Long-term Real-time Measurements of Aerosol Particle Composition in Beijing, China Printer-friendly Version Interactive Discussion Long-term Real-time Measurements of Aerosol Particle Composition in Beijing, China: Seasonal Variations, Meteorological Effects, and Source Analysis Acpd Long-term Real-

This discussion paper is/has been under review for the journal Atmospheric Chemistry and Physics (ACP). Please refer to the corresponding final paper in ACP if available. Abstract High concentrations of fine particles (PM 2.5) are frequently observed during all seasons in Beijing, China, leading to severe air pollution and human health problems in this megacity. In this study, we conducted real-time measurements of non-refractory sub-micron aerosol (NR-PM 1) species (sulfate, nitrate, ammonium, chloride, and organics) 5 in Beijing using an Aerodyne Aerosol Chemical Speciation Monitor for 1 year, from July 2011 to June 2012. This is the first long-term, highly time-resolved (∼ 15 min) measurement of fine particle composition in China. The seasonal average (±1σ) mass concentration of NR-PM 1 ranged from 52 (±49) µg m −3 in the spring season to 62 (±49) µg m −3 in the summer season, with organics being the major fraction (40–51 %), followed by 10 nitrate (17–25 %) and sulfate (12–17 %). Organics and chloride showed pronounced seasonal variations, with much higher concentrations in winter than in the other seasons , due to enhanced coal combustion emissions. Although the seasonal variations of secondary inorganic aerosol (SIA = sulfate + nitrate + ammonium) concentrations were not significant, higher contributions of SIA were observed in summer (57–61 %) than in 15 winter (43–46 %), indicating that secondary aerosol production is a more important process than primary emissions in summer. Organics presented pronounced diurnal cycles that were similar among all seasons, whereas the diurnal variations of nitrate were mainly due to the competition between photochemical production and gas–particle partitioning. Our data also indicate that high concentrations of NR-PM 1 (> 60 µg m −3) are 20 usually associated with high ambient relative humidity (RH) (> 50 %) and that severe particulate pollution is characterized by different aerosol composition in different seasons. All NR-PM 1 species showed evident concentration gradients as a function of wind direction, generally with higher values associated with wind from the south, southeast or east. This was consistent with their higher potential as source areas, as determined 25 by potential source contribution function analysis. A common high potential source area, located to the southwest of Beijing along the Taihang Mountains, was observed during all seasons except winter, when smaller source areas were found. demonstrate a high potential impact of regional transport from surrounding regions on the formation of severe haze pollution in Beijing.


Introduction
Severe haze pollution episodes, characterized by high concentrations of fine particles (PM 2.5 ), occur frequently during all seasons in China (Sun et al., 2013b;Guo et al., 2014;Zheng et al., 2015), not only reducing visibility significantly, but also exerting harmful effects on public health (Cao et al., 2012).The mass concentrations of PM 2.5 often far exceed the China National Ambient Air Quality Standard (NAAQS; 75 µg m −3 as a 24 h average), particularly in the economically developed regions of Beijing-Tianjin-Hebei and Yangtze River Delta (YRD).According to Beijing Environmental Statements, the annual average mass concentration of PM 2.5 was 89.5 and 85.9 µg m −3 in 2013 and 2014, respectively, 2.5 times the NAAQS (35 µg m −3 as an annual average), indicating that Beijing is still facing severe fine particle pollution.While extensive studies have been conducted in recent years to characterize severe haze pollution (e.g., Guo et al., 2014;Huang et al., 2014;Sun et al., 2014;Zheng et al., 2015), most were carried out in a particular season.In reality, the very different compositions, sources, and evolution processes of severe haze pollution among the different seasons mean that a longer-term approach is needed to meet the challenge of mitigating fine particle pollution in Beijing.
A number of long-term measurements and source analyses have been conducted in Beijing during the last decade.Zhao et al. (2009) reported pronounced seasonal variations of PM 2.5 , with higher concentrations in winter than summer.Similarly, Yang et al. (2011) conducted a long-term study of carbonaceous aerosol from 2005 to 2008 in urban Beijing.Both organic carbon (OC) and elemental carbon (EC) showed pronounced seasonal variations, with the highest concentrations occurring in winter and the lowest values in summer.A more detailed investigation of the chemical composition and sources of PM 2.5 in urban Beijing can be found in R. Zhang et al. (2013).Sources Introduction

