Exploring the severe winter haze in Beijing

G. J. Zheng, F. K. Duan, Y. L. Ma, Y. Cheng, B. Zheng, Q. Zhang, T. Huang, T. Kimoto, D. Chang, H. Su, U. Pöschl, Y. F. Cheng, and K. B. He State Key Joint Laboratory of Environment Simulation and Pollution Control, School of Environment, Tsinghua University, Beijing 100084, China Ministry of Education Key Laboratory for Earth System Modeling, Center for Earth System Science, Tsinghua University, Beijing 100084, China Kimoto Electric Co., Ltd, 3-1 Funahashi-cho Tennoji-ku, Osaka 543-0024, Japan Multiphase Chemistry Department, Max Planck Institute for Chemistry, 55128, Mainz, Germany State Environmental Protection Key Laboratory of Sources and Control of Air Pollution Complex, Beijing 100084, China Collaborative Innovation Center for Regional Environmental Quality, Beijing 100084, China

in history.With hourly fine particle (PM 2.5 ) concentrations up to ∼ 900 µg m −3 , outdoor exposure caused adverse health effects (Nel, 2005;Pöschl, 2005;Peplow, 2014), including severe respiratory system related symptoms and deceases (Cao et al., 2014;Ouyang, 2013).Meanwhile the visibility was reduced down to 100 m, which disrupted traffic with canceled flights and closed highways.The government had to adopt emergency response measures to deal with these pollution episodes (http: //english.sina.com/china/p/2013/0113/548263.html).In addition to massive amounts of primary particulate matter, high emissions in China provided plenty of gas pollutants to serve as precursors for secondary aerosols (Zhang et al., 2009).Densely distributed mega-cities (i.e., city clusters) have worsened this situation, contributing to regional air pollution.Once the regional pollution is formed, the advection becomes less effective in scavenging local pollutants (no clean air from upwind).Thus, the regional pollution is more persistent compared with air pollution within a specific city.Moreover, cities within this region could not eliminate their pollution solely by reducing local emissions (Chan and Yao, 2008).
These extreme haze episodes attracted great scientific interest.Regional transport of pollutants was found to contribute considerably to concentrations of PM 2.5 (Z.Wang et al., 2014; L. T. Wang et al., 2014), dust (Yang et al., 2013;Y. Wang et al., 2014), and SO 2 (Yang et al., 2013) in Beijing.Atmospheric dynamic processes during hazy conditions were different from clean conditions, with a significant two-way feedback between PM 2.5 and boundary layer evolution (Z.Wang et al., 2014).Secondary inorganic aerosol species were suggested to be the major contributor to severe haze, based on off-line PM 2.5 analysis (Quan et al., 2014), and on-line non-refractory PM 1 analysis by an Aerosol Chemical Speciation Monitor (Sun et al., 2014).In addition, some studies described unusual atmospheric phenomena taking place under heavily polluted conditions, such as extremely low ozone concentration (less than 5 ppb) in the absence of diurnal variation (Zhao et al., 2013) and the synergistic oxidation of SO 2 and NO 2 (He et al., 2014).These findings suggest need for a better understanding on the haze formation mechanisms.Introduction

Conclusions References
Tables Figures

Back Close
Full In this study, we tried to address the following questions for the winter haze episodes aforementioned: (1) the relative importance of enhanced emission vs. meteorology; (2) the causes of the sharp PM 2.5 increase during the haze episodes in Beijing, whether it was mainly driven by an extremely rapid local chemical production or by regional transport; and (3) the dominant chemical mechanisms of haze formation.

