The Concentrations, Formations, Relationships and Modeling of Sulfate, Nitrate and Ammonium (SNA) Aerosols over China

Sulfate, nitrate and ammonium (SNA) are the dominant composition of secondary aerosols in the atmosphere and have a significant impact on public health, atmospheric chemistry process and climate. In this study, to evaluate SNA pollution in China, a first nationwide investigation derived from almost all published data in the field measurement before 2012 was carried out. The results show that SNA levels in China are about 3–5 times higher than those in USA and Europe. SNA account for 34.2 ± 10.9% in PM2.5 and 28.2 ± 8.5% in PM10. The highest SNA concentrations occur in urban areas of northern China. SNA all have peaks in winter, but the nadirs are in spring for sulfate and ammonium and in summer for nitrate. SOR (sulfur oxidation ratio) and NOR (nitrogen oxidation ratio) values show that the formations of sulfate and nitrate are distinct in different regions and seasons. The low average NO3/SO4 ratio (0.43 ± 0.26) indicates that the stationary emissions from coal combustion remain the main sources. There is a good relationship between (2[SO4] + [NO3]) and [NH4] with near 1 slope, signifying that (NH4)2SO4 and NH4NO3 are the predominant forms which SNA exist in particles in China. Based on the comprehensive observational data in China, the simulation for SNA aerosols by GISS in CMIP5 were evaluated.


INTRODUCTION
Atmospheric particles are complex and consist of all kinds of chemical species which are directly from the natural and anthropogenic emissions or formed as the secondary components (Xu et al., 2012).Sulfate, nitrate and ammonium (abbreviated as SNA) are the dominant secondary species of aerosols (~30-50% in PM 2.5 ).They can determine the chemical characteristic of particle by changing particles' hygroscopicity and acidity, influence the optical properties of particles and the earth's radiation balance by scattering and absorbing solar radiation, change cloud nucleation process by acting as cloud condensation nuclei (CCN) and then affect climate, and the most important, penetrate deeply to human lung and deteriorate human health (Xu et al., 2002;Sun et al., 2004;Cheung et al., 2005;Hu et al., 2008;Pathak et al., 2009;Cheng et al., 2011;Sang et al., 2011;Gao et al., 2012;Yan et al., 2012).
Particle pollution is an increasing concerned issue in current China.SNA have a high loading in the atmospheric particle (See section Contributions to particles) and play a crucial role on the deteriorated air quality, e.g., in the increasing haze formation in the country (Tan et al., 2009;Du et al., 2011;Huang et al., 2011;Zhang et al., 2012c).As the dominant species, SNA have been widely investigated in many regions in recent years, including various urban and suburban/rural sites as shown in Tables S1 and S2, and a lot of SNA characteristics have been gained.However, most of these researches are associated with the specific SNA levels, formation and sources at specific locations.Few are related to the SNA's characteristics in a national scale.Some studies have paid attention to the description of SNA over the country, but they were usually limited to the model simulation (Wang et al., 2003;Li et al., 2010b).Therefore, up to now, the nationwide SNA pollution characteristic from the ground observation is still blank.
In this study, we investigated the levels, formations and sources, and inter-relationships of SNA across the country based on published data in the field measurements before 2012.To our knowledge, this is the first time for such wide and systematic inspection on SNA in China.This may give a holistic understanding of SNA pollution and benefit the further control.Besides, a model simulation was performed and the results were evaluated based on the observations, which provides an insight into differentiating the observation and model simulation and is helpful in the evaluation of SNA's radiative forcing in China.

Database
There are 226 sets of data about SNA in PM 2.5 and 88 sets in PM 10 shown in Tables S1 and S2, which are collected from almost all published literatures before 2012 in the field measurements.These data span from 1997 to 2010 with long term (i.e., annual) and short term (i.e., daily and weekly) averaged concentrations.They are measured at various sites, including urban (i.e., commercial, traffics, industrial and residential sites) and suburban/rural areas (including suburban, rural, park, background, mountain, forest and remote sites).Data with only species without PM mass concentrations or with PM mass without any species concentration in a set (e.g., with PM, SO 4 2-and NO 3 -without NH 4 + ) are not collected, which guarantees the internal consistency of a data set and a minimum of quality control.Northern and southern China are divided by a demarcation line, Qinling Mountains-Huaihe River Line (Gao et al., 2011a) (see supplementary).There are 103 sets of data in PM 2.5 in 13 regions of northern China and 123 sets in 18 areas of southern China.In this study, northern and southern China usually refer to aforementioned designation, unless otherwise specified.