Conclusions References
Tables Figures

Back Close
Full of fine particles also vary greatly among the different seasons; for instance, coal combustion during periods requiring more domestic heating, biomass burning in harvest seasons, and dust storms in spring (Zheng et al., 2005; R. Zhang et al., 2013).Despite this, most previous long-term studies either focused on limited aerosol species, relied upon weekly filter samples, or used one month's worth of data to represent an entire season.Therefore, our understanding of the full spectrum of seasonal variations of aerosol species and sources remains quite poor.
The Aerodyne Aerosol Mass Spectrometer (AMS) is unique in its ability to provide real-time, online measurements of size-resolved submicron aerosol composition (Jayne et al., 2000;Canagaratna et al., 2007).While the AMS has been widely used in China in recent years (Xu, J. et al., 2014 and references therein), real-time, longterm measurements of aerosol particle composition are still rare.Y. Zhang et al. (2013) conducted a four-month measurement campaign of submicron aerosol composition and size distributions using a quadrupole AMS in urban Beijing.Their results showed higher concentration of organics during wintertime and secondary inorganic species in summer.Furthermore, positive matrix factorization (PMF) analysis of organic aerosol (OA) showed higher primary OA (POA) in winter and secondary OA (SOA) in summer.However, measurements over only one month or even less were conducted for each season, due to the high cost and maintenance of the AMS.The recently developed Aerosol Chemical Speciation Monitor (ACSM) is specially designed for long-term routine measurements of submicron aerosol composition (Ng et al., 2011).The ACSM has been proven reliable by several recent long-term field measurements, e.g., in Paris (Petit et al., 2015), north-central Oklahoma (Parworth et al., 2015), and Santiago de Chile (Carbone et al., 2013).Although the ACSM has been deployed at various sites in China (Sun et al., 2012(Sun et al., , 2013b;;Zhang et al., 2015), long-term measurements have yet to be reported.
In this study, the first of its kind, we conducted long-term, real-time measurements of non-refractory submicron aerosol (NR-PM 1 ) composition with an ACSM in Beijing, China, from July 2011 to June 2012.The seasonal variations of mass concentration Introduction

Conclusions References
Tables Figures

Back Close
Full and composition of submicron aerosol were characterized, and the diurnal cycles of aerosol species during the four seasons elucidated.The effects of meteorological parameters, particularly relative humidity and temperature, on aerosol composition and formation mechanisms were investigated.Finally, the potential source areas leading to high concentrations of aerosol species during the four seasons were investigated via potential source contribution function (PSCF) analysis.

Sampling site
The ACSM was deployed on the roof of a two-story building (∼ 8 m) at the Institute of Atmospheric Physics (IAP), Chinese Academy of Sciences (39 • 58 28 N, 116 • 22 16 E, Fig. 1a) from July 2011 to June 2012.The sampling site is located between the north third and fourth ring road in Beijing, which is a typical urban site with influences from local traffic and cooking sources (Sun et al., 2012).The wind rose plots (Fig. 1b) show that southwesterly winds dominate all seasons except winter, when northwesterly and northerly winds prevail.The spring and fall seasons are also characterized by high frequencies of northwesterly and northerly winds.Also note that the prevailing winds with high wind speeds are more frequent during winter and spring than summer.The meteorological parameters, including wind speed (WS), wind direction, relative humidity (RH), and temperature (T ) were obtained from a 325 m meteorological tower at the same location.The parameters of pressure (P ), solar radiation (SR), and precipitation were obtained from a ground meteorological station located nearby.The monthly variations of these meteorological parameters are presented in Fig. 2. Pronounced seasonal variations were observed for all meteorological parameters except WS.RH averaged at > ∼ 60 % in summer and presented its minimum value (< 30 %) in February.Temperature and solar radiation showed similar seasonal cycles, with high values in summer and low values in winter.The monthly variations of WS were relatively flat, Introduction

Conclusions References
Tables Figures

Back Close
Full yet slightly higher values in spring were observed.In addition, a considerable amount of precipitation was observed from June to August, yet it was negligible during wintertime.All the data are reported in Beijing Standard Time in this study.

Aerosol and gas measurements
The submicron aerosol particle composition including organics, sulfate, nitrate, ammonium, and chloride was measured in situ by the ACSM at a time resolution of ∼ 15 min.The ACSM, built upon previous AMSs (Jayne et al., 2000;Drewnick et al., 2005;De-Carlo et al., 2006), is specially designed for long-term routine measurements of fine particle composition (Ng et al., 2011).The ACSM has been successfully deployed at various sites for chemical characterization of submicron aerosol (Ng et al., 2011;Budisulistiorini et al., 2013;Carbone et al., 2013;Sun et al., 2013b;Parworth et al., 2015).
In this study, ambient aerosol particles were delivered to the sampling room through a stainless steel tubing (outer diameter: 1.27 cm) with a flow rate of ∼ 3 L min −1 , out of which ∼ 84 cc min −1 was sampled into the ACSM.A PM 2.5 URG cyclone (URG-2000-30ED) was installed in front of the sampling line to remove coarse particles (> 2.5 µm).
To reduce the uncertainties of collection efficiency (CE), a silica gel diffusion dryer was set up in the front of the ACSM to ensure that the aerosol particles sampled were dry (< 40 %).The ACSM was calibrated routinely with pure ammonium nitrate particles for the response factor following the procedures detailed in Ng et al. (2011).A more detailed description of the ACSM calibration is also given in Sun et al. (2012).

