Understanding the Regional Transport Contributions of Primary and Secondary PM2.5 Components Over Beijing During a Severe Pollution Episodes

Home , Xingtai

Understanding the Regional Transport Contributions of Primary and Secondary PM2.5 Components over Beijing during a Severe Pollution Episodes

Wei Wen1,2, Xiaodong He1*, Xin Ma3**, Peng Wei4, Shuiyuan Cheng5, Xiaoqi Wang5, Lei Liu2

1Institute of Urban Meteorology, China Meteorological Administration, Beijing 100089, China 2 Key Laboratory of Atmospheric Chemistry, China Meteorological Administration, Beijing 100081, China 3 National Meteorological Center, Beijing 100081, China 4 Chinese Research Academy of Environment Science, Beijing 100012, China 5 Key Laboratory of Beijing on Regional Air Pollution Control, Beijing University of Technology, Beijing 100124, China

ABSTRACT

This study applied the CAMx model to study the regional transport of various PM2.5 components in Beijing during a severe pollution episodes. The results revealed that during the episodes, Beijing had the average PM2.5 pollution value of –3 –3 119 µg m . It was 1.58 times of the PM2.5 national air quality standard (75 µg m Level II). The wind speed was low (< 2 m s–1) and relative humidity reached 98%. The anticyclone in Eastern China showed weak local flow fields and southerly winds at the surface and strong temperature inversion under 1000 m, which promote pollution accumulation. The 2– – contribution of monthly regional transport to primary PM2.5 components and SO4 , NO3 , and secondary organic aerosol concentrations in Beijing were 29.6%, 41.5%, 58.7%, and 60.6%, respectively. The emissions from Baoding had the greatest effect on the primary components of PM2.5 (6.1%) in Beijing. The emissions from Tianjin had the greatest influence on the secondary components of PM2.5 concentrations. These values indicated that the secondary components of Beijing’s PM2.5 are more easily affected by transboundary transport than are the primary components. The present findings suggest that control strategies for PM2.5 pollution should include coordinated efforts aimed at reducing secondary aerosol precursors (SO2, NOx, and VOCs) from long-range transport and local generation in addition to primary particulate emissions.

Keywords: PM2.5 pollution; Regional transport; Secondary component; Air quality modeling.

INTRODUCTION NOx, and volatile organic compounds [VOCs], respectively) through not only photochemical reaction but also aqueous PM2.5 pollution is a severe environmental problem in and heterogeneous reaction etc. (Huang et al., 2014). PM2.5 China. It poses a considerable threat to human health because pollution can be controlled through coordinated regional it comprises various metallic elements and toxic organic control. Emphasis should be placed not only on controlling materials (Cao et al., 2012). PM2.5 pollution is a complex primary PM2.5 pollutants from local emissions but also on regional transport problem, as the particles can be transported coordinating the control of precursor emissions (SO2, over long distances (Hatakeyama et al., 2011; Squizzato et VOCs, and NOx) in the region. Beijing is the political and al., 2012; Tao et al., 2012; Lang et al., 2013). Previous cultural center of China. It’s located in the north of China, studies have reported that in North China, a large fraction the most polluted region. This region is characterized by high of PM2.5 is attributable to secondary aerosols (Huang et al., concentrations and frequencies of PM2.5 pollution (MEP, 2014; Zheng et al., 2015). Secondary components, such as 2015). The pollution in Beijing has attracted increasing 2– – SO4 , NO3 , and secondary organic aerosols (SOAs), are attention from the government and public of China. To transformed from precursor gases in the atmosphere (SO2, improve the overall air quality and control PM2.5 pollution in the region, a series of PM2.5 pollution control strategies were implemented by the government (China State Council 2013). For example, in 2014 and 2015, the government executed * Corresponding author. strict emission control strategies to ensure that the Asia- Tel.: +86 10 68400750 Pacific Economic Cooperation meeting and 70th anniversary E-mail address: [email protected] of the victory of the Anti-Japanese War celebration were ** Corresponding author. successfully. Emission control in Beijing and surrounding Tel.: +86 10 58993295 provinces (Hebei, Shanxin, Inner Mongolia, Shandong, and E-mail address: [email protected] Henan) has been correspondingly implemented (Wen et al.,