Experimental methods
On-line ambient observation was conducted from 1-31 January 2013 on the campus of Tsinghua University.The observation site is situated on the rooftop of the Environmental Science Building ( 40• 00 17 N, 116 • 19 34 E), approximately 10 m above ground.
Tsinghua University is located in the northwest part of urban Beijing, close to the North 4th Ring Road, without any major pollution sources nearby.All observation data are hourly unified data.Mass concentrations of fine (PM 2.5 ) and coarse (PM 2.5-10 ) particles were simultaneously measured based on the β-ray absorption method by a PM-712 Monitor (Kimoto Electric Co., Ltd., Japan), which was equipped with a US-EPA PM 10 inlet and a PM 2.5 virtual impactor (Kimoto Electric Co., Ltd., 2012;Kaneyasu et al., 2014).Dehumidification was achieved with the hygroscopic growth correction formula: Dehumidified PM 2.5 mass conc.= Measured PM 2.5 mass conc.× 1 1 + 0.010 × e 6.000 RH 100 (1) where the 0.010 and 6.000 are localized coefficients, and RH is relative humidity in %.
All PM 2.5 hereinafter refer to the dehumidified PM 2.5 data.
A Sunset Model 4 semi-continuous carbon analyzer (Beaverton, OR, USA) was used to measure hourly organic carbon (OC) and elemental carbon (EC) concentrations in PM 2. was corrected accordingly (G.J. Zheng et al., 2014).OM was estimated as 1.6 • OC, based on previous results (Zhang et al., 2014;Xing et al., 2013).The use of fixed OM/OC ratio requires caveats because the ratio might change due to the variable oxidation degree of OM under different conditions.Hourly sulfate and nitrate concentrations in PM 2.5 were measured using an ACSA-08 Monitor (Kimoto Electric Co., Ltd., Japan).The ACSA-08 Monitor measured nitrates using a ultra-violet spectrophotometric method, and quantified sulfates with the BaSO 4based turbidimetric method after addition of BaCl 2 dissolved in polyvinyl pyrrolidone solution (Kimoto et al., 2013).Ammonium was predicted under the assumption that it existed as NH 4 NO 3 and (NH 4 ) 2 SO 4 (He et al., 2012), which might be an overestimation based on the non-refractory PM 1 results (Sun et al., 2014).Thus the predicted ammonium given here should be regarded as an upper limit.
SO 2 and NO 2 concentrations in Beijing, and PM 2.5 concentrations in other cities were acquired from the Atmospheric Environment Monitoring Network (Tang et al., 2012).Daily averaged solar radiation reaching ground data were downloaded from the China Meteorological Data Sharing Service System (http://cdc.cma.gov.cn).Planetary boundary layer (PBL) height was simulated with the Weather Research & Forecasting Model (B.Zheng et al., 2014).

General characteristics of Beijing winter haze
Primary atmospheric pollutant in Beijing during the winter of 2012-2013 was PM 2.5 , which constituted about 70 % of PM 10 .This ratio increased when PM 2.5 pollution became worse (Fig. 1b the highest ever reported in Beijing (Zhao et al., 2009(Zhao et al., , 2011(Zhao et al., , 2013;;Zhang et al., 2014).
The severe PM 2.5 pollution lasted nearly the whole month, characterized by frequent and long-lasting pollution episodes.Here, we define an episode as a set of continuous days with daily PM 2.5 averages exceeding 75 µg m −3 .In total, four episodes were identified in January 2013 (Fig. 1a): 4-8 January (Episode I), 10-16 January (Episode II), 18-23 January (Episode III), and 25-31 January (Episode IV).Maximum episodeaveraged PM 2.5 concentrations reached 245.4 µg m −3 in Episode II (see Table 1 for comparative information on Episodes I to III; Episode IV was not included because of missing data).In addition to the high average concentrations, these episodes were frequent (intervals between episodes were all ∼ 1 day) and long-lasting (5-7 days) compared with typical durations (5 days) and frequencies (1-3 days) of previous Beijing winter haze episodes (Jia et al., 2008).
Another unique feature of the PM 2.5 mass concentrations during this winter haze period was their dramatic hourly fluctuation.The maximum daily variation was 778.6 µg m −3 on 12 January.Hourly PM 2.5 changes of over 100 µg m −3 (increases or decreases) were observed over 40 times during this haze period.Hourly increases or decreases could reach up to 351.8 µg m −3 and −217.7 µg m −3 , respectively.Causes of these sharp transitions are discussed in Sect. 5.The variation of chemical composition with PM 2.5 pollution level, and among episodes, was also explored.We classified PM 2.5 pollution into 4 categories according to the Air Quality Index (http://kjs.mep.gov.cn/hjbhbz/bzwb/dqhjbh/jcgfffbz/201203/t20120302_224166.htm?COLLCC=2906016564&) (Fig. 1b): clean (PM 2.5 ≤ 35 µg m −3 ), slightly polluted (35 < PM 2.5 ≤ 115 µg m −3 ), polluted (115 < PM 2.5 ≤ 350 µg m −3 ), and heavily polluted (PM 2.5 > 350 µg m −3 ), where PM 2.5 refers to the hourly concentration.Under this classification, the slightly polluted, polluted, and heavily polluted levels generally correspond to small, moderate, and large PM 2.5 peaks in Fig. 1b.With increasing pollution level, the EC fraction decreased slightly, OC fraction decreased significantly, while sulfate and nitrate contributions increased sharply (Fig. 2a).It suggests that secondary inorganic aerosol species Introduction