Model Simulation
To evaluate the climate model prediction of aerosols in China, we retrieved aerosol concentrations during 2000-2012 from the Goddard Institute for Space Studies (GISS) atmospheric general circulation model (GCM,Model E2 version) in Coupled Model Intercomparison Project Phase 5 (CMIP5) (Hansen et al., 2002a, b;Schmidt et al., 2006;Schmidt et al., 2014).The GISS_E2 simulation has a resolution of 2° × 2.5° latitude by longitude in horizontal and vertically spans 40 layers with top at 0.1 hPa.The simulation included interactive chemistry, aerosols, and dust driven by emissions (Koch et al., 2011), The descriptions of the tracer code, chemical reactions, species, and model evaluation are detailed in the literatures (Shindell et al., 2006(Shindell et al., , 2013)).

Representativeness of the Collected Data
The data in this study from various papers through different techniques at different sites in China over one decade were analyzed based on the supposition that these data accurately represented the characteristics of particles and SNA during sampling time.The comparisons neglected the temporal variations of the data.
Particle concentrations were obtained by gravimetric measurements.The possible systematic errors may come from the different sampler types, various filters (with Teflon, Quartz, Glass fibre filters) (see Tables S1 and S2) and the relative humidity when weighting the samples before and after sampling.The sampler types usually cause a little error since the filtration devices (including sampling heads) were equivalent to reference methods in the published literatures.The inert filters may result in the variable reduction of NH 4 NO 3 at different temperature.When the temperature is over 20°C.the losses of NH 4 NO 3 can reach 100% from quartz fibre filters (Hering and Cass, 1999;Schaap et al., 2004).This, combined with both negative and positive artefacts for organic matter in particles (Eatough et al., 1993(Eatough et al., , 1996)), may rise the errors of particle mass concentrations.Notably. the losses of NH 4 NO 3 during sampling can affect greatly the seasonal variation of NO 3 -, which will be pointed out in the next discussion when the artefacts may influence the conclusions.There were small errors (< 10%) from the different relative humidity through gravimetric measurements (Putaud et al., 2004), so it is negligible in the following discussion.
In the SNA analysis, we considered the accuracy and consistency of analytical techniques, turning out all SNA in our referenced papers were analyzed by ion chromatograph (IC).According to (Putaud et al., 2000), the results from different laboratories are usually within ± 10% for SNA.

Contributions to Particles
Total fractions of SNA in PM 2.5 and PM 10 were 34.2 ± 10.9% and 28.2 ± 8.5%, respectively, in the country (Fig. 1).In PM 2.5 , sulfate (19.5 ± 7.6%) contributes more to particles than nitrate (7.3 ± 3.6%) and ammonium (7.3 ± 3.3%).The corresponding fractions of the three species in PM 10 are 15.6 ± 6.2%, 7.6 ± 3.0% and 5.1 ± 2.4%, respectively.Notably, the fractions of nitrate in PM 2.5 and PM 10 are comparable.Considering that the mass concentration of PM 10 is higher than that of PM 2.5 (175.28 ± 118.94 µg m -3 vs. 99.63 ± 54.69 µg m -3 in Tables S2 and S1), this signifies that nitrate in coarse particles has a certain contribution to PM 10 mass and can't be ignored (Li et al., 2008;Yan et al., 2012), which is consistent with the results from Putaud et al. (2004) in Europe.Compared to those in USA (49.2%) (Hand et al., 2011) and Europe (38.7%)(Sillanpää et al., 2006), the total SNA fraction in fine particle in China is lower (Figs. 1(a), 2(c) and 2(d)).The contributions of each SNA species to fine particle in China are lightly lower than those in Europe (Sillanpää et al., 2006).However, the proportion of sulfate in China is near that in USA (Hand et al., 2011) and nitrate and ammonium are about a half of those in USA (Hand et al., 2011).

