Impacts of the East Asian summer monsoon on interannual variations of summertime surface-layer ozone concentrations over China

Introduction Conclusions References

and episodic concentrations of exceeding 100 ppbv (T. Wang et al., 2006b;Duan et al., 2008). Concentrations of O 3 are driven by a combination of precursor emissions and the regional meteorological conditions.
Meteorological parameters during summer in eastern China vary with the East Asian summer monsoon (EASM). The EASM prevails in May-September every year, with strong southerlies bringing clean, warm, and moist air from the oceans to eastern China and rain belts that stretch for thousands of kilometers in the west-east direction in eastern China (Tao and Chen, 1987;Wang and Ding, 2008). Previous observational and modeling studies have shown that such patterns of winds and precipitation of the EASM influence the seasonal variations of O 3 in China (Chan et al., 1998;Li et al., 2007;He et al., 2008;Wang et al., 2011) and in the western Pacific region (Pochanart et al., 2002;Tanimoto et al., 2005;Yamaji et al., 2006). He et al. (2008) analyzed the seasonal variations of O 3 concentrations measured over the period [2004][2005][2006] and found that O 3 concentrations peak in spring and autumn with a summer trough in central eastern China and the western Pacific, the areas that are influenced by clean air from the southern oceans during the summer monsoon. Studies by , Lin et al. (2009), and Zhao et al. (2010) reported that the increasing clouds associated with the EASM rainfall suppress photochemical production of O 3 by altering solar radiation, which also contributes to the minimum O 3 concentrations in summer.
The strength of the EASM exhibits large interannual variations as a result of the interactions between the atmosphere and oceans (Webster et al., 1998). No previous studies, to our knowledge, have systematically examined the impacts of the EASM on interannual variations of summertime O 3 in China. Recently, Zhou et al. (2013) analyzed 2000-2010 ozonesonde data from Hong Kong and found a close link between lower tropospheric O 3 and the East Asian monsoon on interannual scales, but their analyses were focused on O 3 in Hong Kong in spring and autumn. We present here a study to examine the impacts of the EASM on interannual variations of summertime surface-layer O 3 concentrations over China, based on 1986-2006 simulations of O 3 concentrations using the global chemical transport model GEOS (Goddard Earth Observing System)-Chem driven by the assimilated meteorological fields. This work is a companion study to the work of Zhu et al. (2012), which investigated the impacts of the EASM on interannual to decadal variations of summertime aerosols in China.
The GEOS-Chem model and numerical experiments are described in Sect. 2. Section 3 presents simulated interannual variations of summertime O 3 in China. Section 4 shows simulated impacts of the EASM on interannual variations of summertime O 3 , and Sect. 5 examines the mechanisms through which the EASM influences the interannual variations. Section 6 compares the impacts of changing monsoon strength with those of changing anthropogenic emissions on O 3 concentrations in China.

GEOS-Chem model
We simulate tropospheric O 3 using the global chemical transport model GEOS-Chem (version 8.2.1, http://acmg.seas. harvard.edu/geos) driven by the assimilated meteorological fields from the Goddard Earth Observing System of the NASA Global Modeling and Assimilation Office (GMAO). The version of the model used here has a horizontal resolution of 2 • latitude by 2.5 • longitude and 30 vertical layers from the surface to 0.01 hPa. The GEOS-Chem model includes a fully coupled treatment of tropospheric O 3 -NO x -VOC (volatile organic compound) chemistry and aerosol components. Tropospheric O 3 is simulated with about 80 species and over 300 chemical reactions (Bey et al., 2001). Photolysis rates are computed using the fast-J algorithm . The cross-tropopause O 3 flux in this version of GEOS-Chem is specified with the synthetic ozone ("synoz") method (McLinden et al., 2000) as implemented by Bey et al. (2001), which includes a passive, ozone-like tracer released into the stratosphere at a constant rate equivalent to that of the prescribed cross-tropopause ozone flux of 499 Tg O 3 yr −1 .