ACSM data analysis
The ACSM data were analyzed for the mass concentrations of NR-PM 1 species including organics, sulfate, nitrate, ammonium, and chloride using ACSM standard data analysis software.The RH in the sampling line, aerosol particle acidity and the fraction of ammonium nitrate (f AN ) in NR-PM 1 are three major factors affecting the uncertainties of CE (Huffman et al., 2005;Matthew et al., 2008;Middlebrook (Zhang et al., 2007) was less than 0.8, the composition dependent CE recommended by Middlebrook et al. (2012), which is CE = max(0.45,0.0833 + 0.9167 × f AN ), was used in this study.The validity of the ACSM data using variable CE in summer and winter was reported previously in Sun et al. (2012Sun et al. ( , 2013b) by comparing the NR-PM 1 with PM 2.5 mass concentration measured by a TEOM system.The correlation between NR-PM 1 and PM 2.5 for the entire year is shown in Fig. S1.The measured NR-PM 1 overall tracked well with that of PM 2.5 , and yet showed different slopes in different seasons.The average ratio of NR-PM 1 /PM 2.5 for the entire year was 0.77 (r 2 = 0.66).It should be noted that the PM 2.5 was measured by a heated TEOM (50 • C), which might have caused significant losses of semi-volatile species, e.g., ammonium nitrate and semi-volatile organics.For example, Docherty et al. ( 2011) found an average loss of ∼ 44 % PM 2.5 mass through use of the heated TEOM compared to that measured with a filter dynamics measurement system.Assuming that the average loss of PM 2.5 mass by the heated TEOM is 30-50 %, the NR-PM 1 /PM 2.5 ratio for the entire study would be ∼ 0.5-0.6,which is close to that reported in Y. Zhang et al. (2013).Figure S1 also shows large variations of NR-PM 1 /PM 2.5 ratios in the different seasons.The reasons for the variations include: (1) the ACSM cannot detect refractory black carbon and mineral dust.For example, low ratios of NR-PM 1 /PM 2.5 (< 0.3) were observed during dust storm periods, when mineral dust is the dominant component of fine particles, (2) the contribution of semi-volatile species to PM 2.5 varied greatly among the different seasons; and (3) the contribution of particles in the range of 1-2.5 µm to the total PM 2.5 might also be different among different pollution episodes.Introduction

Conclusions References
Tables Figures

Back Close
Full

PSCF analysis
The 72 h back trajectories arriving at the IAP study site at a height of 300 m were calculated every 3 h for the entire study period using the National Oceanic and Atmospheric Administration Hybrid Single-Particle Lagrangian Integrated Trajectory model, version 4.8 (Draxler and Rolph, 2003).The back trajectory data were then input into the PSCF (Polissar et al., 2001), which calculates the ratio of the number of points with concentration higher than a threshold value (m i j ) to the total number of points (n i j ) in the ijth grid cell.Higher PSCF values indicate higher potential source contributions to the receptor site.In this study, the domain for the PSCF was set in the range of (34-44 • N, 110-124 • E).The 75th percentile for each aerosol species during the four seasons (Table S1) was used as the threshold value to calculate m i j .To reduce the uncertainties of m i j /n i j for those grid cells with a limited number of points, a weighting function recommended by Polissar et al. (2001) was applied to the PSCF in each season.

Mass concentration and chemical composition
The average mass concentration of NR-PM 1 was 62 µg m −3 in summer (Fig. 3), which is higher than the 50 µg m −3 for July-August 2011 reported in Sun et al. (2012) due to the biomass burning impacts in June 2012.The summer NR-PM 1 level is close to that measured by a High Resolution Aerosol Mass Spectrometer during the Beijing 2008 Olympic Games (Huang et al., 2010), but ∼ 20 % lower than that determined in summer 2006 (Sun et al., 2010).The average NR-PM 1 mass concentrations were relatively similar during the other three seasons, varying from 52 to 59 µg m −3 and with slightly higher concentration during wintertime (Fig. 3).The NR-PM 1 measured in urban Beijing is overall higher than those previously reported in the Yangtze River Delta (YRD) region (27-43 µg m −3 ) (Huang et al., 2012a, b;Zhang et al., 2015) and Pearl Introduction

Conclusions References
Tables Figures
As indicated in Fig. 4, the summer season showed the highest frequency of moderately polluted days, with NR-PM 1 loading in the range of 30-60 µg m −3 (36 % of the time), while the winter season presented the highest frequency of low mass loadings (< 20 µg m −3 , 34 % of the time) due to the prevailing northwesterly winds (Fig. 1b).However, high NR-PM 1 loading (> 90 µg m −3 ) occurred 31 % of the time during the winter season, substantially more than during any of the other seasons (25, 25 and 21 % during summer, fall and spring, respectively), indicating that heavy pollution occurred more frequently during winter than the other seasons.The fall and spring seasons showed similar variations of frequencies, which overall decreased monotonically as a function of NR-PM 1 loadings.Note that heavily polluted days, with NR-PM 1 mass concentrations larger than 150 µg m −3 , occurred during all seasons, on average accounting for 3-7 % of the total time.Such heavily polluted days were mainly caused by agricultural burning in summer and fall, and coal combustion in winter, particularly under stagnant meteorological conditions (Sun et al., 2013b;Cheng et al., 2014).
The NR-PM 1 species varied dramatically and differently during the four seasons (Fig. 3).Overall, organics dominated NR-PM 1 during all seasons, accounting for 40-51 % on average.The dominance of organics in NR-PM 1 has been widely observed at various sites in China, e.g., 31-52 % in the YRD region (Huang et al., 2012a, b;Zhang et al., 2015), 36-46 % in the PRD region (He et al., 2011;Huang et al., 2011;Gong et al., 2012), and 47 % in northwest China (Xu, J. et al., 2014).Organics showed the largest contribution to NR-PM 1 in winter due to a large amount of carbonaceous aerosol emitted from coal combustion (Chen et al., 2005;Zhang et al., 2008).This is also con-Introduction