Wen et al., Aerosol and Air Quality Research, 18: 1720–1733, 2018 1721

2015; Wang et al., 2016). components to local PM2.5 concentrations. Several studies and projects have investigated the regional In this study, the offline model CAMx was applied to 2– transport of PM2.5. Chuang et al. (2008) reported that during study the transport contribution of primary PM2.5 and SO4 , 2– – a typical pollution episode in Taiwan, the SO4 concentration NO3 , and SOA. The particulate source apportionment (35%) was contributed by Shanghai. Wang et al. (2008) technology (PSAT) was implemented in CAMx. This model showed that 28.1% and 59.5% of PM pollution in urban was used to analyze the contributions of emission categories and suburban Beijing was contributed by PM transport from and different source regions. The model not only determines surrounding regions. Streets et al. (2007) found that 50%– the source region and contributions but also provides source 70% of PM pollutants in Beijing were transported from the apportionments for both primary and secondary particulate southern areas of Hebei Province. Wang et al. (2014) used matter. Furthermore, the relative importance of different the CAMx/PSAT (Comprehensive air quality model with source regions can be ranked and further provide a major extensions/ particulate source apportionment technology) contribution amount, which is important for air quality model to analyze a heavy pollution process (November management. Studies on the transport of primary and 2010) in Shanghai; the results revealed that nearly 50% of secondary PM2.5 components can provide scientific and PM2.5 pollution was from surrounding regions. Wang et al. technological support for the collaborative prevention of (2012) applied the CMAQ model in Hebei Province to air pollution. investigate haze pollution and found that regional transport accounted for approximately 35% and 41% of PM2.5 Model Setting and Application pollution in Shijiazhuang and Xingtai, respectively. Lin et Model System al. (2016) extended the CAMx model and reported that in In the present study, the transport components of primary August, local sources respectively accounted for 23.8% and secondary PM2.5 in the Beijing region were simulated and 16.6% of anthropogenic and biogenic SOAs in Beijing. using the CAMx model. The meteorological factors were Furthermore, Wang et al. (2015) applied the MM5–CMAQ simulated using The Weather Research and Forecasting model to estimate the regional source contribution to PM2.5 model (WRF; Version 3.3). The WRF initial conditions were concentrations in most polluted cities in Hebei Province. used as the final global tropospheric analysis data, which They reported that regional transport plays an important were provided by the National Centers for Environmental role in PM2.5 pollution and that the regional joint air Prediction. The PSAT module in CAMx is a framework for pollution controls are crucial. Concurrently, the long-distance apportioning six categories of PM components and gaseous transport of particulate matter and also the species has been precursors. PSAT has been widely used in the CAMx model widely investigated. For example, PM2.5 sulfate in Beirut is (Fann et al., 2012; Zhang et al., 2013). In the present study, mainly from the Eastern European region (Saliba et al., two nested grid architectures were designed to implement 2007). A study reported that pollution in Asia affects the air the modeling system (Fig. 1(a)). During October 2014, quality of the United States and Canada, showing that the several typical severe pollution episodes occurred in the long-distance emission from Asia can be transport and BTH region. The model was applied for a 35-day period in transmitted across the Pacific Ocean (Heald et al., 2006). Beijing, starting from September 25, 2014. The first 5 days However, most studies on the transport of particulate matter were considered as spin-up and were excluded from the are focused on total PM2.5. Few studies have used model analysis. Beijing is enclosed by Hebei Province except to the approaches to comprehensively analyze the transport of southeast, where Tianjin Province is located. The modeling primary and secondary PM2.5 components separately in the domains had the spatial resolution for the modeling Beijing-Tianjin-Hebei (BTH) region under one study domains are 27 km and 9 km respectively. The emission framework. The locations where secondary PM2.5 forms source regions included Beijing and other regions, which might differ from where its precursors are emitted, because were defined on the basis of geographical borders. The the precursors might not be directly converted into PM2.5 emission inventories were obtained from the Multi resolution after emitted into the atmosphere. The time for secondary Emission Inventory for China of the year 2012, which was particle formation to occur depends on the abundance of developed by Tsinghua University. We have previously precursor, temperature, humidity, and other factors. PM2.5 reported additional details of the emission inventory, is the mixture of four aspects (1) Locally emitted primary physical options, and model setting (Wen et al., 2015; Wen PM2.5 and secondary components; (2) Secondary PM2.5 et al., 2016). WRF Version 3.3 was used to analyze the components from local precursor emissions; (3) Secondary meteorological conditions. The physics options included PM2.5 components from precursor emissions transport from the New Thompson, Goddard short-wave, rapid radiative surroundings, which later formed into PM2.5 components in transport model long-wave, Yonsei University planetary that focused region; (4) Primary PM2.5 and secondary boundary layer meteorology, Noah land surface model, and components transport from surrounding regions that already New Grell schemes. formed there. Therefore, the amount of PM2.5 in a specific region is the mixture of PM2.5 formed and transported from Source Apportionment Technology different sources, which can be distinguished according to The particulate source apportionment technology (PSAT) source regions and emission categories. To control PM2.5 module in CAMx was designed for the apportionment of pollution, it is necessary to determine the contribution of primary and secondary PM2.5 components: Sulfate (SO4), the regional transport of primary and secondary PM2.5 Particulate nitrate (NO3), Secondary organic aerosol (SOA),

1722 Wen et al., Aerosol and Air Quality Research, 18: 1720–1733, 2018

(a) Double nesting domains utilized in model simulation

(b) Different source regions for PSAT in the domain Fig. 1. Nesting domains and PSAT source regions used for model simulation.