Conclusions References
Tables Figures

Back Close
Full become more important during polluted periods concerning its contribution to the PM 2.5 .A similar trend was observed for NR-PM 1 (Sun et al., 2014) and off-line samples (Cheng et al., 2014).On average, OC, EC, nitrate, and sulfate comprised 21 %, 3 %, 19 % and 22 % of PM 2.5 (Fig. 2b).Good correlations with PM 2.5 were observed for OC, EC and nitrate (R 2 > 0.8 for these three species) for all data in January 2013, while for sulfate the correlation became weaker, reflecting larger episodic variations (Fig. 2b).
In Episode III, NO 2 exceeded SO 2 by 50 % (Table 1), generally in accordance with previous studies (Meng et al., 2009) (coal combustion, etc.) in local emissions or regional SO 2 -rich air masses transported to Beijing.

Emission enhancement vs. synoptic conditions
Haze episodes were much more severe and frequent in winter 2013 than in 2012.
One possible explanation is that there was an abrupt emission enhancement during 2013.However, we did not find such change in the emission inventory (http: //www.meicmodel.org/).Annual average emissions of primary PM 2.5 , SO 2 and NO x show slight differences between 2013 and 2012 (1.2 %, −1.3 % and 0.8 %, respectively) for the Beijing-Tianjin-Hebei region.The changes of monthly averaged emissions in January were higher than the annual average changes in rates, i.e., 2.1 %, 1.5 % and 2.5 % for primary PM 2.5 , SO 2 , and NO x , respectively; but they are still not significant compared to the changes in pollutant concentrations.Thus, we suspect that these haze episodes arose from the unfavorable synoptic conditions in January 2013.
The relative importance of enhanced emission vs. unfavorable meteorology was estimated by model simulations with three scenarios (Fig. 3).Base scenario (a) was designed to simulate the actual situation, i.e., with both input emission inventory and Introduction

Conclusions References
Tables Figures

Back Close
Full meteorology for January 2013.In scenarios (b) and (c), 2012 meteorology and 2012 emission inventory data were used, respectively.Our model used the Weather Research and Forecasting Model (version 3.5.1)and the Community Multiscale Air Quality (CMAQ, version 5.0.1)modeling system.The CMAQ model was enhanced with more comprehensive heterogeneous reactions (Wang et al., 2012) to improve its performance for sulfate and nitrate simulations.Details of our model configuration, modifications, and its evaluation are described in B. Zheng et al. (2014).
As expected, the influence of emission difference was negligible (Fig. 3a and 3c), resulting in a PM 2.5 difference of within ±10 µg m −3 for the North China Plain (NCP) (Fig. 3e).The negligible influence of variation in emissions was corroborated by excellent correlation between scenarios (a) and (c) in the simulated PM 2.5 results for Beijing (R 2 = 0.97).
In contrast, stable synoptic conditions in January 2013, which favored accumulation of emitted pollutants, were essential to the formation of the severe regional haze.Under the same emission level, changing the meteorological conditions from 2012 to 2013 resulted in a monthly average PM 2.5 increase of 10-40 µg m −3 in the Beijing area, and up to 120 µg m −3 over the whole NCP (Fig. 3a, b, d).This suggests that the severe haze episodes in January 2013 were most likely due to unfavorable meteorology, rather than an abrupt increase in emissions (Fig. 3d, e). Figure 4 compares peak PM 2.5 concentrations in the NCP region during Episodes II to IV and their corresponding surface weather maps, together with surface weather map from a clean hour (Fig. 4g).During severe haze episodes, the regional pollution covered most of Hebei and northern Henan Provinces.In general, Shandong Province was less polluted, except during Episode IV.Beijing borders this polluted region, with mountains to the northwest.Surface weather maps from polluted periods were generally characterized by a weak high-pressure center (1034-1037 hPa) northeast of Beijing, which could result in low surface wind speed and prevent the influx of northwest clean air (Xu et al., 2011;Zhao et al., 2013).During the peak hours of Episode II, Beijing was located near a low-pressure trough, where air masses from Introduction