Spatial and Temporal Distributions Seasonal Distributions
From Table 1, the concentration trends of SNA in China are winter > summer > autumn > spring, winter > autumn > spring > summer, winter > autumn > summer > spring, respectively.Peak concentrations occur consistently in winter for SNA in China, but the lowest values are in spring for sulfate and ammonium and in summer for nitrate, which are attributed to the distinct climate and source emissions.The high sulfate concentration in winter may be attributed to the heavy SO 2 emissions and primary sulfate emissions (see Section Formation of sulfate and nitrate aerosols from gaseous precursor conversion), e.g., the increasing emissions from the extra coal burning for heating in the northern China, which also resulted in the highest sulfate concentration in the cold season in the north.The low rate of removal via less wet deposition in this season as shown in Table 2 is also an important factor to contribute higher sulfate level.In summer and autumn, the temperature and RH are relative high (Table 2), and there are higher SO 2 emissions in late autumn, which will activate the photochemical reactions and result in higher sulfate concentrations in the two seasons (Hu et al., 2002;Sun et al., 2004).In comparison, in spring, RH and temperature are lower (Table 2) and there are more violent winds, which can inhibit the formation of sulfate (Yang et al., 2007).In northern China, sulfate adopts the same pattern as the whole country.While in southern China, the level of sulfate in autumn is comparable to that in winter and higher than that in summer, which is different from that in northern China with the lower value in autumn than those in winter and summer.This may be attributed to the absence of both the pronounced SO 2 emissions from extra coal combustion in winter and obviously higher temperature in summer than in autumn in the south.Nitrate exhibits an obvious seasonal variation with the highest level in winter and the lowest in summer in northern China, southern China and the whole country.It should be pointed out that the losses of NH 4 NO 3 are huge during the sampling and may reach 100% from quartz fibre filters in summer (Hering and Cass, 1999;Schaap et al., 2004).When we multiply the measured nitrate concentrations in summer by 2 to reevaluate its pattern in season, it is found that even so, the nitrate levels in winter are still the highest in northern China, southern China and the whole country.It is well known that nitrate concentration is affected by NO 2 emissions and the meteorological factors (such as temperature).NO 2 is mainly from the vehicular emissions and coal burning emissions.Due to the almost even distribution of traffic emissions over the year round, the variations of nitrate may be strongly associated with the coal burning and the meteorological factors (Wang et al., 2006b;Zhang et al., 2009b).In the whole country, NO x emissions are slight higher in winter than those in other three seasons due to the extra emissions from coal burning for heating in northern China.The relative high NO x emissions, the less wet precipitation and favorable shift from gas phase to particle phase as a result of lower temperature in winter jointly lead to high nitrate concentration.In summer, NO x emissions are relative low and temperature is high, which benefits the conversion of gaseous NO 2 to HNO 3 , but is adverse for gaseous HNO 3 to particulate nitrate.These, combined with high wet precipitation, probably result in the low nitrate concentration.It should be pointed out that the artefacts from the volatilization of NH 4 NO 3 on the filter during the sampling in summer should also have a significant contribution to the lower nitrate concentration.In northern China, in winter, NO x emissions are heavier due to the heating (Table 1), which, combined with the low temperature, which is favorable for the shift from gaseous HNO 3 to particulate nitrate, causes the high nitrate concentration (Yang et al., 2007).In comparison, in southern China, the higher nitrate concentration in winter may be attributed to lower temperature and less wet precipitation than other seasons (Table 2).In summer, whether northern or southern China, the same reasons as above mentioned result in the lowest nitrate concentrations.
The seasonal pattern of ammonium resembles sulfate with the highest concentration in winter in the country due to the preferred reaction of NH 3 with H 2 SO 4 more than with HNO 3 (Tang et al., 2004).The conversion of NH 3 to ammonium aerosol relies on the concentrations of sulfate and nitrate in the atmosphere, temperature and available water, as well as the NH 3 level (Zhang et al., 2011).In winter, the more neutralization reactions of NH 3 with SO 2 and NO x , lower temperature and higher acidic species favor the shift from the NH 3 (g) to (NH 4 ) 2 SO 4 (s) and NH 4 NO 3 (s) (He et al., 2001;Wang et al., 2006a;Qu et al., 2008).In contrast, the higher temperature in summer is unfavorable for the conversion of NH 3 to ammonium aerosol.Meanwhile, the sampling artifact from the NH 4 NO 3 composition on quartz fibre filter at relative high temperature should also have a certain contribution to the relative low ammonium concentration.In northern China, the ammonium follow completely the seasonal pattern of sulfate in the region.While in southern China, ammonium levels in autumn and winter are higher than those in spring and summer.The significant low concentration in summer is probably related to the losses of NH 4 NO 3 in the sampling and the shift of particulate ammonium to gas NH 3 .