Wen et al., Aerosol and Air Quality Research, 18: 1720–1733, 2018 1723 and Six categories of primary PM (Elemental carbon, grid cells that are averaged together to provide multi-cell Primary organic aerosol, Crustal fine, Other fine, Crustal average tracer concentrations. The entire domain was divided coarse, and Other coarse). In general, a single tracer can into 14 source regions, namely Beijing, Tianjin, Zhangjiakou, track primary PM species whereas secondary PM species Chengde, Qinhuangdao, Tangshan, Baoding, Langfang, require several tracers to track the relationship between Cangzhou, Shijiazhuang, Hengshui, Xingtai, Handan, the gaseous precursors and the resulting PM. The PSAT tracers non-BTH region, as shown in Fig. 1(b). These regions were for each type of PM are listed: Sulfur (Primary SO2 emissions defined according to their geographical borders. Detailed and Particulate sulfate ion from primary emissions plus PSAT algorithms are reported in the manual of the CAMx secondarily formed sulfate), Nitrogen (Reactive gaseous model (http://www.camx.com). nitrogen including primary NOx emissions plus nitrate radical, nitrous acid, dinitrogen pentoxide, Gaseous peroxyl RESULTS AND DISCUSSION acetyl nitrate plus peroxy nitric acid, Organic nitrates, etc.), Secondary Organic (Aromatic secondary organic aerosol Model Verification precursors, Isoprene secondary organic aerosol precursors, In this study, the model simulation results were assessed etc.). The source apportionment results actually includes the in terms of normalized mean bias (NMB), normalized mean contributions from two aspects: (1) transport of precursors gross error (NME), and correlation coefficient (RC). Daily from surrounding regions to Beijing, e.g., SO2 for sulfate, average PM2.5 concentration monitoring data for Beijing NOx for nitrate, VOC for SOA, which later transformed into were collected from the China National Environmental PM2.5 components in Beijing; (2) direct transport of specific Monitoring Centre (CNEMC) for the target month; these PM2.5 components that formed in surrounding regions. The data included an average concentration of 23 monitoring 2– – model estimates the contribution of specific emission sites. The simulated particle components (SO4 , NO3 , SOA) sources to receptor PM2.5 concentrations. It can determine were compared with the monitoring data. The components the spatial and temporal relationship between ambient PM2.5 monitoring data were from our previous study paper, which levels and various emission source regions and categories, sampling site was at the Beijing Normal University (BNU) which can be used to rank the relative importance of various (wen et al., 2015). Table 1 presents the statistical results of source regions to further develop control policies (Wu et the comparison the CAMx simulation results with the daily al., 2013). On the basis of the PSAT results, the important average concentrations of the monitored PM2.5 and its sources can be controlled to achieve certain air quality components. The simulated meteorological parameters were targets. PSAT has been used as a source apportionment tool also evaluated against observations. The meteorological in many studies. The PSAT module can provide apportion data were obtained from the Meteorological Information of particulate matter in each single-grid cell among different Comprehensive Analysis and Process system of the China sources and regions. The different source groups in the Meteorological Administration. The evaluation results of PSAT methodology are defined in terms of their geographical the meteorological parameters affecting air pollution, region and emission category. In the present study, Beijing including temperature (T2), relative humidity (RH), wind was selected as the receptor region. The CAMx receptor speed at 10 m (WSP10), and sea-level pressure (P) are also region setting was not a point in Beijing, but a group of provided in Table 1. The near-surface temperature simulation

Table 1. Comparison of simulated results with monitored data (24 h average). –1 WSP10 Simulation (m s ) 2.3 P Simulation (hpa) 989.5 Monitored (m s–1) 1.9 Monitored (hpa) 1020.6 NMB 19.22% NMB –3.04% NME 28.94% NME 3.04% RC 0.76 RC 0.99 T2 Simulation (°C) 13.58 RH Simulation (%) 50.4 Monitored (°C) 13.28 Monitored (%) 65.5 NMB 1.75% NMB –21.09% NME 9.13% NME 21.32% RC 0.84 RC 0.86 –3 2– –3 PM2.5 Simulation (µg m ) 136.7 SO4 Simulation (µg m ) 8.09 Monitored (µg m–3) 131.2 Monitored (µg m–3) 10.56 NMB 4.50% NMB –6.38% NME 29.60% NME 48.56% RC 0.89 RC 0.68 – –3 –3 NO3 Simulation (µg m ) 30.62 SOA Simulation (µg m ) 4.8 Monitored (µg m–3) 35.40 Monitored (µg m–3) 5.2 NMB –13.51% NMB –92.28% NME 63.46% NME 93.11% RC 0.82 RC 0.35

1724 Wen et al., Aerosol and Air Quality Research, 18: 1720–1733, 2018 results showed a closer fit to the observations. The NMB These acceptable WRF-simulated meteorological results in and RC for the domain were 1.75% and 0.84, respectively. addition to emission categories can drive the air quality WSP10 was slightly over predicted with NMBs (19.22%). model. Fig. 2 presents the simulated and monitored PM2.5 RH and sea-level pressure were in good agreement with the concentrations in Beijing. PM2.5 pollution was much more RC at 0.86 and 0.99, respectively. In general, the comparison frequent in October 2014 and was the highest on October –3 results indicate that an acceptable agreement between the 9, 2014 (330 µg m ). The value was 4.4 times of the PM2.5 simulated and observed concentrations has been achieved national air quality standard (75 µg m–3 Level II). The trend the guidelines of the U.S. Environmental Protection of the simulated values matched well with the observation Agency (2007). WRF yielded acceptable performance. data. The average monthly spatial distribution of PM2.5

Fig. 2. Comparison between the simulated and observed daily concentrations of PM2.5.

(simulated) (monitored) –3 Fig. 3. Average simulated and monitored monthly spatial distribution of PM2.5 pollution (µg m ).