Conclusions References
Tables Figures

Back Close
Full south, west and northeast converged.During Episode III, Beijing was located in a saddle between two pairs of high-and low-pressure centers, which also led to enhanced stability.In contrast, weather patterns for the clean hours were characterized by strong high-pressure centers (up to 1046 hPa) northwest of Beijing, i.e., the Siberian Anticyclone.With sharp pressure gradient, synoptic conditions produce effective convection and strong northerly winds, bringing dry and clean air masses into Beijing.
Local meteorology, controlled by synoptic conditions, could have "deterministic impacts" on air pollution levels (Xu et al., 2011).Compared with the clean periods, the polluted periods were associated with significantly lower wind speed and PBL, and higher temperature and RH (Fig. 5).Besides changes in the average level, diurnal pattern of temperature in polluted periods could also differ from clean periods, with diminished overnight (0:00 to 6:00 a.m.) temperature drop.

Local chemical production vs. regional transport
As shown in Fig. 6, Episode II consists of several sharp-increase events, in which PM 2.5 concentrations increased by over 400 µg m −3 within 1-3 h (maximum mass growth rate up to 351.8 µg m −3 h −1 ).Earlier studies have attributed this dramatic rate of increase to fast local chemical production (Y.Wang et al., 2014).However, we found that the apparent rapid changes are more likely to be caused by the regional transport of clean/polluted air masses.In winter, the Siberian Anticyclone could bring clean air masses into NCP (Jia et al., 2008;Liu et al., 2013) while southerly winds refill the areas with polluted air masses.The transition between clean and polluted air masses may result in an "apparent" sharp build-up of particle concentrations.In other words, these events reflected "interruption" and rapid recovery of pollution from adjacent areas, rather than merely local chemical production.The impact of transport is supported by the temporal variations in the regional distribution of PM 2.5 concentrations, the surface weather maps, and the specific humidity (Figs. 6 and 7).The first evidence is that these sharp PM 2.5 build-up events were unique Introduction

Conclusions References
Tables Figures

Back Close
Full to Beijing among all the 8 cities around/in the NCP (Fig. 6).Chengde and Zhangjiakou are situated to the north of NCP with mountains in between (Fig. 6a).Among the NCP cities, Beijing is located at the northern tip, with mountains to the north and west shielding the city (Fig. 6a2).When conditions favor transport of clean air from north or northwest (i.e. with the advent of a cold air current), Beijing is the first one among NCP cities to embrace it, which resulted in a sharp drop of PM 2.5 concentrations.In this case, PM 2.5 levels in Beijing became similar to the upwind cities, i.e., Chengde and Zhangjiakou (yellow solid circles; Fig. 6b1).However, these cold air currents were too weak to go further, leaving the rest NCP cities unaffected.Not surprisingly, the influence of these weak cold air currents soon receded and the polluted air parcels were transported back to Beijing, which lead to a sharp increase in the PM 2.5 level similar as the rest NCP cities (e.g., Shijiazhuang, Baoding, Tianjin, Langfang, and Tangshan) (yellow solid circles; Fig. 6b2 and b3).
In accordance with the above description, surface weather maps showed that the sharp PM 2.5 increase/decrease events in Beijing during January 2013 were always accompanied with quick transition between low/high pressure systems.As shown in Fig. 7b, the two sharp drops in PM 2.5 concentration on 11 and 12 January corresponded to a weak high-pressure system developed in the mountains northwest of Beijing, which brought clean air mass into the city.When the high-pressure systems diminished, a low-pressure system developed southwest of Beijing, and the air mass in Beijing was again affected by the regional background pollution, resulting in a sharp increase in PM 2.5 concentration.
The observed variation of the specific humidity, an indicator for the origin of air masses (Jia et al., 2008), also supports our explanation (Fig. 7a).Air masses from the south were usually warmer and wetter than the northern air masses, thus possessing a higher specific humidity.During the rapid changes of PM 2.5 , the trend of specific humidity nicely followed the variations of PM 2.5 (Fig. 7a, pink and yellow rectangles marked periods), which reflected the quick transition of air parcel origins.It has been suggested that the decrease of PBL height will compress air pollutants into a shallow Introduction

Conclusions References
Tables Figures

Back Close
Full layer, resulting in elevated pollution levels (Liu et al., 2013).However, our results indicated that the compression was not really happening.Rather, the decrease of PBL height hindered the vertical mixing of pollutants, resulting in a faster accumulation and higher concentrations.As shown in Fig. 7a, the time lag between variations in PBL and its effects on PM 2.5 concentration is a clear evidence demonstrating that the PBL was not "compressing" air pollutants into a shallower layer.Otherwise, concurrent increase in PM 2.5 will be found during the decrease of PBL height.

