transport and photochemical reactions. The model results of the higher-resolution NAQPMS seem much close to the observation even though the global model used for the boundary conditions overestimates the observed O3 at Mondy from July to December, and at Waliguan from September to January. Sudo et al. [2002] simulated more reasonable seasonal variation at Mondy using another global model (CHASER) with higher resolution of 2.8° × 2.8°.

Figure 3.

Simulated monthly averaged surface ozone variations at Waliguan and Mondy, two remote sites close to the boundary. The simulated (open triangles, regional model; open circles, global model) data at every 0000 UTC are from March 2004 to February 2005. The observed monthly mean data at Mondy are from March 2004 to February 2005. The observations at Waliguan are in 2001.

[27] Figure 4 presented the regional spatial distribution of simulated O3 on near ground (below 2 km) and at 500 hPa in summer by GEOS-CHEM and NAQPMS. It should be noted that only the values at 0000 UTC are compared considering that GEOS-CHEM output the results once a day in the present study. The global and regional models are generally in good agreement in simulating the distribution of O3 near ground except over Tibetan plateau both at near the ground and 500 hPa. As depicted in Figure 3, the comparison with observation at Waliguan, a remote station at about 500 hPa level in Tibetan plateau, shows overestimation tendency of the global model and more reasonable agreement of the regional model.

Figure 4.

Simulated averaged concentrations of O3 at (a and b) the surface and (c and d) 500 hPa in summer 2004 by GEOS-CHEM (Figures 4a and 4c) and NAQPMS (Figures 4b and 4d).

[28] In summer, the highest concentrations of 60–65 ppbv at near the ground are simulated with NAQPMS in east-central China, where emissions rates are high. Maximum values are extended to the downwind region over Bohai Bay according to GEOS-CHEM possibly because of coarse resolution. In winter, NAQPMS simulated ozone concentrations in northern and northeastern China are generally 40–50 ppbv, but is generally below 40 ppbv by GEOS-CHEM (figure not shown). The strong advection and vertical wind shear due to frequent cyclones in winter and spring over the region bring high ozone from the stratosphere. The higher resolution enables NAQPMS to simulate stronger advection from the upper atmosphere, which contributes to higher ozone simulated by the region model.

3.3. Regional Model Versus Observations

3.3.1. Surface Ozone

[29] Because of the coarse grid resolution (81 km) of NAQPMS, the mountainous sites are only several hundred meters in the model. It is not suitable to choose the lowest layer for comparison with the observations. Here we use O3 of the eighth layer (900–1200 m) to match the heights of these mountainous sites. Figure 5 shows the seasonal cycles of surface O3 in both observations and model results. In general, magnitudes and seasonal variations are reproduced by the model reasonably well, with mixing ratios mostly lying within one standard deviation of the measurements. Simulated results of the secondary peak in fall at Mount Tai and Mount Huang is weaker than observations. The coarse horizontal grid resolution (81 km) fails to treat properly a subgrid-scale variability in emission and may result in the underestimation. Further, the emission data used may not reflect properly the seasonality and spatial distribution of biomass burning of agricultural waste in this region, and may have led to the underestimation of the fall peak. Burning crop residue in the field after harvesting, as a way of disposing the waste, is a common practice in China in September–October. Figure 6 shows the hotpots from satellite image over Anhui province in east-central China where Mount Huang is located ( However, these high emissions are not well reproduced by the inventory used in the model.

Figure 5.

Observed mean monthly ozone mixing ratios in ppbv (solid squares), model calculated ozone mixing ratios (open circles) at Japanese sites (Rishiri, Sado-seki, Oki, Yusuhara, Ryori, Yonagunijima), Russian sites (Mondy), a Krygyzstan site (Issyk-kul) and Chinese mountain sites (Mount Tai, Mount Huang, and Mount Hua). Monthly observed ranges with a mean of 1 standard deviation (the whiskers) at some sites (Ryori, Yonagunijima, Mount Tai, Mount Huang, and Mount Hua) are also shown. The locations are given in Figure 1.