Wen et al., Aerosol and Air Quality Research, 18: 1720–1733, 2018 1725 pollution simulated data and the monitored data were higher than those in the northern BTH region. In the northern plotted in Fig. 3, which also confirmed the well simulation BTH region, Langfang had the highest PM2.5 monthly average –3 and observation trends. The correlation coefficient between value of 131 µg m . Beijing had a PM2.5 pollution value of –3 the simulated and observed PM2.5 concentrations was 0.89. 119 µg m , which was 1.58 times of the PM2.5 national air The simulation results obtained using the CAMx model can quality standard (75 µg m–3 Level II). In the southern BTH be considered to reflect the change in PM2.5 concentrations region, Xingtai had the highest PM2.5 monthly average value in October. of 167 µg m–3, which was 2.22 times of the national standard. The NMB value for the CAMx simulation of PM2.5 in In the eastern BTH region, Tangshan yielded the highest –3 Beijing was 4.5%. The NME of CAMx was 29.6%. It was PM2.5 monthly average value of 106 µg m . According to the slightly overestimated by CAMx. The model did not simulate calculation data, the average monthly PM2.5 concentrations well during heavy pollution events. For example, the observed were lowest in Zhangjiakou in the BTH region (26 µg m–3). –3 PM2.5 concentration was 330 µg m in Beijing in October The emissions, meteorological conditions, and terrain th 9 2014; however, the simulated PM2.5 concentration was could be obvious reasons for such a spatial distribution of –3 only 279 µg m . The simulation NMB and NME values of PM2.5 concentrations. Industrial activities in Shijiazhuang secondary components were relatively larger than PM2.5 and Xingtai in the southern BTH region is intensive (i.e., concentration. The most of correlation coefficients of cement, coke, and metallurgical industries); thus, PM2.5 secondary species between simulation and monitored data emissions were higher than those in the northern BTH 2– - were strong (SO4 with 0.68, NO3 with 0.82), except SOA region according to the emission inventory (i.e., Shijiazhuang –1 –1 –1 –1 results (0.35). The model underpredicted PM2.5 concentrations [9.0 ton year area ], Xingtai [9.6 ton year area ], and –1 –1 on heavy polluted days and PM2.5 secondary species Zhangjiakou 2.7 ton year area ]) (Wen et al., 2015). possibly due to its mechanisms. These underpredictions Meanwhile, the atmosphere over the southern areas of Hebei may also reflect the meteorological simulation bias, which was highly stable. Shijiazhuang and Xingtai are located in oversimulated of the WS and underestimated RH. The the Taihang Piedmont Plain. Southerly or easterly air meteorological simulation bias resulted in accelerating currents over this region coincide with the westerlies, which pollutant diffusion. The model lacks chemical reaction cross the Taihang Mountain range. These air flows sink and processes such as missing model calculated oxidants reactions converge, resulting in a highly stable atmosphere. This unique and stat-of-science concerning SOA formation (Li et al., meteorological factor and terrain tend to trap precursor 2015), which may also lead to some underestimation. gases as well as primary PM2.5 emissions, which contributes According to Morris et al. (2006), the performance of the to pollution in the region. Even with the same emission in the simulation of organic carbons are usually poor. The densities as in other regions, the southern BTH region NMB of carbons simulation results usually over 100% typically experiences heavier PM2.5 pollution than do other (Zhang et al., 2013). These underpredictions are the inherent regions (Wang et al., 2013). characteristics of the model (Wang et al., 2013). The model The meteorological conditions were highly conducive to results can clearly simulate the PM2.5 pollution process PM2.5 pollution. The meteorological parameters in the typical despite of the underestimations. The CAMx simulation pollution episodes (6–9th) in October were rigorously results in this study were acceptable compared with those analyzed (Fig. 4). The meteorological parameters affecting air reported by related studies (Liu et al., 2010; Wu et al., 2013; pollution, including wind direction, WS, total cloud amount, Zhang et al., 2013). and RH. As shown in Fig. 4, the heavy air pollution can be divided into several stages PM2.5 pollution ascent, continuous Regional PM2.5 Pollution Characteristics and high-value stage and descent. As shown in Fig. 5(a), in the Meteorological Conditions significant ascent stage of PM2.5 concentrations (October The meteorological conditions and PM2.5 concentration in 6–9, 2014), the wind direction changed between 0° and 350°. –1 the BTH region affected the transport of PM2.5 components The WS was relatively low (< 2 m s ). Static wind occurred in Beijing. The data (CNEMC) on PM2.5 concentration at night or in early morning (UTC). Under stable weather monitoring in all cities in the BTH region during October conditions, the RH and cloud amount had significantly were determined (Fig. 4). increased [Fig. 4(b)]. These conditions were conducive to The cities in the BTH region were divided into three pollutant accumulation in the local areas, which resulted in categories (northern, eastern, and southern) for calculation. a significant increase in PM2.5 concentrations. In the The daily PM2.5 concentrations in these cities are plotted in continuous high-value stage of PM2.5 (October 8–10, 2014), Fig. 4. The BTH region experienced several severe PM2.5 the wind direction changed between 50° and 200°. The pollution episodes in October 2014. The monitoring data of easterly and southerly winds WS was still low (< 2 m s–1). different cites showed similar trends of PM2.5 changes with The clouds were more stable, and the RH increased slowly. time, indicating that the BTH region experienced similar The RH reached 98%, which was close to saturation and PM2.5 pollution processes in October. This observation also was conducive to secondary pollutant formation. In the revealed that PM2.5 pollution can be characterized as a PM2.5 concentration descent stage (October 11–12, 2014), regional pollution concern. Severe pollution episodes occurred the wind direction was mainly toward the north. The WS during October 2014, notably from the 6th to the 12th, 16th increased from 2 m s–1 to 6 m s–1; the RH was 12%–20%. st nd th to the 21 , and 22 to the 26 . Fig. 4 shows that PM2.5 The amount of cloud change was scarce, and the PM2.5 concentrations in the southern BTH region were typically concentration decreased rapidly. During heavy pollution,

1726 Wen et al., Aerosol and Air Quality Research, 18: 1720–1733, 2018

Fig. 4. Daily average concentration monitoring of PM2.5 in the Beijing-Tianjin-Hebei region. obvious variations were observed in the WS, wind direction, center was located over the North China Plain. Under the high cloud amount, and RH in each stage of PM2.5 pollution. The pressure control, the wind was weaker, with speeds less results reveal that pollution was dominated by meteorological than 4 m s–1. The north was dominated by weak local flow factors. Furthermore, the large-scale synoptic pattern fields and southerly winds, while the south was controlled by influences the changes in meteorological factors, which weaker northerly currents. Furthermore, the anticyclone in results in changes in PM2.5 concentration. The special weather Eastern China showed a weak local flow fields and southerly background often results in large-scale fog haze and heavy winds at the surface and strong temperature inversion near pollution events in the area. To verify this point, the large- the ground (Fig. 7), which promote pollution accumulation. scale synoptic pattern were studied using the WRF model. In the severe pollution episodes, the surface weather system Fig. 6 shows the daily average large-scale synoptic pattern in the BTH region was highly stable. The anticyclone the PM2.5 concentration ascent stage, which included sea- provides a stable large-scale synoptic pattern for the heavy level pressure and WSP10 on the ground. The eastern lands pollution events in autumn. The weaker wind field on the of China were controlled by high pressure, and the pressure ground and the subsidence inversion layer at a high altitude

Wen et al., Aerosol and Air Quality Research, 18: 1720–1733, 2018 1727

(a)

(b)

Fig. 5. Variations in meteorological factors over time in Beijing.