Spatial Distributions
Fig. 2 exhibits spatial variations of sulfates, nitrates and ammonium in PM 2.5 .From the figure, high SNA concentrations occur in North China Plain (e.g., Beijing), Yangtze River delta (e.g., Shanghai) and Sichuan Basin (e.g., Chongqing), which is in accord with the results from Wang et al. (2013) and Li et al. (2010b).This is probably related to their higher precursors emissions derived from the extra coal consumption for collective heating in cold season in northern China, rapid economic development and high population density in Yangtze River delta and the terrain unfavorable for the air diffusion in Sichuan Basin.Low SNA levels are observed at sparsely populated mountain, forest and background sites, such as Jianfengling (Li et al., 2010a), Mt.Tai (Deng et al., 2010) and Hok Tusi (Lai et al., 2007).The maxima in urban are 60.90, 35.60 and 29.80 µg m -3 for SNA, respectively, may reach 28, 274 and 298 times of the minima at mountain, forest and background sites, suggesting a vast inhomogeneity in the distribution.
Besides this, the SNA distributions in PM 2.5 in urban and rural areas are also different.SNA show the highest concentrations in urban sites with the averages of 19.83, 8.70 and 8.03 µg m -3 , then slightly low levels in suburban sites with the mean values of 14.25, 5.97 and 5.46 µg m -3 , which are comparable with those (15.01,5.44 and 6.52 µg m -3 for SNA, respectively) in rural sites.The lowest SNA concentrations occur in background sites with the percentages of 59.7% (11.84 µg m -3 ), 37.6% (3.27 µg m -3 ) and 47.7% (3.83 µg m -3 ) of SNA in urban areas.Fig. 3 shows SNA patterns in PM 2.5 in urban and rural areas (including suburban, rural, background, mountain, and forest sites in Table S1) in northern and southern China.Obviously, SNA are higher in northern China, especially the urban areas, than those in southern China.This may be contributed two reasons: firstly, the emissions of precursors of SO 2 and NO x in the north are much higher than that in the south (e.g., 12.40 Tg in the north versus 9.28 Tg in the south for SO 2 emissions and 13.50 Tg in the north versus 10.22 Tg in the south for NO x emissions) (China Statistical Yearbook, 2002-2013) due to more heavy industries and collective heating in winter in the north; secondly, the subtropical high pressures, typhoon, heavy monsoon rain (e.g., the annual average wet precipitation: 7995.4 mm in the north versus 18693.1 mm in the south (China Statistical Yearbook, 2002-2013) in the south favor the removal and dispersion of SNA.In the comparisons between urban and rural areas, it is found that SNA concentrations are higher in urban than those in rural areas in northern China.This may be attributed to more emissions of industry, vehicle exhaust and construction activities caused by dense residents in cities to elevate SNA levels (Sun et al., 2004;Wang et al., 2009;He et al., 2011;Zhang et al., 2012c).While in the south, the SNA levels are comparable between in urban and rural areas.This is probably due to the consistency of emission sources and/or the transportations of pollutants from urban to suburban/rural sites.