Conclusions References
Tables Figures

Back Close
Full sistent with the highest contribution of chloride, with coal combustion being a major source in winter (Zhang et al., 2012).High concentrations of organics were also observed during late June and late October, due to the impacts of agricultural burning in these two months.Secondary inorganic aerosol (SIA = sulfate + nitrate + ammonium) contributed the largest fraction of NR-PM 1 during the summer season (59 %) and the lowest fraction during the winter season (44 %).Such seasonal differences in PM composition reflect the different roles played by primary emissions and secondary formation.While photochemical production of secondary aerosol associated with higher O 3 and stronger solar radiation (Fig. 2) plays a dominant role in affecting aerosol composition in summer, primary emissions play enhanced roles in winter when photochemical processing is weaker (Sun et al., 2013b).It is interesting to note that nitrate, on average, showed a higher contribution than sulfate during the four seasons.Compared to previous AMS measurements in Beijing (Huang et al., 2010;Sun et al., 2010), the nitrate contribution to NR-PM 1 appears to show an increasing trend.The ratio of NO varied from 1.3-1.8 in this study, which is overall higher than those (0.8-1.5) observed during the four seasons in 2008 (Y.Zhang et al., 2013).This result likely indicates a response of secondary inorganic aerosol composition to the variations of precursors of NO x and SO 2 in recent years.For instance, a continuous effort to reduce SO 2 emissions is accompanied with a gradual increase in NO x emissions (S.X. Wang et al., 2014), which results in an increasingly more important role played by nitrate in PM pollution in Beijing.Indeed, a recent model analysis of the response of SIA to their precursors from 2000-2015 showed that the increase of nitrate would exceed the reduction of sulfate in northern China, assuming no change to NH 3 emissions (Wang et al., 2013).A higher concentration of nitrate than sulfate has also been frequently observed at urban and rural sites in China in recent years, e.g., Nanjing, in the YRD region (Zhang et al., 2015), and Changdao Island (Hu et al., 2013).Introduction

Conclusions References
Tables Figures

Back Close
Full

Seasonal variations
The monthly average NR-PM 1 mass concentration stayed relatively constant throughout the year, with the average value ranging from 46 to 60 µg m −3 , except in June 2012 (Fig. 5).The month of June presented the highest NR-PM 1 (89 µg m −3 ) due to the impact of agricultural burning.Consistently, a higher concentration of NR-PM 1 was observed in the summer of 2008 (5 June-3 July) than the other seasons in Beijing (Y.Zhang et al., 2013).Zhao et al. (2009) also observed the highest concentration of PM 2.5 in June 2007, due to the influences of agricultural burning.These results indicate that agricultural burning is a large source of PM pollution in Beijing in summer.The lowest concentration of NR-PM 1 in this study occurred in July, mainly due to the abundant precipitation and high temperatures, which facilitated wet scavenging and convection of PM, respectively (Fig. 2).Similarly lower concentrations of PM 2.5 in summer than in the other seasons were also observed previously at an urban site in Beijing (Zhao et al., 2009).Among the NR aerosol species, organics and chloride presented pronounced seasonal variations, showing higher concentrations in winter than the other seasons (Fig. 5).The concentration of organics increased from 17 µg m −3 in July to ∼ 30 µg m Introduction

Conclusions References
Tables Figures

Back Close
Full (2.8-3.3 µg m −3 ) was approximately six times that (0.5 µg m −3 ) in summer.The contribution of chloride to NR-PM 1 showed a similar seasonal trend, with the lowest contribution in summer (∼ 1 %) and the highest in winter (∼ 5-6 %) (Fig. 6).High concentrations of chloride in winter are associated with enhanced coal combustion emissions (Sun et al., 2013b), but also with low ambient temperature, which facilitates the formation of particle-phase ammonium chloride.Also note that chloride showed a twice as high concentration and contribution in June than the other two months in summer because agricultural burning is also a large source of chloride (Viana et al., 2008;Cheng et al., 2014).
The seasonal variation of sulfate is different from organics and chloride.The sulfate concentration gradually decreased from 10.1 µg m −3 in August to 4.9 µg m −3 in November, which was associated with a synchronous decrease in solar radiation and O 3 (Fig. 2).The contribution of sulfate to NR-PM 1 showed a corresponding decrease from 19 to 10 %.The sulfate concentration then increased to 8.3-8.8 µg m −3 in December and January, likely due to a significant increase of precursor SO 2 associated with an increased demand for domestic heating during the winter season, which can be oxidized to form sulfate via either gas-phase oxidation or aqueous-phase processing (W.Y. Xu et al., 2014).Sulfate showed the highest concentration in June (13.5µg m −3 ) due to secondary production, but possibly the impact of biomass burning as well.Indeed, a recent study in the YRD region also found a large enhancement of sulfate in biomass burning plumes in summer (Zhang et al., 2015).Nitrate showed minor seasonal variation, with the monthly average concentration ranging from 8 to 15 µg m −3 , except in June (23 µg m −3 ).It is interesting that a higher concentration of nitrate was observed in summer and spring than in winter.On average, nitrate accounted for ∼ 25 % of NR-PM 1 during summertime, but decreased to ∼ 15 % during wintertime (Fig. 6).
Although high temperatures in summer favor the dissociation of ammonium nitrate particles to gas-phase ammonia and nitric acid, the correspondingly high RH and excess gaseous ammonia facilitate the transformation of nitric acid to aqueous NH 4 NO 3 particles (Meng et al., 2011;Sun et al., 2012).The lowest concentration of nitrate during Introduction

Conclusions References
Tables Figures

Back Close
Full wintertime might be primarily caused by the weak photochemical production associated with low solar radiation and oxidants (e.g., O 3 ).In addition, the higher particle acidity in winter (Liu, 2012) and lower mixing ratio of gaseous ammonia may also suppress the formation of ammonium nitrate particles (Zhang et al., 2007).The seasonal variation of ammonium is similar to that of sulfate and nitrate because ammonium primarily exists in the form of NH 4 NO 3 and (NH 4 ) 2 SO 4 .