(0–200 m) were beneficial for the local accumulation. Since of Beijing’s PM2.5 pollution are more easily affected by October 11, 2014, the main weather situation effect changed transboundary transport than are primary components. This in Eastern China. The high pressure was from Mongolia. behavior of secondary components was also reported by a The frontal of high pressure control almost the entire previous study that revealed that the average contribution continent. This meteorological background in Northern of local emissions to primary PM2.5 was 64% and that less China was consistent with systemic northerly winds. The than 10% of SOA was contributed by local emission sources wind was strong, reaching 10 m s–1, which is highly in Paris (Skyllakou et al., 2014). This is mainly because favorable for pollution removal. It was the main systematic component formation is more complex and can be affected pattern for the PM2.5 concentration descent. by other factors. First, secondary components are mainly enriched in smaller particles, which are conducive to long- Regional Transport of Primary and Secondary PM2.5 distance transport (Li et al., 2013). Second, various ways of 2– Components secondary particles are transported (i.e., SO4 ). (1) Primary Table 2 presents the contribution of monthly transport to sulfates (directly emitted from various sources) are transported 2– – primary PM2.5, SO4 , NO3 , and SOA concentrations in from other region. (2) Secondary sulfates are formed by Beijing during the simulation months from the surrounding chemical reactions in the region and are transported to regions. Beijing. (3) SO2 and NOx are transported from the region 2– – Significant variation was observed in the contribution ratios to Beijing, where SO4 , NO3 , and SOA are formed through among different pollutant compositions. The contribution for oxidation. Table 2 shows that the effect of transboundary different components transported from surrounding regions transport from the surrounding regions was the highest on to Beijing was 30%–61%. The contribution of local the concentration of SOA (61%) among the studied fine 2– – emission sources to primary PM2.5, SO4 , NO3 , and SOA particles components in Beijing. Identical results were concentrations in Beijing were 70.4%, 58.5%, 41.3%, and reported in a previous study, in which a simulation indicated 39.4%, respectively. Furthermore, the contribution of monthly that long-distance sources dominated SOA in Beijing (Lin 2– – regional transport to primary PM2.5, SO4 , NO3 , and SOA et al., 2016). The reasons are as follows: (1) Compared concentrations in Beijing were 29.6%, 41.5%, 58.7%, and with the surrounding region, Beijing has modest VOC 60.6%, respectively, indicating that secondary components emissions; the formation of SOA from freshly emitted

1728 Wen et al., Aerosol and Air Quality Research, 18: 1720–1733, 2018

. . . Otc Otc Otc - - -

th th th 7 10 13

. m wind field m -

. . . Otc Otc Otc - - -

th th th 6 9 12 level pressure level and 10 - Sea

. 6

. Fig

. . . Otc Otc Otc - - -

th th th 5 8 11

Wen et al., Aerosol and Air Quality Research, 18: 1720–1733, 2018 1729

Fig. 7. Vertical and dew-point temperatures in Beijing.

VOCs required several hours (Li et al., 2013). During the abundant emissions of SOA precursors to form semivolatile reaction, VOCs emitted in Beijing were transported out of oxidation products and condense to SOAs (Kim et al., Beijing before oxidation and converted into semivolatile 2011). (3) SOAs comprise all types of secondary compounds, oxidation products, which can condense to form SOA which are formed by the reactions of evaporated primary 2– (Wagstrom et al., 2012; Skyllakou et al., 2014). (2) SO4 organic aerosols, intermediate VOCs, and VOCs. SOA was formed in the atmosphere primarily through the formation processes are very complicated, thus making it very oxidation of SO2, which is mainly emitted from coal difficult to simulate SOA concentration in the atmosphere. – burning. Moreover, NO3 was derived from the emission of The module used in this study may not contain all reaction NOx, which is mainly an exhaust gas from power plants processes, which may result in some errors. The CAMx and vehicles. In contrast, the emission sources of SOA SOA module consists of two parts: gas-phase oxidation precursor gases were highly uncertain; the sources include chemistry that forms condensable gases (CGs), and biogenic sources and those from other anthropogenic areas. equilibrium partitioning between gas and aerosol phases for Although VOCs cannot be transported too far because of each CG/SOA pair. The simulated SOA concentration using their short life span, during transport, they are consistent of the the two-product approach are substantially underestimated,