Figure 6.

AVHRR satellite image of crop residue burning area over Anhui province in east China during September–October 2004. The burning areas are marked by the red solid cycles. The black dots are Mount Huang.

[30] A series of statistical measures of agreements or disagreements between the prediction and observation [Eder et al., 2006; Monache et al., 2006] are applied here to the ozone concentrations at five sites (Ryori, Yonagunijima, Mount Tai, Huang and Hua) to quantify the model performance. Depending on the measure of interest, bias for example, numerous variations exist each with their own advantages and disadvantages. In this evaluation, two standard and well-accepted measures of model bias are selected: the Mean Bias (MB), and the Normalized Mean Bias (NMB). Likewise, three accepted measures of model error are selected: the Root Mean Square Error (RMSE), Normalized Mean Error (NME) and the Unpaired Peak Prediction Accuracy (UPPA). The results are summarized in Table 1 together with the correlation coefficient(r). Examination of Table 1 reveals that NAQPMS simulates ozone concentrations well, as correlations are about 0.7. The predicted NMB of −8–9% and NME of 16–22% are within the criteria recommended by US EPA, where the values of NMB and NME for a good performance of ozone are ≤15% and ≤30%, respectively. The UPPA of 15–25% indicates a good model performance in high ozone. Figure 7 shows most of simulated ozone concentrations are within a factor of 2 of the corresponding measured concentrations.

Figure 7.

Scatterplots showing the relationship between observed and simulated daily averaged ozone at Ryori, Yonagunijima, Mount Tai, Mount Huang and Mount Hua. The characters are the station name; two lines indicate y = 0.5x and y = 2x.

Table 1. Statistical Summaries of Comparisons of the Model Results With Observations
 Naequation imagebequation imagebrMBNMB, %RMSENME, %UPPA, %
  • a

    N is the number of paired samples.

  • b

    Cm and Co are average concentrations of modeled and observed species.

Mount Tai, ppbv32554580.67−4.2−7.212.416.219.7
Mount Huang, ppbv34348500.68−2.2−4.411.017.018.5
Mount Hua, ppbv19551480.653.16.410.916.224.3
Ryori, ppbv34845410.673.68.98.616.514.9
Yonagunijima, ppbv35638400.69−2.0−5.010.921.624.5

3.3.2. Ozone Soundings

[31] Unfortunately, no long-term ozone vertical profile measurements over China are available for the use of model validation. Instead, here we use the climatologically averaged profile from two long-term ozonesonde stations, Naha (26.2°N, 127.7°E, 27 m a.s.l) and Tsukuba (Taneto) (36°N, 140°E, 31 m a.s.l) in Japan [Logan, 1999]. Figures 8a and 8b present the simulated and observed O3 in four seasons at Naha and Tsukuba, respectively. The model captures the vertical gradients and magnitudes of ozone at both sites, with mixing ratios mostly lying within one standard deviation of the measurements. In summer, the model underestimates tropospheric ozone below the tropopause at Tsukuba, while it overestimates low-tropospheric ozone (below 700 hPa) at Naha. A possible cause of the discrepancy is a limitation to use the climatologically averaged value since the launching frequency is rather low at these stations (1989–1995 for Naha and 1980–1995 for Tsukuba). The relatively high standard deviation in summer means high meteorological variability, which may bring error for the evaluation. In fact, at Yonagunijima (123.02°E, 24.47°N, 30 m a.s.l) and Sadoseki (138.40°E, 38.25°N, 110 m a.s.l) (two regional sites close to Naha and Taneto), the observed surface ozone in the same period is well reproduced by the model as seen in Figure 5.

Figure 8.

(a) Observed (solid squares) and modeled (open circles) ozone profile (ppbv) at Naha (26.2°N, 127.7°E, 27 m a.s.l). The observations are the climatologically averaged value (1989–1995); the model results are during March 2004 to February 2005. (b) Same as Figure 8a but at Taneto (36°N, 140°E, 31 m a.s.l).