Formations and Sources Formation of Sulfate and Nitrate Aerosols from Gaseous Precursor Conversion
It is well known that sulfate can be produced in the atmosphere by two ways: gaseous phase oxidation and aqueous phase reaction.In the gaseous phase, SO 2 can react with OH radical followed by condensation of H 2 SO 4 to form sulfate.In comparison, the aqueous phase reaction is more important due to forming the majority of sulfate in most circumstances (Langner and Rodhe, 1991;Guo et al., 2012;Harris et al., 2013), which includes the oxidation of SO 2 with H 2 O 2 /O 3 and transition metal in cloud and fog droplets and the heterogeneous reactions of SO 2 on aerosol surfaces (Deng et al., 2010).Sulfur oxidation ratio (SOR), which is defined as S nss-SO4 2-/(S nss-SO4 2-+ S SO2 ), is a measure of the oxidation degree of sulfur.When SOR is smaller than 0.10, sulfate can be considered from the primary emissions, otherwise, it is produced through the oxidation of SO 2 (Wang et al., 2011).Similarly, nitrate is generally produced by the oxidation of nitrogen oxides and then reacts with NH 3 via homogeneous and heterogeneous reactions (Zheng et al., 2008;Ye and Chen, 2009;Zhang et al., 2009b;Ma et al., 2011;Lei and Wuebbles, 2013).Apart from the photochemical oxidation of NO x to HNO 3 , the formation of nitrate in particle is also influenced by the temperature dependent equilibrium between gaseous phase HNO 3 and particulate nitrate: NH 3 (g) + HNO 3 (g) ↔ NH 4 NO 3 (s) (Zhou et al., 2009).Nitrogen oxidation ratio (NOR), which is defined as N NO3 /(N NO3 + N NO2 ), is usually used to express the oxidation degree of nitrogen (Zhang et al., 2011).Higher SOR and NOR values suggest more secondary aerosols produced by oxidation of gaseous species in the atmosphere.
SOR and NOR values are calculated in this investigation based on almost all published literatures before 2012 with four seasons as shown in Table 3. From the table, in northern China, SOR is significantly larger in summer than those in other three seasons.The gaseous phase oxidation of SO 2 to sulfate by OH radical has a strong dependence on temperature (Wang et al., 2006b); while high RH favors the aqueous heterogeneous formation of sulfate.In summer, both temperature and RH are high (Table 2), which may enhance the formation of sulfate by gaseous and aqueous phase reactions, so there are relative high SOR values.In winter, the SOR values are lower, even less than 0.1 in some areas, suggesting the conversions of SO 2 to sulfate through neither gas nor aqueous phase are active in the low temperature, and primary emissions from coal combustion become the principal sources.In southern China, the SOR values are comparable in four seasons with all larger than 0.1 (Wang et al., 2006b;Xu et al., 2012;Zhang et al., 2012a), suggesting SO 2 has active photochemical reactions in the atmosphere.Considering that when RH is larger than 75%, sulfate is largely from heterogeneous reactions of photochemical oxidation (Wang et al., 2006b), the average RH of 69.98-75.90% in southern China (Table 2) means the enhancement of sulfate formation is probably more related to heterogeneous reactions.The comparable SOR in four seasons may be attributed to the warm and humid weather throughout the whole year in the region.
NOR values are lower than SOR values (Table 3).NOR values are the highest in summer or in winter in Beijing (Wang et al., 2005;Deng et al., 2011) and Ji'nan (Gao et al., 2011b) of northern China, suggesting that in different samplings, either the oxidization of NO x to HNO 3 , e.g., in summer, or the equilibrium between gaseous phase HNO 3 and particulate nitrate, e.g., in winter, determines the particulate nitrate formation.In contrast, in the two cities of southern China, Xiamen (Zhang et al., 2012a) and Fuzhou (Xu et al., 2012), the highest NOR values occur in winter, meaning the equilibrium between gaseous phase HNO 3 and particulate nitrate is dominated in the particulate nitrate formation.The high temperature and RH in summer are favorable for the conversion of NO x to HNO 3 , but also promote the shift of NH 4 NO 3 from particulate phase to gaseous phase.Moreover, the higher the temperature is, the larger amount H 2 SO 4 produced from SO 2 oxidation, which competes with HNO 3 for NH 3 , slowing the production of nitrate.These, along with the emissions of NO x (Table 2), determine the formation of nitrate in the particulate phase is higher in summer or winter in northern China and in winter in southern China.Notably, the artefacts from the losses of NH 4 NO 3 in the sampling in summer can reduce the NOR values and probably have a significant effect on the season trends of NOR.