1730 Wen et al., Aerosol and Air Quality Research, 18: 1720–1733, 2018

Table 2. Regional transport of primary and secondary PM2.5 components (%). 2– – PM2.5 (Primary) SO4 NO3 SOA Beijing 70.4% 58.5% 41.3% 39.4% Baoding 6.7% 6.5% 10.1% 9.6% Tianjin 6.1% 9.4% 12.9% 10.2% Tangshan 3.4% 5.6% 5.8% 4.6% Zhangjiakou 2.2% 3.3% 3.1% 8.5% Shijiazhuang 2.1% 3.5% 6.1% 3.8% Langfang 2.0% 2.8% 4.0% 5.1% Chengde 1.8% 2.9% 3.4% 7.9% Xingtai 1.6% 2.4% 4.0% 2.2% Cangzhou 1.5% 1.2% 3.5% 3.9% Hengshui 1.1% 1.3% 2.3% 2.9% Handan 1.0% 2.3% 2.9% 1.3% Qinhuangdao 0.1% 0.3% 0.6% 0.6%

when compared to the corresponding observations in most transport for Beijing PM2.5 primary , sulfate, nitrate, SOA cases. A large OA source in the free troposphere was not concentration were 22.6, 26.8, 42.0, and 48.0%, respectively. included in the model (Lin et al., 2016). These were lower than those during the periods with PM2.5 The daily variations in the contributions of emission concentration of over 75 µg m–3. During the high pollution –3 sources from the surrounding regions to different PM2.5 levels (i.e., > 75 µg m ), the transportation were 36.7, components in Beijing under different PM2.5 pollution 45.5, 62.0, and 63.7% for PM2.5, sulfate, nitrate, SOA, levels are shown in Fig. 8. During the periods of higher respectively. Moreover the contribution from further south PM2.5 concentrations, the contribution of regional transport of Hebei (e.g. Handan, Xingtai, Shijiazhuang) had a greater to different PM2.5 components was also higher (i.e., on contribution to Beijing pollution, when PM2.5 concentration –3 October 8, 2014: PM2.5 concentration, 258 µg m vs. was in high pollution level. The results listed in Table 3 – regional contribution ratio of NO3 , 68%). Relatively weaker confirm the positive correlation between transport from polluted days corresponded to lower regional contributions regions and its PM2.5 concentrations. The results suggest –3 (i.e., on October 13, 2014: PM2.5 concentration, 13 µg m that the mitigation of PM2.5 pollution not only requires the – vs. regional contribution ratio of NO3 , 13%). A positive reduction of local emissions, but also needs the cooperation correlation was observed between the contribution of the of its surrounding regions, especially during the heavy transport from surrounding regions and local PM2.5 pollution periods. concentration in Beijing. The result indicated a major role of transboundary transport in Beijing’s PM2.5 concentrations CONCLUSIONS during the high pollution episodes. The sea-level pressure and 10-m wind field (Fig. 6) revealed that during the PM2.5 In this study, the offline model CAMx (PSAT) was used ascent stage, the weak wind field to the south was to study the transport of primary and secondary PM2.5 2– – advantageous for transporting pollutants from the south of components (SO4 , NO3 , and SOA) in October 2014 in Hebei Province. The meteorological conditions were Beijing. That month, several severe pollution episodes favorable for pollution accumulation. Table 2 and Fig. 9 occurred in Beijing. The model performance reflects the show the average contributions of transport from different process of PM2.5 pollution and agree well with the regions to PM2.5 components. Pollution emissions from observations. The southern BTH region experienced heavier Baoding and Tianjin had the greatest effect on PM2.5 PM2.5 pollution than did the other regions. By studying the concentrations in Beijing because these cities are close to meteorological conditions, it revealed that the large-scale Beijing and have high pollutant emissions. The emissions synoptic in the BTH region was highly stable during the from Tianjin had the greatest influence on the secondary severe pollution episodes. The anticyclone in Eastern China 2– – components of PM2.5 concentrations in Beijing; SO4 , NO3 , showed weak local flow fields and southerly winds at the and SOA concentrations were 9.4%, 12.9%, and 10.2%, surface and strong temperature inversion, which promote respectively. The emissions from Baoding had the greatest pollution accumulation. The contribution of monthly 2– – effect on the primary components of PM2.5 (6.7%) in Beijing. regional transport to primary PM2.5, SO4 , NO3 , and SOA Based on the PM2.5 standards, the contribution ratios of concentrations in Beijing were 29.6%, 41.5%, 58.7%, and emission contributions to PM2.5 components in Beijing from 60.6%, respectively. The secondary components of Beijing’s its surrounding regions under different PM2.5 pollution PM2.5 were more easily affected by transboundary transport levels were calculated. The results are listed in Table 3. than were the primary components. The higher concentration Two different PM2.5 pollution levels were examined, namely of PM2.5 pollutants corresponded to the greater contribution –3 the daily PM2.5 concentration of 0–75 µg m , and greater of regional transport to PM2.5 components. The severe than 75 µg m–3, respectively. During the periods when the pollution episodes were promoted by the long-range –3 PM2.5 concentrations were 0–75 µg m , the contribution of transport-and-transmission generation processes from the

Wen et al., Aerosol and Air Quality Research, 18: 1720–1733, 2018 1731 south of Hebei Province. Emissions from Baoding and Beijing. The results suggest that the regional strategies aimed Tianjin had the greatest effect on PM2.5 concentrations in at reducing PM2.5 pollution will require coordinated effort to

Fig. 8. Daily variations in the contribution of emission transport to different PM2.5 components, as studied using the CAMx 2– 2– – – model (TR-SO4 : transportation contributions to SO4 ; TR-NO3 : transportation contributions to NO3 ; TR-SOA: transportation contributions to SOA).

Fig. 9. Average contribution of transport to different PM2.5 components from different regions (%).

1732 Wen et al., Aerosol and Air Quality Research, 18: 1720–1733, 2018

Table 3. Regional transport of primary and secondary components under different PM2.5 concentration level.