3.3.3. Carbon Monoxide

[32] To evaluate the capability of our model in simulating primary pollutants, it would be meaningful to compare modeled carbon monoxide (CO) with observations. However, regional representative long-term CO data are very scarce over China except the measurements in Hong Kong, Lin'an and sampling at Waliguan. Figure 9 shows the time series of monthly mean CO at the WMO GAW sites shown in Figure 1. Figure 9 clearly indicates that the model can reproduce the seasonal variations in CO at most sites. For example, at most sites, the CO mixing ratios have a summer minimum and a winter-spring maximum variation, which is captured well by the simulation. However, the mixing ratios are slightly underestimated at Yonagunijima and Ryori, which is also reported in the study of Tanimoto et al. [2002] suggesting an underestimation of CO emissions in east Asia [Arellano et al., 2004].

Figure 9.

Observed monthly mean CO mixing ratios in ppbv (solid squares) and model calculated CO mixing ratios (open circles) at Plateau Assy, SaryTaukam, Waliguan, Yonagunijima, Ryori and Ulaan Uul. The characters are the station name.

[33] We may conclude that NAQPMS is capable of reproducing observed monthly mean concentrations of surface ozone and its precursor (CO) over east Asia during the simulated period. The reasonable model performance gives some confidence in the model-derived continental attributions.

4. Simulated Results and Discussion

4.1. Seasonal Cycles of Ozone in the Lower Troposphere Over East Asia

[34] Figure 10 presents the simulated seasonal mean concentration of ozone (colored) and its horizontal transport fluxes (vectors) calculated by NAQPMS in the lower troposphere (below 2 km). Figure 11 shows the net photochemical production of ozone in the model domain.

Figure 10.

Seasonal mean ozone mixing ratios (ppbv, shaded) and ozone horizontal fluxes (10−5 mole/m2/s, vector) in the boundary layer (below 2 km) in (a) winter (December–February), (b) spring (March–May), (c) summer (June–August), and (d) fall (September–November).

Figure 11.

Seasonal mean net photochemical production of ozone (ppbv/d, shaded) in the boundary layer (below 2 km) in (a) winter (December–February), (b) spring (March–May), (c) summer (June–August), and (d) fall (September–November).

[35] In winter (December–February), because of the strong zonal pressure gradient between Siberian/Mongolia High and Aleutian Low, a winter monsoon with strong northern wind is predominant in the lower troposphere over east Asia. Consequently, the major outflow transport pathway for O3 to the western Pacific is at 25–40°N (Figure 10a), where the ozone mixing ratios in the lower troposphere remain 45–50 ppbv. Since the solar radiation is weak and ozone net production (ONP) is low (Figure 11a), there is no significant difference between O3 concentration over the high-polluting regions and remote regions. Also Figure 11a shows the net ozone loss due to the titration by nitric oxide over high NOx emission areas (i.e., Beijing, Shanghai, Seoul and Tokyo), in agreement with previous study [Zhang et al., 2002]. Nevertheless, it should be noted that the observed and model simulated concentrations of ozone in east Asia is much higher than the typical concentration in Europe in winter, 20–30 ppb as widely reported [Derwent et al., 1998; Pochanart et al., 2001a; Donev et al., 2002]. This would be ascribed to much weaker effective deposition rate on surface in east Asia compared to Europe in winter due to lower latitude.

[36] In spring (March–May), the O3 concentrations increase over the domain, because of the higher ONP (Figure 11b). For example, O3 rises to 55–60 ppbv in the midlatitudes. Meantime, because of winter monsoon weakness, the transport pathway and the band of high ozone move northward to 30–50°N, and the continental outflow prevails over Japan in this season. The ozone fluxes toward the West Pacific reach the maximum about 3 × 10−5 mol/m2/s in the lower troposphere. This is consistent with the results of Zhang et al. [2003] and Yamaji et al. [2006]. In southern part of China (<25°N and >110°E) O3 slightly falls from winter because of inflow of oceanic air by summer monsoon, which is confirmed by the observations at Hong Kong [Lam et al., 2001]. Meanwhile, it should be noted that ozone concentration in the continental southeast Asia may be underestimated in spring because of underestimate of biomass burning emission used in the model calculation considering the observational data [Pochanart et al., 2001b].