Emissions
Global emissions of O 3 precursors, aerosol precursors, and aerosols in the GEOS-Chem model follow Park et al. (2003Park et al. ( , 2004, but anthropogenic emissions of NO x , CO, SO 2 , and NH 3 over East Asia are overwritten by the emissions inventory of Streets et al. (2003). Global anthropogenic emissions of nonmethane hydrocarbons are from the GEIA (Global Emissions Inventory Activity) inventory (Piccot et al., 1992). Biomass burning emissions are taken from the GFED-2 inventory (van der Werf et al., 2006). These inventories are then scaled for 2005 on the basis of economic data and energy statistics as described by van Donkelaar et al. (2008). The biogenic emissions in the GEOS-Chem model are simulated using the MEGAN module (Model of Emissions of Gases and Aerosols from Nature; Guenther et al., 2006;Wiedinmyer et al., 2007). Soil NO x emissions are computed using a modified version of the algorithm proposed by Yienger and Levy (1995). Lightning emissions follow Price and Rind (1992), with the NO x vertical profile proposed by Pickering et al. (1998).
The simulations of tropospheric O 3 by the GEOS-Chem model have been evaluated in previous studies for the United States (Fiore et al., 2005;Wu et al., 2008 Wang et al., , 2011Jeong and Park, 2013;Lou et al., 2014). The model was found to be able to capture the magnitude and spatial distribution of O 3 in China.

Experiments
In this study concentrations of O 3 in China for years 1986-2006 are simulated using the GEOS-4 meteorological fields.
To identify the key processes that influence O 3 concentrations in different monsoon years, we perform the following simulations:

East Asian summer monsoon index
The interannual variations in strength of the EASM are commonly represented by the EASM index (EASMI). The EASMI introduced by Li and Zeng (2002) is used in this study. The formulation for calculating EASMI based on the GEOS-4 meteorological parameters was given in Zhu et al. (2012). Positive values of EASMI indicate strong monsoon years whereas negative values indicate weak monsoon years. Physically, a strong summer monsoon in China is characterized by strong southerlies extending from southern China to northern China, a deficit of rainfall in the middle and lower reaches of the Yangtze River, and large rainfall in northern China. On the contrary, in a weak summer monsoon year, China experiences weak southerlies, large rainfall in southern China, and a deficit of rainfall in northern China. The movement of the rain belts is associated with the strength of the southerlies.  , and those at the Ryori site are from the WMO (World Meteorological Organization) World Data Center for Greenhouse Gases (WDCGG, http: //http://ds.data.jma.go.jp/gmd/wdcgg/). At Hok Tsui, simulated O 3 concentrations are higher than the observations in JJA. This discrepancy may be due to the model's overestimation of O 3 in marine boundary layers in summer . Interannually, the model captures well the peaks and troughs of the observed JJA O 3 concentrations, with a high correlation coefficient of +0.87. The model underestimates JJA O 3 concentrations at Ryori, probably due to the uncertainties with local emissions, but captures mostly the years with maximum or minimum O 3 levels with a correlation coefficient of +0.47. The simulated APDM values at Hok Tsui and Ryori are both 7 %, smaller than the observed interannual variations of 22 and 8 % at these two sites, respectively, which can be attributed to the fixed anthropogenic emissions of O 3 precursors in our O 3 _TOT simulation.