PM SO4 NO3 SOA > 75 µg m–3 < 75 µg m–3 > 75 µg m–3 < 75 µg m–3 > 75 µg m–3 < 75 µg m–3 > 75 µg m–3 < 75 µg m–3 Beijing 63.3% 77.4% 54.5% 73.2% 38.0% 58.0% 36.3% 52.0% Tianjin 9.2% 3.0% 12.2% 3.9% 15.5% 7.7% 12.1% 5.7% Tangshan 4.6% 2.3% 6.4% 3.8% 5.9% 4.9% 6.2% 3.2% Baoding 6.9% 6.6% 6.3% 4.9% 10.5% 6.8% 8.9% 7.9% Shijiazhuang 2.2% 2.0% 3.3% 2.8% 5.7% 5.0% 3.4% 3.0% Langfang 2.6% 1.2% 3.2% 1.5% 4.5% 2.8% 5.5% 3.6% Handan 1.5% 0.4% 3.0% 0.8% 3.5% 1.4% 2.6% 0.5% Xingtai 2.1% 1.0% 2.7% 1.3% 4.3% 2.5% 4.9% 1.3% Zhangjiakou 2.0% 2.8% 2.6% 3.6% 2.0% 4.1% 6.1% 10.5% Chengde 1.7% 2.0% 2.2% 2.9% 2.3% 3.9% 5.0% 8.9% Cangzhou 2.2% 0.7% 1.6% 0.5% 4.5% 1.5% 5.0% 1.7% Hengshui 1.5% 0.5% 1.6% 0.6% 2.7% 1.1% 3.4% 1.3% Qinhuangdao 0.2% 0.1% 0.4% 0.2% 0.7% 0.4% 0.7% 0.4% reduce both the long-range transport and also local generation Chemical composition of aerosols and layered structure of SO2, NOx, and VOCs, in addition to particulate primary of an air mass over the East China Sea. Aerosol Air emissions. The techniques used in this study are effective Qual. Res. 11: 497–507. for developing PM2.5 pollution control policies. Heald, C.L., Jacob, D.J., Park, R.J., Alexander, B., Fairlie, T.D., Yantosca, R.M. and Chu, D.A. (2006). Transpacific ACKNOWLEDGMENT transport of Asian anthropogenic aerosols and its impact on surface air quality in the United States. J. Geophys. This study was supported by the Beijing Natural Science Res. 111: 14310. Foundation (No. 8172051), LAC/CMA (No. 2017B05), Huang, R.J., Zhang, Y.L., Bozetti, C., Ho, K.F., Cao, J.J., Natural Sciences Foundation of China (No. 41305130 and Han, Y.M., Daellenbach, K.R., Slowik, J.G., Platt, S.M., No. 41305104), Natural Sciences Foundation of Beijing Canonaco, F., Zotter, P., Wolf, R., Pieber, S.M., Bruns, (No. 8161004), National Science and Technology Support E.A., Crippa, M., Ciarelli, G., Piazzalunga, A., Project of China (No. 2014BAC23B03), National Science Schwikowski, M., Abbaszade, G., Schnelle-Kreis, J., and Technology Support Project of Beijing (No. Zimmerman, R., An, Z.S., Szidat, S., Baltensperger, U., Z151100002115045), and Special Scientific Research Fund Hadded, I.E. and Prevot, A.S. (2014). High secondary of Meteorological Public Welfare Profession of China (No. aerosol contribution to particulate pollution during haze GYHY201306061). The authors thank the editors and events in China. Nature 514: 218–222. anonymous reviewers for their insightful comments. Kim, Y. Sartelet, K. and Seigneur, C. (2011). Formation of secondary aerosols over Europe: Comparison of two REFERENCES gas-phase chemical mechanisms. Atmos. Chem. Phys. 11: 1457–1477. Cao, J.J., Xu, H.M., Xu, Q., Chen, B.H. and Kan, H.D. Lang, J.L., Cheng, S.Y., Li, J.B., Chen, D.S., Zhou, Y., (2012). Fine particulate matter constituents and Wei, X. and Wang, H.Y. (2013). A monitoring and cardiopulmonary mortality in a heavily polluted Chinese modeling study to investigate regional transport and city. Environ. Health Perspect. 120: 373–378. characteristics of PM2.5 pollution. Aerosol Air Qual. Res. China State Council (2013). Air pollution Prevention and 13: 943–956. Control (Action Plan). http://www.gov.cn/zwgk/2013- Li, L., An, J.Y., Zhou, M., Yan, R.S. and Chen, C.H. 09/12/content_2486773.htm. (2015). Source apportionment of fine particles and its Chuang, M.T., Fu, J.S., Jang, C.J., Chan, C.C., Ni, P.C. and chemical components over the Yangtze River Delta, China Lee, C.T. (2008). Simulation of longrange transport during a heavy haze pollution episode. Atmos. Environ. aerosols from the Asian Continent to Taiwan by a 123: 415–429. Southward Asian high-pressure system. Sci. Total Environ. Li, X., Wang, L. and Ji, D. (2013) Characterization of the 406: 168–179. size-segregated water-soluble inorganic ions in the Jing- Fann, N., Baker, K.R. and Fulcher, C.M. (2012). Jin-Ji urban agglomeration: Spatial/temporal variability, Characterizing the PM2.5-related health benefits of emission size distribution and sources. Atmos. Environ. 77: 250– reductions for 17 industrial, area and mobile emission 259. sectors across the U.S. Environ. Int. 49: 141–151. Lin, J., An, J., Qu, Y., Chen, Y., Li, Y., Tang, Y.J., Wang, F. Hatakeyama, S., Hanaoka, S., Ikeda, K., Watanabe, I., and Xiang, W.L. (2016). Local and distant source Arakaki, T., Sadanaga, Y., Bandow, H., Kato, S., Kajii, contributions to secondary organic aerosol in the Beijing Y., Sato, K., Shimizu, A. and Takami, A. (2011). Aerial urban area in summer. Atmos. Environ. 124: 176–185. observation of aerosols transported from East Asia - Liu, X.H., Zhang, Y., Cheng, S.H., Xing, J., Zhang, Q.A.,