[37] In summer, NAQPMS simulates high ozone (60–65 ppbv) in the north China plain, where pollutant emissions are strong and ozone net production is high (Figure 11c.). It should be noted that the seasonal averaged concentration is close to the ozone standard in China (80 ppbv, Grade I) (State Environmental Protection Administration of China, and many other countries (60 ppbv, in 1 hour or several hours average), including Japan (Ministry of Environment Japan, The concentration is well over the new WMO guideline of 100 μg m−3 (50 ppbv) for 8-h mean [World Health Organization, 2005]. The seasonal average ozone concentrations decrease to below 40 ppbv in the southeastern parts of China and Japan because of the stronger summer monsoon. Under this condition, ONP also decreases to below 4 ppbv/d in the southeast of China (<30°N), much less than ONP (over 15 ppbv/d) in east-central China (Figure 11c). It should be noted, however, that ONP over Japan is higher than 8 ppb/d, which implies that high ozone concentration will occur under the meteorological condition of weak summer monsoon. Figure 10c shows the typical ozone transport pathway moved northward to 40°N, but the amplitude of fluxes become weak. High ozone concentration is also seen over Tibetan plateau. The relative contribution of long-range transport from polluted region and stratospheric influence is still controversial [Zhu et al., 2004; Ma et al., 2005; Ding and Wang, 2006].

[38] In fall, the ozone mixing ratios are 40–50 ppbv except on Tibetan Plateau, where it is about 55 ppbv (Figure 10d). Compared to O3 in summer, ozone concentrations decrease in east-central China, and increase in southeastern China and Japan. As shown in Figure 11d, ONP are nearly the same in east-central China and southern China.

[39] In summary, the distribution of ozone mixing ratios has an apparent seasonal cycle in east Asia. Over the east part of east Asia (>110°E), Continental or marine regional transport and photochemical reactions play key roles in the seasonal cycle.

4.2. Maximum of Monthly Mean O3 in East-Central China

[40] As discussed in sections 2.3 and 4.1, east-central China (region II) shows a high ozone levels with unique seasonal cycle. In the proceeding section, the controlling factors ozone concentration, transport, net photochemical production and deposition, at the three mountain sites have been discussed. Here, every term affecting the ozone budget is evaluated quantitatively on the 0–2.5 km column (above ground level) in an area of 400 km × 400 km around these stations.

[41] From Trainer et al. [2000], the time derivative is the result of all of the chemical and dynamical processes that effect the ozone concentration:

equation image

where O3chem represents the net ozone production; O3loss includes all removal of ozone,(dry deposition is the major process); O3trans represents the sum of the horizontal and vertical transport. In NAQPMS, every term is calculated, on the basis of an internal 5 min time step, and results are presented on an hourly basis.

[42] Table 2 presents the seasonal ozone budgets at the three mountain sites. Our results can be compared with the global model analysis by Mauzerall et al. [2000]. In their assessment of the role of the photochemistry with a 2.8° × 2.8° resolution, they examined the horizontal distribution of ONP over east Asia, and found that BL ONP over east-central China lay within 10–20 and 0–3 ppbv/d in July and January, respectively. Thus our result is comparable to their results except much higher value in summer time Mount Tai in our study.

Table 2. Summary of the Ozone Budgets for Three Mountain Sites in Different Months, Calculated in the NAQPMS Modela
  • a

    Unit is ppbv/d.

  • b

    Transport, Chem and Deposition represent the net effect of transport vertical and horizontal, the net ozone photochemical production, and dry deposition, respectively.