The Main Sources, Coal Combustion or Vehicle Exhaust Emissions?
The mass ratio of nitrate to sulfate has been used as a tracer for the relative importance of mobile vs. stationary sources of nitrogen and sulfur in the atmosphere (Wang et al., 2005;Lai et al., 2007;Shen et al., 2007;Cao et al., 2009;Li et al., 2009).It has been indicated that high nitrate to sulfate ratios will be observed when the influences of vehicular emissions surpass those from the coal combustion emissions (Yang et al., 2011).In China, the estimated NO x to SO x ratios from gasoline burning, diesel burning, and coal combustion are 13:1, 8:1 and 1:2 (Yao et al., 2002;Wang et al., 2005), respectively, thus it is reasonable to use nitrate and sulfate as the indicators of mobile and stationary emissions, respectively.
The mass ratios of NO 3 -/SO 4 2-are 0.03-2.14 with an average of 0.43 ± 0.26 in China, which is lower than those in America (0.63) (Hand et al., 2011) and comparable to Europe (0.47) (Sillanpää et al., 2006).The ratio is less than 1 and suggests that stationary sources (coal combustions) are still dominated in the country (Lai et al., 2007;Li et al., 2009).In order to examine the seasonal patterns of NO 3 -/SO 4 2-ratios, the measurements with all four seasons were selected from Table S1 as shown in Table 4.There are obvious seasonal variations for the ratios with much lower in summer and higher in winter and spring.This may be attributed to the high temperatures, RH and radiations in summer, which are favorable for the formation of sulfate, but not for the formation of NH 4 NO 3 (s), which is apt to decomposing in the high temperature.In addition, the NH 4 NO 3 composition on quartz filter during the sampling also should have a non-ignorable contribution to lower NO 3 -/ SO 4 2-ratios.In early spring and winter, the low temperature favors a shift from the gaseous phase as HNO 3 to the particulate phase as NH 4 NO 3 , but against for sulfate (see Section Formation of sulfate and nitrate aerosols from gaseous precursor conversion), which results in higher NO 3 -/SO 4 2-ratios in the two seasons, even more than one in part areas (Duan et al., 2006;Yang et al., 2007;Yang et al., 2011).Another meaningful inspection was performed between the correlations of sulfate and nitrate.The mass concentrations of sulfate and nitrate exhibit a relative weak correlation (R 2 = 0.42, p = 0.00) (Fig. 4).However, there is a significantly negative correlation between the mass percentages of sulfate and nitrate when there are high sulfate plus nitrate loadings in PM 2.5 , for example, the proportions of sulfate plus nitrate in PM 2.5 are ≥ 40% (R 2 = 0.53, p = 0.00) and 30-40% (R 2 = 0.62, p = 0.00) (Fig. 5).It suggests that the results from the former study in Shanghai, Hangzhou and Guangzhou (Kong et al., 2014) are not local phenomena, but a universal rule across the country.The negative correlation signifies that more nitrates will probably be produced when sulfate formation decreases along with the reduction of SO 2 from current emission control strategies.Nitrate aerosol will become more crucial in the future atmosphere along with the increase in nitrogen oxide emissions and probably cause serious air pollution just like that of sulfate in China.

Relationships among Sulfate, Nitrate and Ammonium
In order to explore the existent forms of SNA across China, the correlations of SNA were determined by regression analysis.In Fig. 4, sulfate and ammonium has a good correlation with a correlation coefficient of 0.62 (p = 0.00), implying that they are associated with the same particulate system.In the atmosphere, ammonium is mainly produced by the reaction between gaseous NH 3 and acidic sulfate and nitrate particles (Yang et al., 2005).The molar ratio of sulfate to ammonium for (NH 4 ) 2 SO 4 and NH 4 HSO 4 are 0.50 and 1.00, respectively.So the sulfate to ammonium slope of 0.28 in Fig. 6 suggests ammonium is rich and sulfate is completely neutralized by ammonium, and (NH 4 ) 2 SO 4 are formed in aerosols (Wang et al., 2006b;Li et al., 2009).Moreover, nitrate also shows a good correlation with ammonium (R 2 = 0.66, p = 0.00) with a slope of less than 1, meaning that the formation of NH 4 NO 3 and the possible existences of other anions (such as Cl -, F -, Br -, C 2 O 4 2-…) in particles (He et al., 2001;Hu et al., 2002;Yang et al., 2005;Li et al., 2008;Yang et al., 2012).Notably, the regression fitted between (2[SO 4 2− ] + [NO 3 -]) and [NH 4 + ] shows a strongly linear correlation with a slope very closely to 1, indicating a completely neutralized system, and (NH 4 ) 2 SO 4 and NH 4 NO 3 may be the major compounds that SNA exist in the atmospheric particles in China.