Impacts of the EASM on interannual variations of summertime O 3 in China
The simulated JJA surface-layer O 3 concentration averaged over all of China (defined by the national borders of China) is shown in Fig. 3a    demonstrate that the EASM strength has large impacts on JJA O 3 concentrations over China.
In order to quantify the impacts of the EASM on O 3 concentrations over China, we show in Fig. 4a and b, respectively, the absolute and percentage differences between O 3 concentrations averaged over the five weakest EASM years (1988, 1989, 1996, 1998, and 2003) and those averaged over the five strongest EASM years (1990, 1994, 1997, 2002, and 2006). These weakest (or strongest) monsoon years are selected within years 1986-2006 based on the five largest negative (or positive) values of the normalized EASMI as shown in Fig. 3a. Relative to the concentrations in the strongest monsoon years, O 3 levels in the weakest monsoon years are lower over almost all of China, with the largest reductions exceeding 3 ppbv (or 6 %) in northeastern and southwestern China and over or near the Tibetan Plateau. Concentrations of O 3 in the weakest monsoon years are simulated to be higher than those in the strongest monsoon years by 3-5 ppbv (or 6-15 %) over the East China Sea. The pattern of the differences in O 3 concentrations agrees with the spatial distribution of the correlation coefficients (Fig. 3b). Averaged over China, the O 3 level in the weakest monsoon years is lower than that in the strongest monsoon years by 2.0 ppbv (or 4 %). Note that the monsoon region covers almost all of China except the northwest (Gao, 1962;An et al., 2000). Our simulated monsoon-induced changes in O 3 in China (Fig. 4) are mostly within the monsoon region. The changes in O 3 over Siberia are large, which can be explained by the anomalous northerlies over the north border of China between the five weakest and strongest EASM years (weakest-strongest) (Fig. 5) that transport O 3 to China.    exhibit increases in the weakest monsoon years, with maximum increases of 4-7 ppbv (3-6 %) over 80 and 130 • E, as a result of the anomalous horizontal convergence at these layers.

Impacts of the EASM on transboundary transport of O 3
Considering that O 3 concentrations in almost all of China are lower in the weakest monsoon years than in the strongest monsoon years and that tropospheric O 3 has a relatively long lifetime of 3-4 weeks (Seinfeld and Pandis, 2006), we examine firstly the impacts of the EASM on transboundary transport of O 3 . Figure 5 shows JJA horizontal winds at 850 and 500 hPa averaged over years 1986-2006 as well as the composite differences in JJA horizontal winds between the five weakest and five strongest EASM years at these two layers. The typical features in winds during the EASM can be seen in Figs. 5a and b. The southerlies prevail in southeastern China in the lower and middle troposphere in JJA. Figure 5c and d also present the differences in winds between the weakest and strongest EASM years (weakest minus strongest). In JJA, anomalous southerlies are found in southern China and anomalous westerlies are found in southeastern China, as the winds in the five weakest monsoon years are compared with those in the five strongest monsoon years. Such differences in winds between weak and strong EASM years were also reported in Li and Zeng (2002) and Huang (2004). The differences in winds in different monsoon years lead to differences in transboundary transport of O 3 . We show in Fig. 6 the differences in simulated horizontal mass fluxes of O 3 at the four lateral boundaries of the selected box (85-120 • E, 20-46 • N, from the surface to 250 hPa) and in Table 1 the summary of the composite analysis on fluxes of O 3 in and out of this box, based on simulation O 3 _TOT. This box is selected to capture the features of transboundary O 3 transport that influence O 3 concentrations in China. The location of the box is shown in Fig. 5d. Relative to the strongest monsoon years, the weakest monsoon years have less inflow by 0.1 Tg season −1 at the west boundary, larger inflow fluxes of O 3 by 4.2 Tg season −1 at the south boundary and by 5.5 Tg season −1 at the north boundary, and larger outflow by 12.9 Tg season −1 at the east boundary (Table 1, Fig. 6), as mass fluxes are summed over JJA. The net effect is a larger transboundary outflow of O 3 by 3.3 Tg season −1 in the weakest monsoon years than in the strongest monsoon years. The anomalous westerlies in southeastern China are especially important, which bring polluted air with high O 3 concentrations to the coastal areas and the East China Sea (Fig. 6d), leading to reductions in O 3 concentration in China. The role of changes in transboundary transport of O 3 can be further quantified by the simulation O 3 _TB with natural and anthropogenic emissions of O 3 precursors in China turned off. The spatial distribution of O 3 from simulation O 3 _TB (referred to as TBO 3 hereafter) is presented in Fig. 7a. TBO 3 concentrations show a distinct spatial gradient over China, decreasing from about 55 ppbv in northwestern China to about 10 ppbv in southeastern China. The JJA TBO 3 concentration averaged over the whole China is 30 ppbv, which is about 62 % of the average concentration simulated in O 3 _TOT, indicating that a significant fraction of surface-layer O 3 over China is from transboundary transport. The differences in TBO 3 concentrations between the weakest and strongest monsoon years (Fig. 7d) are, to a large extent, similar to those from simulation O 3 _TOT (Fig. 4a), in terms of both distributions and magnitudes. Averaged over China, the difference in TBO 3 between the weakest and strongest monsoon years is −1.6 ppbv, which accounts for 80 % of the corresponding difference obtained in O 3 _TOT. The JJA mass fluxes of TBO3 for the selected box (85-120 • E, 20-46 • N, from the surface to 250 hPa) are also similar to those simulated in O 3 _TOT (Table 1), with a net horizontal outflow of 2.8 Tg season −1 . These model results indicate that the differences in transboundary transport of O 3 is a dominant mechanism through which the EASM influences interannual variations of JJA O 3 concentrations in China.  (1988, 1989, 1996, 1998, and 2003) and five strongest monsoon years (1990, 1994, 1997, 2002, and 2006), and the differences are calculated as (weakest-strongest