Diurnal variations
As demonstrated in Fig. 7, the diurnal cycles of organics during the four seasons were overall similar, characterized by two pronounced peaks occurring at noon and during the evening time.PMF analysis of OA suggested that the noon peak was primarily caused by cooking emissions, while the evening peak was driven by different primary emissions (e.g., cooking, traffic, and coal combustion emissions) among the different seasons (Sun et al., 2012(Sun et al., , 2013b)).It should be noted that the noon peaks in summer were more significant than those in fall and winter.Indeed, the cooking emissions, determined by subtracting the background (10:00-11:00) from the noon peak (12:00-13:00), were ∼ 1.5-2 µg m −3 from September to the following March, which were lower than the ∼ 3.5 µg m −3 calculated for June and July.This seasonal trend agreed with that of temperature, indicating that cooking emissions are temperature dependent, probably because of increased cooking activity in hot summers than cold winters.Relatively flat diurnal cycles were observed for sulfate during most months, indicating the regional characteristics of sulfate.In fact, multi-day build-up of sulfate was frequently observed during all seasons (Fig. 3), supporting the notion of regional influences on sulfate in Beijing.It should be noted that the daytime photochemical production of sulfate from gas-phase oxidation of SO 2 might be masked by an elevated planetary boundary layer (PBL).Considering the dilution effect of the PBL, Sun et al. (2012) found that sulfate increased gradually from morning to late afternoon, demonstrating the daytime photochemical production of sulfate.In this study, sulfate in May, June and October showed an evident daytime increase until late afternoon, indicating an impor-14562 Introduction

Conclusions References
Tables Figures

Back Close
Full tant role played by gas-phase photochemical processing in driving the sulfate diurnal cycle.
Nitrate showed substantially different diurnal cycles among different months.A clear daytime increase starting from ∼ 08:00 until ∼ 19:00 was found in the five months of January, February, March, November and December, indicating that such a diurnal pattern is more significant during wintertime compared to the fall and spring seasons.Figure 2 shows that the temperature during these five months was generally low (< 10 • C), under which the partitioning of NH 4 NO 3 into gaseous NH 3 and HNO 3 would not be significant.As a result, photochemical production would be the primary factor driving the diurnal variations.The photochemical production rate calculated from the daytime increase was 0.6-0.8µg m −3 h −1 during winter and ∼ 0.2-0.3µg m −3 h −1 in November and March.Nitrate presented pronounced diurnal cycles in summer (June, July and August), with the concentrations gradually decreasing during daytime and reaching a minimum at ∼ 16:00.Similar diurnal cycles have been observed on many occasions in summer in Beijing (Huang et al., 2010;Sun et al., 2012;Y. Zhang et al., 2013).
The evaporative loss of NH 4 NO 3 associated with high temperatures, which overcomes the amount of photochemical production, plays the major role in driving such diurnal cycles.The rising PBL plays an additional role in the low concentrations of nitrate during daytime (Sun et al., 2012).The diurnal cycle of nitrate in May and September was also significant, characterized by a pronounced morning peak occurring at ∼ 10:00, when photochemical production dominated over the gas-particle partitioning of NH 4 NO 3 .
Nitrate showed a relatively flat diurnal cycle in April, indicating a combined effect of various nitrate formation mechanisms.
Chloride in this study was mainly detected as ammonium chloride, which shows similar volatile properties as NH 4 NO 3 .As shown in Fig. 7, two different diurnal cycles were observed throughout different months.For the months of July, August, September, April and May, chloride presented a morning peak when both temperatures and the PBL were at their lowest, and then rapidly decreased to a low ambient level at ∼ 18:00.Such a diurnal cycle was likely primarily driven by temperature dependent gas-particle par-Introduction

Conclusions References
Tables Figures

Back Close
Full titioning (Hu et al., 2008).The diurnal cycles of chloride during the remaining months were also significant, all of which were characterized by high concentrations at night.Coincidentally, these months fell during the season of high domestic-heating demand, which usually starts on 15 November and ends on 15 March.Coal combustion has been found to be a large source of chloride (Zhang et al., 2012;Sun et al., 2013b).
Therefore, the diurnal cycle of chloride is likely dominantly driven by coal combustion emissions that are intensified at night for domestic heating.