Wen et al., Aerosol and Air Quality Research, 18: 1720–1733, 2018 1733

Streets, D.G., Jang, C., Wang, W.X. and Hao, J.M. haze pollution over the southern Hebei area of China (2010). Understanding of regional air pollution over using the CMAQ model. Atmos. Environ. 56: 69–79. China using CMAQ, part I performance evaluation and Wang, L.T., Wei, Z., Yang, J., Zhang, Y., Zhang, F.F., Su, J., seasonal variation. Atmos. Environ. 44: 2415–2426. Meng, C.C. and Zhang, Q. (2013). The 2013 severe haze MEP (Ministry of Environment Protection of People’s over southern Hebei, China: model evaluation, source Republic of China) (2015). http://www.zhb.gov.cn/gk apportionment, and policy implications. Atmos. Chem. ml/hbb/qt/201602/t20160204_329886.htm?COLLCC=3 Phys. 13: 28395–28451. 065852892&. Wang, Y.J., Li, L. and Chen, C.H. (2014). Source appointment Morris, R.E., Koo, B., Guenther, A., Yarwood, G., McNally, of fine particulate matter during autumn haze episodes in D., Tesche, T. W, Tonnesen, G., Boylan, J. and Brewer, Shanghai, China. J. Geophys. Res. 4: 1903–1914. P. (2006). Model sensitivity evaluation for organic Wang, Z.F., Li, L.N., Wu, Q.Z., Gao, C. and Li, X. (2008). carbon using two multi-pollutant air quality models that Simulation of the impacts of regional transport on simulate regional haze in the southeastern United States. summer ozone levels over Beijing. Chin. J. Nat. 30: Atmos. Environ. 40: 4960–4972. 194–248. Saliba, N.A., Kouyoumdjian, H. and Roumie, M. (2007). Wen, W., Cheng, S.Y., Chen, X.F., Wang, G., Li, S., Wang, Effect of local and long-range transport emissions on the X.Q. and Liu, X.Y. (2015). Impact of emission control elemental composition of PM10-2.5 and PM2.5 in Beirut. on PM2.5 and the chemical composition change in Atmos. Environ. 41: 6497–6509. Beijing-Tianjin-Hebei during the APEC summit 2014. Skyllakou, K., Murphy, B.N., Megaritis, A.G., Fountoukis, Environ. Sci. Pollut. Res. 105: 432–436. C. and Pandis, S.N. (2014). Contributions of local and Wen, W., Cheng, S.Y., Liu, L., Wang, G. and Wang, X.Q. regional sources to ﬁne PM in the megacity of Paris. (2016). Source apportionment of PM2.5 in Tangshan, Atmos. Chem. Phys. 14: 2343–2352. China-Hybrid approaches for primary and secondary Squizzato, S., Masiol, M., Innocente, E., Pecorari, E., species apportionment. Front. Environ. Sci. Eng. 10: 6. Rampazzo, G. and Pavoni, B. (2012). A procedure to Wu, D.W., Fung, J.C.H, Yao, T. and Lau, A.K.H. (2013). A assess local and long-range transport contributions to study of control policy in the Pearl River Delta region by PM2.5 and secondary inorganic aerosol. J. Aerosol Sci. using the particulate matter source apportionment 46: 64–76. method. Atmos. Environ. 76: 147–161. Streets, D.G., Fu, J.S. and Jang, C.J. (2007). Air quality Zhang, Y., Olsen, K.M. and Wang, K., (2013). Fine scale during the 2008 Beijing Olympic Games. Atmos. Environ. modeling of agricultural air quality over the southeastern 41: 480–492. United States using two air quality models. Part I. Tao, M.H., Chen, L.F., Su, L. and Tao, J.H. (2012). Application and evaluation. Aerosol Air Qual. Res. 13: Satellite observation of regional haze pollution over the 1231–1252. North China Plain. J. Geophys. Res. 117: 48–50. Zhang, Y. and Wu, S.Y. (2013). Fine scale modeling of Wagstrom, K.M. and Pandis, S.N. (2011b). Contribution of agricultural air quality over the southeastern united long range transport to local ﬁne particulate matter states using two air quality models. Part II. Sensitivity concerns. Atmos. Environ. 45: 2730–2735. studies and policy implications. Aerosol Air Qual. Res. Wang, Gang., Cheng, S.Y., Lang, J.L., Yang, X.W., Wang, 13: 1475–1491. X.Q., Chen, G.L., Liu, X.Y. and Zhang, H.Y. (2017). Zheng, B., Zhang, Q., Zhang, Y., He, K.B., Wang, K., Characteristics of PM2.5 and assessing effects of emission– Zheng, G.J., Duan, F.K., Ma, Y.L. and Kimoto, T. (2015). Reduction measures in the heavy polluted city of Heterogeneous chemistry: A mechanism missing in Shijiazhuang, before, during, and after the Ceremonial current models to explain secondary inorganic aerosol Parade 2015. Aerosol Air Qual. Res. 17: 499–512. formation during the January 2013 haze episode in North Wang, L.T., Wei, Z., Wei, Z.T. and Ma, S.M. (2015). China. Atmos. Chem. Phys. 15: 2031–2049. Source apportionment of PM2.5 in top polluted cities in Hebei, China using the CMAQ model. Atmos. Environ. 122: 723–736. Received for review, October 22, 2017 Wang, L.T., Xu, J., Yang, J., Zhao, X.J., Wei, W., Cheng, Revised, December 19, 2017 D.D., Pana, X.M. and Sua, J. (2012). Understanding Accepted, December 23, 2017