Impacts of the EASM on vertical transport of O 3
Vertical circulations in China have some unique features during the EASM. In summer, two major ascending branches of winds (or strong convections) are observed throughout the entire troposphere (Fig. 8a). One branch is located over the Yangtze River valley, associated with the Mei-yu front (rain belt) of the EASM (Ding and Chan, 2005). The other branch is over the Tibetan Plateau, because the Tibetan Plateau during summer serves as a large heat source, which uplifts heated air to the upper troposphere and even to the stratosphere (Ye, 1981).  Figure 8b shows the differences in JJA vertical wind velocity and simulated vertical mass flux of O 3 at 500 hPa between the weakest and strongest monsoon years. Relative to the strongest monsoon years, anomalous convections are found over central and western China in the weakest monsoon years, leading to enhanced upward transport of O 3 from the surface to upper troposphere at these locations. The differences in vertical winds shown in Fig. 8b agree with those reported in Huang (2004). Table 1 also summarizes the composite analysis on vertical fluxes of O 3 through the top side of the selected box (85-120 • E, 20-46 • N, from the surface to 250 hPa). In simulation O 3 _TOT, the upward flux of O 3 through the top plane of the box in the weakest monsoon years is larger than that in the strongest monsoon years by 1.4 Tg season −1 . This anomalous vertical outflow of 1.4 Tg season −1 is smaller than the anomalous horizontal transboundary outflow of 3.3 Tg season −1 (Sect. 5.1), indicating that the differences in vertical transport of O 3 also contribute to lower JJA O 3 concentrations in China in the weakest monsoon years than in the strongest monsoon years, but the impact of the differences in vertical transport is smaller than that of the differences in transboundary transport of O 3 .

Impact of cross-tropopause transport on surface-layer O 3 concentrations
The cross-tropopause transport of O 3 from the stratosphere is an important source of tropospheric O 3 . Simulation O 3 _ST is performed to quantify the impact of cross-tropopause transport on JJA surface-layer O 3 concentrations. The simulated JJA concentrations of O 3 in simulation O 3 _ST (referred to as  McLinden et al. (2000) to represent O 3 in the stratosphere, in which the ozone vertical profiles across the tropopause are relaxed back toward climatological profiles and hence cross-tropopause O 3 flux varies with time step and location. We have tested using linoz instead of synoz and found that these two schemes obtain same conclusion about the impact of cross-tropopause O 3 on JJA surface-layer O 3 in China.