Weekend effects
Because the switch between clean periods and pollution episodes arising from different source areas happens frequently in Beijing (Sun et al., 2013b;Guo et al., 2014), the diurnal cycles of aerosol species can vary greatly due to the influences of different occurrences of clean periods between weekdays and weekends (Sun et al., 2013b).Therefore, periods with low aerosol loadings (NR-PM 1 < 20 µg m −3 ) were excluded from the results (Fig. 8) for a better investigation of the weekend effects (for the average diurnal cycles with clean periods included, see Fig. S2).As shown in Fig. 8, there were no clear weekend effects in the summer, except for slightly lower concentrations of organics, sulfate and nitrate in the late afternoon at weekends.This suggests that no significant differences in anthropogenic activity between weekdays and weekends in summer.Although some enhanced traffic emissions between 00:00 and 06:00 at weekends might have occurred, as indicated by the higher concentration of NO (Fig. S3), they appeared to have negligible impacts on secondary sulfate and nitrate.While the diurnal variations of organics and chloride were similar between weekdays and weekends during the fall season, sulfate and nitrate showed pronounced weekend effects, with persistently higher concentrations at weekends throughout the day.An explanation for this is the stronger photochemical production of secondary species associated with higher O 3 and solar radiation at weekends (Fig. S3).Consistently, SOA showed similar weekend effects as those of secondary inorganic species, while POA did not (Sun et al., 2015).Because of the regional characteristics of secondary aerosols, fur-14564 Introduction

Conclusions References
Tables Figures

Back Close
Full ther analysis is needed to address the impacts of regional transport on the weekend effects of secondary species.Winter showed the most pronounced weekend effects for all aerosol species.All aerosol species showed much lower concentrations at weekends than on weekdays across the entire day, which was consistent with those of NO, SO 2 , and CO (Fig. S3).These results clearly indicate much reduced anthropogenic activity at weekends during wintertime because of low ambient temperature (−4 to −3 • C).Further evidence is provided by the diurnal cycles of organics, which presented pronounced noon peaks at weekends during all seasons except winter.This observation was consistent with much reduced cooking activity at weekends during wintertime.Similar to summer, no evident weekend effects were observed in spring.The weekend effects of aerosol species in this study are overall consistent with those observed by Han et al. (2009), in which similar diurnal cycles of primary elemental carbon, CO, and CO 2 between weekdays and weekends under weak wind conditions were observed during the three seasons other than winter.

Meteorological effects
Figure 9 shows the RH and T dependent distributions of NR-PM 1 and WS for the entire year.The distribution of NR-PM 1 showed an obvious concentration gradient as a function of RH.NR-PM 1 showed the lowest mass loading, generally less than 20 µg m −3 at RH < 20 %, and had no clear dependence on T .This can be explained by the high WS (often larger than 5 m s −1 ; Fig.The RH-and T -dependent distributions of major aerosol species (Fig. 10) allow us to further investigate the RH/T impacts on the formation of aerosol species.While all aerosol species showed similar concentration gradients as a function of RH to that of NR-PM 1 , the T -dependent patterns varied greatly.Organics generally showed the highest concentrations under low T (< 6 • C) and high humidity conditions -very similar to the behavior of chloride, which is mainly derived from combustion sources, e.g., coal combustion or biomass burning (Zhang et al., 2012;Cheng et al., 2014).The results suggest that high concentrations of organics during wintertime are primarily caused by coal combustion emissions during the domestic-heating season, particularly from residential coal combustion (Zhang et al., 2008).In fact, a previous study by our group found that nearly one-third of OA during wintertime is primary coal combustion OA (CCOA) (Sun et al., 2013b).In contrast, organics showed much lower concentrations under the conditions of higher RH and higher T , for which one of the reasons was probably far fewer coal combustion emissions during summertime (Zheng et al., 2005;R. Zhang et al., 2013).Consistently, CCOA has not yet been resolved from PMF analyses of AMS OA in summer in Beijing (Huang et al., 2010;Sun et al., 2010).Note that the region with a high concentration of organics corresponded to a high concentration of NR-PM 1 .In this region, organics accounted for the largest fraction of NR-PM 1 (approximately 40-50 %), indicating that severe PM pollution under low temperature and high humidity conditions is dominantly contributed to by organics.The mass fraction of organics, however, showed an opposite distribution to that of mass loading.The RH/T dependence of secondary inorganic species showed somewhat different behaviors from that of organics.Sulfate presented two high concentration regions, with the highest values occurring during wintertime when T was below 0 • C and RH was above 70 %.Aqueous-phase oxidation, mostly fog processing, has been found to play a dominant role in sulfate formation under such meteorological conditions (Sun et al., 2013a).Surprisingly, the semi-volatile nitrate showed a relatively homogeneous distribution across different temperatures at RH > 40 %.Despite high temperature in summer, high humidity facilitates the transformation of gaseous species into aqueousphase nitrate particles (Sun et al., 2012), particularly in the presence of high abundance of gaseous ammonia (Ianniello et al., 2010).In fact, nitrate showed the highest contribution (>∼ 25 %) to NR-PM 1 mass under high T and high RH conditions, which were also the conditions under which high concentrations of NR-PM 1 were observed.The fact that nitrate contributed more than sulfate (∼ 15-20 %) to NR-PM 1 mass during these conditions suggests an important role played by nitrate in summer haze formation.While the concentration of nitrate at various temperatures was close, its contribution to NR-PM 1 was generally lower at low temperatures due to the greater enhancement of organics during wintertime.Also note that the two semi-volatile species, i.e., nitrate and chloride, show the lowest contributions to NR-PM 1 in the top-left region with the highest T and lowest RH.This illustrates the evaporative loss process of ammonium nitrate and ammonium chloride under high temperatures in summertime.However, sulfate shows a relatively higher contribution in this region since ammonium sulfate is less volatile than ammonium nitrate and chloride (Huffman et al., 2009).