Formation of secondary aerosols
Compared with clean conditions, the hazy days are characteristic of weaker radiation and higher RH.The RH depends on the synoptic conditions while the radiation reduction is due to the direct radiative effects of aerosol particles (Crutzen and Birks, 1982;Ramanathan and Carmichael, 2008;Ramanathan et al., 2001;Cheng et al., 2008;Wendisch et al., 2008).Secondary aerosols (inorganic and organic) are major components in fine particles in China (Yang et al., 2011).In this section, we will evaluate the impact of changes in radiation and RH on the formation of secondary aerosols.

Weakened photochemistry
The radiative forcing imposed by aerosol particles is particularly strong during haze episodes in Beijing because of extremely high particle concentrations.During haze episodes, the amount of solar radiation reaching the ground was low (down to 2.77 MJ m −2 , 13 January) (Fig. 8a), rendering high photochemical activity impossible.
The reduction of radiation intensities will change the atmospheric photochemistry and oxidant concentrations (hydroxyl radical (OH) and ozone (O 3 )), which will consequently change the production and aging of secondary organic aerosols (SOA) (Hallquist et al., 2009;Jimenez et al., 2009).Introduction

Conclusions References
Tables Figures

Back Close
Full Usually, we can tell the increase of SOA formation from its elevated concentrations.During the haze event, however, the reduced boundary layer height (dilution/mixing effect) itself is already enough to increase pollutant concentrations.For example, concentrations of EC, which is merely of primary origin, continuously increased from clean to heavily polluted conditions (Fig. 8b1).Therefore, we used the ratio of OC to EC to account for the boundary layer effect.Though OC concentrations increased with pollution levels (Fig. 8b2), the OC/EC ratios show a decreasing trend as air pollution become more severe (Fig. 8b5).Compared to OC/EC of 10.8 under clean conditions, the ratios decrease to 6.7 ∼ 7.4 for polluted conditions.Such decreasing trend suggests that the radiation attenuation suppressed SOA formation.
Besides the total OC concentrations, investigations of secondary organic carbon (SOC) lead to the same conclusion.The SOC was estimated using the EC-tracer method (Lim and Turpin, 2002).Briefly, SOC was estimated using these formulae: The basic assumptions and underlying principles of this method are discussed in Lim and Turpin (2002) and Lin et al. (2009).In our study, data pairs with the lowest 10 % percentile of ambient OC/EC ratios were used to estimate the primary OC/EC ratio (Fig. 9a).York regression (York et al., 2004) was used to estimate the intercept N and the slope, i.e., values of (OC/EC) pri, according to Saylor et al. (2006).
The SOC/OC ratio dropped from 30 % under clean conditions to 15 % during heavily polluted periods, suggesting a reduced production rate due to weakened photochemical activity.The relatively small proportion of SOC in total OC (∼ 30 %, Fig. 9b) also supported weakened photochemical activity during haze episodes.This estimation was consistent with those from off-line PM 2.5 samples from winter 2009 (∼ 32 %, Cheng et al., 2011), and positive matrix factorization-resolved NR-PM 1 samples in winter 2011-2012 (∼ 31 %, Sun et al., 2013b).Overall, the SOC to OC ratio during haze Introduction

Conclusions References
Tables Figures

Back Close
Full episodes was much lower than that recorded in summer (roughly 50 %; Cheng et al., 2014).Moreover, changes in the diurnal profile of SOC/EC with pollution level (Fig. 8b6) also reflected the suppressed SOA formation due to weakened photochemistry.Like OC, SOC was also scaled by EC to account for the boundary layer effect.In clean periods, average diurnal trends in SOC/EC have both an afternoon maximum and a nighttime maximum.These two peaks both diminished as the pollution got worse, and instead a stable diurnal pattern was observed during heavily polluted periods.Decrease in the nighttime SOC/EC peaks could be explained by the more stable night temperatures (and corresponding stable RH) with rising pollution levels (Fig. 5) (Lim and Turpin, 2002), while afternoon SOC/EC peaks were typically associated with photochemical activity.The weakening of the afternoon peak indicated a reduction in photochemical activity.