Evaluation for Climate Model
Climate models are useful to predict aerosol concentrations and radiative forcing, despite considerable uncertainties of simulations in key source regions such as China (Bauer et al., 2007;Han et al., 2011;Zhang et al., 2012b;Cheng et al., 2013).We retrieved data during 2000-2012 from GISS model in CMIP5 to simulate SNA's spatial and temporal distributions.The simulated average SNA by GISS are in good agreements with the observations in spatial patterns as shown in Figs.6-9.The GISS simulation well captured the winter peaks of nitrate (Fig. 6) and ammonium (Fig. 7), although the ammonium concentration was slightly underpredicted in middle China during winter season.This underestimation of ammonium may be caused by uncertainties of emissions and/or photochemical parameterization in GISS model.However, sulfate concentrations in the model are obvious lower in winter than summer levels, contrasting to the observed maximum sulfate concentration in winter (Fig. 8).The model performed well for the prediction of sulfate during summer, however, sulfate concentration in winter were significant underpredicted in northern China (Fig. 9).Such underprediction of winter sulfate in northern China is also reported in other climate models, such as    GEOS-Chem (Wang et al., 2013) (Fig. S2).This means that although SO 2 emissions in China peak in winter (Table 2), the photochemical production of sulfate probably isn't yet well resolved in the model in the severe polluted area in northern China, such as Northern China Plain (NCP).

CONCLUSIONS
In this paper, we investigate SNA in aerosols in China based on almost all data published before 2012 in the field measurements.The results show that SNA in China are much higher than those in USA and Europe.In fine particles, sulfate is the most abundant species among SNA with the average concentration of 17.96 ± 10.23 µg m -3 compared to 7.66 ± 5.68 µg m -3 and 7.26 ± 5.10 µg m -3 for nitrate and ammonium, respectively.SNA occupy 34.2 ± 10.9% of PM 2.5 mass and 28.2 ± 8.5% of PM 10 mass in the country.In the seasonal distribution, SNA all have the highest concentrations in winter, but the lowest levels are in spring for sulfate and ammonium and in summer for nitrate.In the spatial pattern, the high SNA concentrations occur in North China Plain, Yangtze River delta and Sichuan Basin, especially high in the urban areas in North China Plain.The formation of sulfate and nitrate are different in different regions (e.g., northern and southern China) and seasons.Sources analyses find that the mean NO 3 -/SO 4 2ratio is 0.43 ± 0.26, signifying that stationary sources (coal combustions) are still dominant in China.The regression analysis suggests that (NH 4 ) 2 SO 4 and NH 4 NO 3 are the main existence forms of SNA in China.On these observations, the model simulation for SNA aerosols are evaluated.The simulation by GISS model in CMIP5 in both summer and winter show good agreement with SNA concentrations based on the observations.However, the simulated sulfate concentrations present the maximum in summer, contrasting to the highest observational concentration in winter when the simulated nitrate and ammonium concentrations are consistent with the observations in seasons, which is probably attributed to the underestimation of sulfate in the model in northern China in winter.These results provide useful information on SNA in China and is significant for the establishment of policy for legislator to reduce particle pollution in the country.

Fig. 1 .
Fig. 1.The fractions of sulfate, nitrate and ammonium in (a) PM 2.5 and (b) PM 10 of China, in PM 2.5 of (c) USA and (d) Europe, respectively.

Fig. 9 .
Fig. 9. Comparison between model and observation for latitude trend of sulfate concentrations in summer (a) and winter (b).The model shows the average between 110 to 120° in longitude.Sites of observation are shown in Fig. 8.

Table 1 .
The average concentration of SNA in four seasons in China, North and South in China.

Table 2 .
The emissions of SO 2 and NO x in four seasons of China; the average wet precipitation, rainfall frequency in four seasons of China; the average RH and temperature in northern and southern China.

Table 3 .
The SOR and NOR in four seasons in China.

Table 4 .
The mass ratios of NO 3