Source analysis
In summer, all NR-PM 1 species showed evident wind sector gradients, with higher concentrations in association with winds from the east (E) and southeast (SE), and lower concentrations with northwest (NW) wind (Fig. 11).The average NR-PM 1 concentration from the SE was 89.5 µg m −3 , which was more than twice that (39.4 µg m −3 ) from Introduction

Conclusions References
Tables Figures

Back Close
Full the NW.All aerosol species increased as wind sectors changed along the N-NE-E-SE gradient, and then decreased along the SE-S-SW-W gradient.Such wind sector dependence of aerosol composition is remarkably consistent with the spatial distribution of fine particles in Beijing in 2013 (Beijing Environmental Statement 2013).These results suggest an inhomogeneous distribution of air pollution around the IAP sampling site in summer.Organics dominated NR-PM 1 across different sectors (37-43 %), followed by nitrate (21-28 %), sulfate (15-20 %), and ammonium (15-17 %).While chloride contributed a small fraction of NR-PM 1 (0.7-1.8 %), the mass concentration showed the largest difference between SE and NW.The fall season showed a similar aerosol composition dependence as that in summer, with higher concentrations from the E, SE, and S.However, the gradients of wind sectors appeared to be smaller.For example, the average NR-PM 1 concentration ranged from 46.3 to 72.7 µg m −3 in all eight sectors except NW.Organics showed a similar dominance in NR-PM 1 , accounting for 47-55 %, and the contribution was ubiquitously higher than in summer for all wind sectors.It should be noted that the NW sector showed the largest difference between mean and median values for all species.The much lower median values suggest a dominance of clean days for most of the time in this sector.In contrast, the summer season showed higher median concentrations from the NW, indicating a higher regional background during this season.The winter season showed consistently high concentrations of PM across the different wind sectors, except for NW, where the mass con- As Fig. 12 shows, the potential source areas for aerosol species varied among the four seasons.In summer, high potential source areas were mainly located to the south, southwest and southeast of Beijing.Organics had a relatively small high potential source region in the south of Beijing (< 100 km) and a small source region located around Baoding -one of the most polluted cities in Hebei Province.A narrow and visible source area to the southeast of Beijing, including Tianjin and the Bohai Sea, was also observed.Nitrate and chloride showed similar source areas to organics.The high potential source area to the southeast Beijing was mainly caused by open agricultural burning in June in northern China.Sulfate showed a distinct source region characterized by a narrow high PSCF band along Hengshui-Baoding-Langfang-Beijing.Such a pollution band agrees well with the topography of the North China Plain, with the Taihang Mountains to the west and Yan Mountains to the north.The wide area of high PSCF for sulfate also indicates a regional characteristic of sulfate that is formed from gas-phase oxidation or cloud processing of precursor SO 2 , which is particularly high in Hebei Province (Ji et al., 2014).Secondary nitrate showed a similar, yet much smaller, PSCF region compared to sulfate.One reason for this might be due to the evaporative loss of ammonium nitrate during the long-range transport in summer.
All aerosol species showed similar PSCF spatial distributions during the fall season, with high potential source regions located in a narrow area from Hengshui, Baoding to Beijing.These results suggest that regional transport from the southwest plays a dominant role in formation of severe haze pollution in fall.The wintertime results showed largely different PSCF distributions from the other seasons.High PSCF values were mainly located in a small region (< 50 km) in the south and southeast of Beijing.Although Hebei Province often has worse air pollution than Beijing during wintertime (Ji et al., 2014), the cities far away from Beijing appear not to be a very important source of wintertime air pollution in Beijing.One explanation for this is that stagnant meteorological conditions occur more frequently in winter due to low WS and T inversions.Thus, local emissions and transport from nearby regions would play a more significant role in affecting the pollution level in Beijing.While the spring season showed similarly Introduction

Conclusions References
Tables Figures

Back Close
Full small high potential source regions to those during wintertime, an obvious high potential source area in Hebei Province was also observed.The transport of air pollution from the SW to the NE along the Taihang Mountains in northern China has been observed many times in previous studies (L.T. Wang et al., 2014;Z. Wang et al., 2014).
Given that many cities located on this pathway are often highly polluted, such as Shijiangzhuang, Baoding, and Hengshui, regional transport from these areas would have a potentially high impact on the formation of severe haze pollution in Beijing.

Conclusion
This paper presents the results from a year-long, real-time measurement study of submicron aerosol particle composition using an ACSM, conducted at an urban site in Beijing from July 2011 to June 2012.The mass concentration of NR-PM 1 varied dramatically, with the seasonal average concentration ranging from 52 to 62 µg m −3 .Organics comprised a major fraction of NR-PM 1 during all seasons, accounting for 40-51 % on average.The average contribution of nitrate to NR-PM 1 (17-25 %) exceeded that of sulfate (12-17 %) during all seasons, suggesting an enhanced role of nitrate in PM pollution in recent years.Organics and chloride were two species showing pronounced seasonal variations in both mass concentrations and mass fractions.The higher concentrations of organics and chloride in winter than summer were largely due to enhanced coal combustion emissions.We also observed high concentrations of organics and chloride in June and October -two months with strong agricultural burning impacts.The seasonal variations of secondary sulfate and nitrate were not significant because of the large variations of precursor concentrations, photochemical production, and also meteorological effects in different seasons.However, higher contributions of SIA in summer (57-61 %) than in winter (43-46 %) were still observed, indicating a more significant role of secondary production in summer.The diurnal cycles of organics were similar during all seasons, all characterized by two pronounced peaks.While the diurnal cycles of secondary sulfate were overall relatively flat during Introduction