Enhanced heterogeneous chemistry
Unlike OM, contributions of sulfate and nitrate to PM 2.5 were increasing during the haze events (Fig. 2).Again, we used their ratios to EC to account for the boundary layer effect.Significant increase of SO 2− 4 /EC and NO − 3 /EC ratios were found from clean periods (3.03 and 3.33, respectively) to heavily polluted periods (6.35 and 5.89, respectively), suggesting enhanced chemical productions.The SOR and NOR (molar ratio of sulfate or nitrate to sum of sulfate and SO 2 or nitrate and NO 2 ) have been used as indicators of secondary transformation (Sun et al., 2006).The fact that SOR and NOR increased much more rapidly than SO 2 and NO 2 as pollutions became more severe (Column 3 in Fig. 10), is another evidence of elevated secondary formations of sulfate and nitrate during severe haze events.
Both gas-phase and heterogeneous reactions could contribute to the formation of sulfate and nitrate from SO 2 and NO 2 , and thus elevating the SOR and NOR.Sulfate is formed through oxidation of SO 2 by gas-phase reactions with OH (Stockwell and Calvert, 1983;Blitz et al., 2003) and stabilized Criegee intermediate (which is formed 17920 Introduction

Conclusions References
Tables Figures

Back Close
Full by O 3 and alkenes) (Mauldin et al., 2012), and by heterogeneous reactions with dissolved H 2 O 2 or with O 2 under the catalysis of transition metal (Seinfeld and Pandis, 2006).Nitrate formation is dominated by the gas-phase reaction of NO 2 with OH during daylight, and the heterogeneous reactions of nitrate radical (NO 3 ) during nighttime (Seinfeld and Pandis, 2006).
As discussed in Sect.6.1, gas-phase production of sulfate and nitrate is unlikely to be important owing to the weakened photochemistry, which implied low gas-phase oxidant concentrations of OH and ozone (Y.Wang et al., 2014;Zhao et al., 2013).This was further supported by the weak correlation between SOR and temperature (Fig. 11a), for gas-phase SO 2 oxidation by OH radicals was proposed to be strongly correlated with temperature (Seinfeld and Pandis, 2006;Sun et al., 2006;Blitz et al., 2003).Thus we infer that the rapid increase in SOR and NOR during polluted periods (Fig. 10) was mainly due to heterogeneous reactions, especially reactions associated with aerosol water.
Dependence of SOR and NOR on RH (Fig. 11) also supported our inference.Both SOR and NOR were constant under dry conditions (RH < 50 %) (Fig. 11a and b) while started increasing when RH > 50 %, resulting in average values around 0.34 and 0.28 at RH 70-80 %, respectively.This suggests important contributions from heterogeneous reactions with abundant aerosol water under wet conditions (Sun et al., 2013a).The observed SOR value was high compared with previously reported values of 0.24 (Wang et al., 2006) and 0.29 (Zhao et al., 2013) during hazy days in Beijing.The NOR values for this study were higher than spring hazy days in 2001-2004 (0.22;Wang et al., 2006), but significantly lower than for the hazy episode in January 2010 (0.51; Zhao et al., 2013).
Concerning the SOA formation, the contribution of heterogeneous reactions might be possible, but it should be much less significant than for sulfate and nitrate.For RH > 50 %, SO 2− 4 /EC and NO − 3 /EC ratios rose significantly (Fig. 11d) while SOC/EC ratios remained constant (Fig. 9c).By using HOA (hydrocarbon-like organic aerosol) Introduction

Conclusions References
Tables Figures

Back Close
Full instead of EC, Sun et al. (2013a) found similar phenomena.Apparently, SOC does not have a heterogeneous formation pathway as effective as those of sulfate and nitrate.