Impacts of the EASM on local chemical production of O 3
The  Figure 7f presents the differences in simulated LOCO 3 between the weakest and strongest EASM years. Relative to the strongest monsoon years, LOCO 3 concentrations in the weakest monsoon years are lower by 2-5 ppbv over southern China and slightly higher by up to 1 ppbv over central China. Averaged over China, the difference in LOCO 3 between the weakest and strongest monsoon years is −0.4 ppbv, which accounts for 20 % of the corresponding difference in TOTO 3 (O 3 concentrations in simulation O3_TOT) and hence reflects the small impacts of monsoon strength on local chemical production of O 3 . The small impacts of monsoon strength on local chemical production of O 3 can be further justified by examining the net chemical production of O 3 within the selected box (85-120 • E, 20-46 • N, from the surface to 250 hPa). The sum over the selected box, the net chemical production (chemical production − chemical loss) averaged over the weakest monsoon years, is 39.0 Tg season −1 in JJA, which is smaller than that averaged over the strongest monsoon years by 0.5 %.  by 111, 56, and 9 %, respectively, leading to increases in JJA surface-layer O 3 by 9-15 ppbv in southeastern China (Fig. 9). Note the locations of large increases in O 3 here are different from those of large differences in O 3 between the weakest and strongest monsoon years (Fig. 4b). Averaged over China, the change in JJA surface-layer O 3 concentration owing to changes in emissions over years 1986-2006 is +5.3 ppbv, which is larger than the difference in the JJA surface-layer O 3 of 2.0 ppbv between the selected weakest and strongest monsoon years. However, the difference in surface-layer O 3 between the weak monsoon year 1998 and the strong monsoon year 1997 is −4.2 ppbv, indicating that the impacts of the EASM on JJA O 3 can be particularly strong for certain years.

Conclusions
We examine the impacts of the East Asian summer monsoon With fixed anthropogenic emissions, simulated JJA O 3 concentrations averaged over China exhibit strong positive correlation (with a correlation coefficient of +0.75) with the East Asian summer monsoon index (EASMI) in the time period of 1986-2006, indicating that JJA O 3 concentrations are lower (or higher) in weaker (or stronger) EASM years. Relative to JJA surface-layer O 3 concentrations in the strongest EASM years (1990, 1994, 1997, 2002, and 2006), O 3 levels in the weakest EASM years (1988, 1989, 1996, 1998, and 2003) are lower over almost all of China with the national mean O 3 concentration lower by 2.0 ppbv (or 4 %).
Sensitivity studies are performed to identify the key processes through which the variations in EASM strength influence interannual variations of JJA O 3 in China. The difference in transboundary transport of O 3 is found to be the most dominant factor that leads to lower O 3 concentrations in the weakest EASM years than in the strongest EASM years. Relative to the strongest EASM years, the weakest EASM years have less inflow by 0.11 Tg season −1 at the west boundary, larger inflow fluxes of O 3 by 4.2 Tg season −1 at the south boundary and by 5.5 Tg season −1 at the north boundary, and larger outflow by 12.9 Tg season −1 at the east boundary, as horizontal mass fluxes of O 3 at the four lateral boundaries of the selected box (85-120 • E, 20-46 • N, from the surface to 250 hPa) are calculated. As a result, the weakest EASM years have larger outflow of O 3 than the strongest EASM years, which, together with the enhanced vertical convections in the weakest EASM years, explain about 80 % of the differences in surface-layer O 3 concentrations between the weakest and strongest EASM years.
We also perform a sensitivity simulation O 3 _EMIS to compare the impact of changing monsoon strength with that of changing anthropogenic emissions on JJA O 3 concentrations. Averaged over China, the change in JJA surface-layer O 3 concentration owing to changes in emissions over years 1986-2006 is +5.3 ppbv, which is larger than the difference in JJA surface-layer O 3 of 2.0 ppbv between the selected weakest and strongest monsoon years. However, the difference in surface-layer O 3 between the weak monsoon year 1998 and the strong monsoon year 1997 is −4.2 ppbv, indicating that the impacts of the EASM on JJA O 3 can be particularly strong for certain years. Note that while the largest increases in O 3 by anthropogenic emissions are located over southeastern China, the largest impacts of EASM on O 3 are found over central and western China.