Conclusions References
Tables Figures

Back Close
Full most months of the year, those of nitrate varied greatly in different seasons.It was evident that the diurnal cycles of nitrate are driven by gas-particle partitioning and daytime photochemical production in summer and winter, respectively.The winter season showed substantially different concentrations of aerosol species between weekdays and weekends, with much lower concentrations on weekends.However, no significant weekend effects were observed during the other seasons.Meteorological conditions play important roles in the formation of serve PM pollution in Beijing.In this study, we illustrate the influences of RH and T on aerosol loading and chemistry in different seasons.All aerosol species increased significantly under stagnant meteorological conditions associated with high RH and low WS.NR-PM 1 showed two high concentration regions (> 60 µg m −3 ) at RH > 60 %.While organics comprised a major fraction of NR-PM 1 in these two regions, the abundances of sulfate and nitrate and air temperature were largely different, suggesting they play different roles in causing PM pollution during different seasons.Under drier conditions (RH < 30 %), the NR-PM 1 concentration was generally low and organics contributed more than 50 % of its mass, indicating the importance of organics during clean periods.The semi-volatile nitrate presented the largest contribution under high RH and high T , highlighting the importance of nitrate formation via aqueous-phase processing in summer.All NR-PM 1 species showed obvious dependence on wind direction, with higher concentrations commonly associated with winds from the S, E and SE.This was consistent with the results from PSCF analysis, which showed that the high potential source areas were mainly located to the S and SW of Beijing.The high potential source areas varied differently during the four seasons.A common high potential source area to the SW of Beijing, along the Taihang Mountains, was observed during all seasons except winter, demonstrating the potentially high impact of regional transport on severe PM pollution in Beijing.The winter season showed a much smaller source region compared to the other seasons, indicating that local and regional transport over a smaller regional scale are more important.High potential source areas to the SE of Beijing were also observed for organics, nitrate and chloride in summer, likely due to agricultural burning.Introduction

Conclusions References
Tables Figures

Back Close
Full  Full Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Figure9shows the RH and T dependent distributions of NR-PM 1 and WS for the entire year.The distribution of NR-PM 1 showed an obvious concentration gradient as a function of RH.NR-PM 1 showed the lowest mass loading, generally less than 20 µg m −3 at RH < 20 %, and had no clear dependence on T .This can be explained by the high WS (often larger than 5 m s −1 ; Fig.9b) at low RH levels associated with clean air masses from the north and/or northwest.Previous studies have also found a strong association between low aerosol loading and high WS in Beijing(Han et al., 2009;Sun et al., 2013b).NR-PM 1 showed moderately high concentrations (∼ 20-40 µg m −3 ) at low RH (20-40 %), which rapidly increased to a high concentration level (> 60 µg m −3 ) at RH > 50 %.These results indicate that severe haze episodes in Beijing mostly occur under high humidity conditions, when WS is low as well.Two different regions with high concentrations of NR-PM 1 are apparent in Fig.9a: one in the top-right region with As shown in Fig. 10, organics presents the highest contribution to NR-PM 1 (∼> 50 %) in the lefthand region with low RH, indicating the dominance of organics during periods with low NR-PM 1 mass loadings.Such a distribution is independent of temperature, suggesting a ubiquitously organics-dominant composition during clean days in all seasons.Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | centrations were approximately half of those in the other sectors.The average NR-PM 1 ranged from 55.0 to 84.4 µg m −3 , with organics being the major fraction, accounting for 46-54 %.The spring season showed a similar wind sector dependence on aerosol composition as the fall season.The average NR-PM 1 ranged from 49.0 to 74.4 µg m −3 for all of the wind sectors except the N (38.5 µg m −3 ) and NW (24.7 µg m −3 ), which had much lower mass concentrations.Similar to other seasons, organics dominated NR-PM 1 throughout the different sectors (36-53 %), followed by nitrate (19-27 %) and sulfate (11-16 %).Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Hu, W. W., Hu, M., Yuan, B., Jimenez, J. L., Tang, Q., Peng, J. F., Hu, W., Shao, M., Wang, M., Zeng, L. M., Wu, Y. S., Gong, Z. H., Huang, X. F., and He, L. Y.: Insights on organic aerosol aging and the influence of coal combustion at a regional receptor site of central eastern China, Atmos.Chem.Phys., 13, 10095-10112, doi:10.5194/acp-13-10095-2013,2013.Huang, R.-J., Zhang, Y., Bozzetti, C., Ho, K.-F., Cao, J.-J., Han, Y., Daellenbach, K. R.Discussion Paper | Discussion Paper | Discussion Paper |

Figure 1 .Figure 2 .Figure 3 .
Figure 1.(a) Map of the sampling site (IAP).(b) Wind rose plots, color coded by wind speed for each season.The frequencies are set to the same scales for all seasons.
Because aerosol particles were dry and overall neutralized for most of the time, except some periods when the ratio of measured NH + 4 to predicted NH + Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | et al., 2012).

Table 1 .
Summary of mass concentrations of NR-PM 1 species, gaseous pollutants and meteorological parameters during the four seasons and entire study period.