Mount Tai
Mount Hua
Mount Huang

[43] As shown in Table 2, the budgets at these sites indicate east-central China is a strong photochemical source of ozone with maximum ONP during the summertime. Among these sites, Mount Tai (located in the center of east-central China) presents the highest net photochemical production (31.8 ppbv/d) in summer, while Mount Hua and Huang are about 10–15 ppbv/d. On the contrary the contribution of transport is the highest at Mount Huang. A strong variability of ONP and transport among the sites results in the different mechanism of seasonal variation of ozone maximums among the sites.

[44] Mount Tai is a weak net importer and a week photochemical ozone producer in winter (January), and the deposition plays quite an important role in the ozone budget in comparison with their two terms. In April, with increases in solar radiation and temperature, ozone photochemical production of 12.8 ppbv/d has became the most important process in the budget in comparison with the exported and deposition of −6.4, and −2.3 ppbv/d, respectively. In June, the budget clearly indicates that photochemical production (31.8 ppbv/d) around the site has a significant increase and becomes the dominant factor in the maximum of ozone. This would be mainly attributed to the strongest solar radiation and the biomass burning of agriculture waste in the region. Because of the local harvest of winter wheat in this region from June to early July, strong fires activities can be found around Mount Tai [Fu et al., 2007]. The observed fire hot spots counts from satellites totally are over 1000 ( Tang et al. [2003], in their assessment of the influences of biomass burning, also confirmed that the biomass burning air masses emitted from Southeast Asia have very high ozone production efficiency. In comparison with net photochemical production, the transport of 0.8 ppbv/d plays a minor role in the budget. Thus it can be concluded that Mount Tai area is the typical “source region” of ozone in east-central China in spring-summer. In July, a greater net export (−5.0 ppbv/d) and lower net production (25.9 ppbv/d), attributed to the impact of summer monsoon and few biomass burnings, lead to the marked ozone decrease than in June. However, the net production of 25.9 ppbv/d, because of the still high pollutants emissions and temperature, keeps summer ozone to be higher level than winter, unlike with those sites in West Pacific (Oki, Yonagunijima) with stronger import of marine air and much weaker ONP (Figure 2).

[45] With regard to Mount Hua, its ozone budget indicates the net ozone photochemical production dominates in spring and summer In June, although this site are located in an ozone exporter (−1.6 ppbv/d) area, the net production reach to 15.1 ppbv/d (annual maximum), which maximizes ozone level.

[46] Different from the situations at Mount Tai and Hua, ozone budgets at Mount Huang suggest that the net production and transport show comparable importance to ozone maximum in May. In April and May, the net import reaches to 5.9 and 10.1 ppbv/d, close to the magnitudes of net production, 8.9 and 13 ppbv/d, respectively. The increase of fraction of import is partly explained by the impact of emissions in north China plain. In May, the site is located in the interface between north and south air masses. Under this situation, a pollutants transport convergence appears in middle and low troposphere around the site with quite a few pollutants accumulation from anthropogenic sources in the north China plain [Wang, 2000; Yamaji et al., 2006]. In addition, the transport from the higher altitudes resulting from the downdraft at the edge of subtropical pressure also favors the situation.

[47] Compared with the previous observations and model results over western North Pacific [Davis et al., 2003], close enough to mainland China, the net photochemical production at the three mountain sites are higher. For instance, during the PEM-West B carried out from February to April 1994, averaged net photochemical production (daytime flight) for 25–45°N was in the range of 1.0–1.3 ppbv/d [Davis et al., 2003]. Davis et al. [2003] reported mean net photochemical production of 0.7–3.1 ppbv/d for 25–45°N in TRACE-P (February–April 2001) capturing the polluted events. As shown in Table 2, even mean net ozone productions (February–April) at the three mountain sites reach to 2.3–5.8 ppbv/d, which suggests the importance of this region to the whole east Asia.