Conclusions
The severe haze pollution during January 2013 was not a Beijing-localized phenomenon.Rather, it was the result of local pollutants superposed on background regional pollution, which affected the whole NCP.Although pollutant emissions were high, there was no abrupt enhancement in 2013.The occurrence of the severe winter haze resulted from stable synoptic meteorological conditions over a large area of northeastern China.Surface weather maps from hazy periods were characterized by a weak high-pressure center northeast of Beijing, while the termination of a haze episode was always accompanied by the Siberian Anticyclone (Xu et al., 2011;Jia et al., 2008;Liu et al., 2013).Rather than chemical productions, transport of polluted regional air masses was suggested as the key process leading to the sharp increases of aerosols concentrations.Beijing pollution was temporarily flushed away by strong winds associated with the arrival of a weak cold air current, but as its influence weakened, the polluted regional air mass readily reoccupied the Beijing area, resulting in an apparent rapid build-up of PM 2.5 .This hypothesis was supported by data on the PM 2.5 levels around Beijing, specific humidity and PBL height, as well as surface weather maps.
Atmospheric chemistry and physics during severe haze pollutions are illustrated in a conceptual model (Fig. 12).With the onset of stable synoptic conditions, RH rises, primary pollutants began to accumulate and regional pollution began to form.If the stable conditions last long enough, PM 2.5 build-up occurs, such that solar radiation is reduced at the ground.This inhibits surface temperature fluctuation, making easier the formation of inversed layer and rending the atmosphere into a more stable condition.Meanwhile, photochemical activity is weakened under low solar radiation, and secondary aerosol formation via this pathway becomes less important.However, un-Introduction

Conclusions References
Tables Figures

Back Close
Full der high RH, and enhanced precursor concentrations, heterogeneous reactions play a more important role, especially those associated with the aerosol water phase.This results in the rapid build-up of sulfates and nitrates, enhancing PM 2.5 pollution.The sulfates and nitrates absorb more water into the particulate phase, promoting subsequent heterogeneous reactions.This cycle terminates with the incursion of a strong cold front, usually the Siberian Anticyclone.Introduction

Conclusions References
Tables Figures

Back Close
Full  Full      .Variation of SO2, NO2, SOR, NOR, SO4 2-and NO3 -with pollution level."C", "S", "P", "H" refer to "clean", "slightly polluted", 676 "polluted" and "heavily polluted", respectively.Normalized X in Column 3 refers to the average concentration of X in any pollution level, scaled 677 by its average concentration during clean periods.4 and NO − 3 with pollution level."C", "S", "P", "H" refer to "clean", "slightly polluted", "polluted" and "heavily polluted", respectively.Normalized X in Column 3 refers to the average concentration of X in any pollution level, scaled by its average concentration during clean periods.Introduction

Conclusions References
Tables Figures

Back Close
Full episodes in the winter of 2012-2013 engulfed Beijing, as well as other cities in southeastern China, causing one of the worst atmospheric pollution events Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | 5 .A NIOSH (National Institute for Occupational Safety and Health) temperature protocol was used and the calculation discrepancy under high ambient concentrations Discussion Paper | Discussion Paper | Discussion Paper | ). Monthly average PM 2.5 concentration reached 121.0 µg m −Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

Figure 2 .
Figure 2. Percentile composition of major components in PM 2.5 with respect to (a) pollution level and (b) PM 2.5 mass concentration.

Figure 3 .
Figure 3. Model simulation results under different scenarios.(a) Base scenario.Actual 2013 emission and 2013 meteorology data were used.(b) 2012 meteorology data were used, and (c) 2012 emissions were used.The difference caused by meteorology (d; equivalent to a-b) and emission (e, equivalent to a-c) is also shown.

Figure 4 .Figure 5 .Figure 7 .Figure 9 .Figure 9 .
Figure 4. Surface weather maps (a, c, e, g) and PM 2.5 concentrations (b, d, f) of the North China Plain on 12 January LT 18:00 (a, b), 18 January LT 20:00(c, d), 29 January LT 13:00 (e, f), and 1 January LT 8:00 (g).The location of Beijing is indicated as a white star on the weather maps, and as the shaded area on the PM 2.5 concentration maps.PM 2.5 concentrations in Beijing at the four selected time points are also shown on the left for reference.

Figure 11 .Figure 12 .Figure 12 .
Figure 11.Importance of heterogeneous chemistry in sulfate and nitrate formation.(a-b) SOR and NOR plotted against RH, colored with temperature.(c-d) EC-scaled precursors (SO2 and NO2) and product (SO4 2-and NO3 -) plotted against RH.Change of EC concentration with RH level was also shown for reference.

Table 1 .
General information on severe haze episodes in January 2013.