[48] Finally, net photochemical ozone productions in east-central China can be compared to the results in Europe and North America. An analysis of ozone tendency in air masses arriving Britain and continental Europe calculated by in situ peroxy radical measurements showed that the mean daytime (0600–1900 UT) net ozone productions during summer (June–July 1996) and spring (May 1997) were 13 and 37 ppbv/d, respectively [Salisbury et al., 2002]. Connor et al. [2004] estimated that daytime net ozone production in the boundary layer over central Europe during July/August 2000 was within a range of 13–39 ppbv/d. The model results showed significantly higher net photochemical production in the BL in American Midwest than over central Europe [Connor et al., 2004]. As shown in Table 2, net ozone productions at the three sites in east-central China show a similar magnitude of 10–32 ppbv/d (including daytime and nighttime) in spring and summer. Considering net ozone destruction on nighttime, daytime net photochemical production in east-central China may be higher.

5. Summary and Conclusions

[49] In this study, we have simulated ozone seasonal variation in lower troposphere over east Asia with a regional three-dimensional chemical transport model called NAQPMS, based on measurements at three mountain sites (Mount Tai, Hua and Huang, more than 1500 m a.s.l) newly founded in east-central China and several other sites from EANET and WDCGG. Ozone budgets in east-central China were assessed to investigate causes of ozone maximum at three mountain sites in east-central China and were compared with the reported data in west North Pacific, Europe and North America. This data set in three mountain sites is a unique long-term (12 months) data series of ozone in the upper boundary layer in east-central China and hence it provides an excellent opportunity to perform such a study. Meantime, this is the first modeling work concerning the long-term observed results of regional ozone seasonal cycle over well-known mountain stations in east-central China.

[50] A striking pattern of ozone at three mountain sites in east-central China is two sharp peaks in May–June and September–October. The pattern is distinctly distinguished with that obtained at other rural mountain-top sites over high polluting region in Europe and North America, where O3 showed a single broad peak in May–August.

[51] The regional model performance was compared with a global chemical transport model (GEOS-CHEM), surface observations (including three sites in east-central China) and climatologically averaged ozone sonde profile. Ozone seasonal cycles and CO in lower troposphere over east Asia were generally well reproduced by both the global and the regional models. However, the global model tended to overestimate ozone at Mondy and Waliguan. The regional model underestimates ozone mixing ratios in fall at Mount Tai, possibly because of the underestimation of biomass burning emissions.

[52] Simulated seasonal cycles of ozone and its horizontal transport flux in lower troposphere over east Asia show that high ozone and transport flux are mainly located in the middle latitudes. In winter, there is a typical transport pathway over 25–40°N, with the 45–50 ppbv ozone mixing ratios in the lower troposphere. In spring, the concentrations of O3 from east China to Japan in the north of 25°N reach to 55–60 ppbv and slightly falls for the transport of low ozone from marine air masses in the southern region (<25°N and >110°E). Meantime, the typical transport pathway moves northward to 30–50°N with the maximum of transport fluxes. In summer, NAQPMS simulates high ozone (60–65 ppbv) in the north China plain and over Tibetan Plateau.

[53] Seasonal ozone budgets show that ozone maximums at three mountain sites in east-central China are caused by different mechanisms. At Mount Tai and Hua, net photochemical productions dominate the formation of maximum in June. Differently, the transport and net photochemical productions show comparable importance to ozone maximum in May at Mount Huang.

[54] Finally, a comparison shows that east-central China presents much stronger net photochemical productions than the western North Pacific, which is thought to be close enough to mainland China. Its net photochemical productions are comparable to anthropogenic sources regions in Europe and North America.


[55] The authors would like to acknowledge the local staffs of Mount Tai, Huang and Hua weather stations, the EANET and WDCGG. We also thank R. A. Zaveri for providing the CBM-Z source code and the three anonymous reviewers for their helpful comments and suggestions. Work of IAP has been funded by the Key Project of Chinese Academy of Sciences (KZCX2-YW-205), Natural Science Foundation of China (40305018) and Chinese Ministry of Science and Technology (2005CB422205). The study has been supported by RR2002 grant (MEXT, Japan) and Global Environmental Research Fund, B051 (Ministry of the Environment, Japan). The authors gratefully acknowledge D. Streets for offering emission inventory of east Asia and Alex Bagida for English checking.