Contributions of Trans-boundary Transport to Summertime Air Quality in Beijing , China 1 2

Abstract. In the present study, the WRF-CHEM model is used to evaluate the contributions of trans-boundary transport to the air quality in Beijing during a persistent air pollution episode from 5 to 14 July 2015 in Beijing–Tianjin–Hebei (BTH), China. Generally, the predicted temporal variations and spatial distributions of PM2.5 (fine particulate matter), O3 (ozone), and NO2 are in good agreement with observations in BTH. The WRF-CHEM model also reproduces reasonably well the temporal variations of aerosol species compared to measurements in Beijing. The factor separation approach is employed to evaluate the contributions of trans-boundary transport of non-Beijing emissions to the PM2.5 and O3 levels in Beijing. On average, in the afternoon during the simulation episode, the local emissions contribute 22.4 % to the O3 level in Beijing, less than 36.6 % from non-Beijing emissions. The O3 concentrations in Beijing are decreased by 5.1 % in the afternoon due to interactions between local and non-Beijing emissions. The non-Beijing emissions play a dominant role in the PM2.5 level in Beijing, with a contribution of 61.5 %, much higher than 13.7 %, from Beijing local emissions. The emission interactions between local and non-Beijing emissions enhance the PM2.5 concentrations in Beijing, with a contribution of 5.9 %. Therefore, the air quality in Beijing is generally determined by the trans-boundary transport of non-Beijing emissions during summertime, showing that the cooperation with neighboring provinces to mitigate pollutant emissions is key for Beijing to improve air quality.


Introduction
Beijing, the capital of China, has become an environmentally stressed city due to a growing population, increasing transportation activity, and city expansion (Parrish and Zhu, 2009).Beijing is situated in northeastern China, surrounded from the southwest to the northeast by the Taihang Mountains and the Yanshan Mountains and open to the North China Plain (NCP) in the south and east.Unfortunately, the NCP has become one of the most polluted areas in China due to rapid industrialization and urbanization (Zhang et al., 2013).When south or east winds are prevalent in the NCP, air pollutants originating in the NCP are transported to Beijing and surrounding areas and subject to accumulation due to the mountain blocking, causing heavy air pollution in Beijing (Long et al., 2016).
PM 2.5 (fine particulate matter) and O 3 (ozone) are considered to be the most serious air pollutants of concern in Beijing during summertime (e.g., Xie et al., 2015;Zheng et al., 2015;Chen et al., 2015;Wang et al., 2016).The mean summertime PM 2.5 mass concentration was about 80 µg m −3 in 2013 (R. K. Li et al., 2015), exceeding the second grade of National Ambient Air Quality Standards (NAAQS) in China and also exceeding the average PM 2.5 concentration of 78.1 µg m −3 during the period from 2004 to 2012 (Liu et al., 2015).During haze pollution events in summer 2014, the PM 2.5 concentration generally reaches 100 µg m −3 , and even exceeds 150 µg m −3 in Beijing (Wang et al., 2016).An in-creasing O 3 trend has been observed in Beijing from 2002 to 2010 (Wang et al., 2012;Y. H. Wang, 2013).The average maximum 1 h O 3 concentration has been reported to achieve 140 µg m −3 during summertime of 2013 in Beijing (L.T. Wang et al., 2014).Wang et al. (2016) have demonstrated that the summertime O 3 mass concentration reached high levels in 2014 in Beijing, with a daily average of up to 110 µg m −3 .Chen et al. (2015) have further shown that the average maximum daily O 3 concentrations were higher than 150 µg m −3 during the summer in 2015 at most of the monitoring sites in Beijing.
In recent years, Beijing has implemented aggressive emission control strategies to ameliorate the air quality (Parrish and Zhu, 2009).Both NO x (NO+NO 2 ) and total VOCs (volatile organic compounds) in Beijing have decreased linearly since 2002, while the daytime average O 3 concentration still increased rapidly (Tang et al., 2009;Wang et al., 2012;Zhang et al., 2014).Zhang et al. (2014) have highlighted the importance of the trans-boundary transport and the cooperation with neighboring provinces to control the O 3 level in Beijing.Pollutants transported from outside of Beijing and formed locally together determine the air quality in Beijing (Meng et al., 2006;Zhang et al., 2012).
Several studies have been performed to investigate the role of trans-boundary transport in the air quality of Beijing based on observational analyses and model simulations.Using the US EPA (Environmental Protection Agency) Model-3/CMAQ (Community Multiscale Air Quality) model simulation in the Beijing area, Streets et al. (2007) have pointed out that Hebei Province can contribute 50-70 % of Beijing's PM 2.5 concentration and 20-30 % of O 3 concentration.Wang et al. (2009) have indicated that O 3 formation in Beijing is not only affected by local emissions, but also influenced by Tianjin and the south of Hebei Province.The intense regional transport of pollutants from south to north in the NCP has been proposed to be the main reason for the heavy haze pollution in January 2013 in Beijing (Sun et al., 2014;Tao et al., 2014;Z. Wang et al., 2014).Jiang et al. (2015) have demonstrated that the transport from the environs of Beijing contributed about 55 % of the peak PM 2.5 concentration in the city during a heavy haze event in December 2013.
Since September 2013, the "Atmospheric Pollution Prevention and Control Action Plan" (hereafter referred to as APPCAP) has been implemented, which was released by the Chinese State Council to reduce PM 2.5 by up to 25 % by 2017 relative to 2012 levels.After implementation of the APPCAP, high PM 2.5 mass concentrations still can be observed and the O 3 pollution has deteriorated during summertime since 2013 in Beijing (Chen et al., 2015;Wang et al., 2016).Hence, to support the design of mitigation strategies, studies are imperative to explore the O 3 and PM 2.5 formation from various sources and evaluate the pollutants' contributions from local production and trans-boundary transport in Beijing.
The purpose of the present study is to evaluate the contributions of trans-boundary transport of emissions outside of Beijing to the air quality in Beijing and interaction of emissions in and outside of Beijing after APPCAP using the WRF-CHEM model.The model configuration and methodology are described in Sect. 2. Model results and sensitivity studies are presented in Sect.3, and conclusions and discussions are given in Sect. 4.

WRF-CHEM model
The WRF-CHEM model used in the study is developed by Li et al. (2010Li et al. ( , 2011aLi et al. ( , b, 2012) ) at the Molina Center for Energy and the Environment, with a new flexible gas-phase chemical module and the CMAQ aerosol module developed by the US EPA.The aerosol component of the CMAQ model is designed to be an efficient and economical depiction of aerosol dynamics in the atmosphere (Binkowski and Roselle, 2003).The particle size distribution in the study is represented as the superposition of three lognormal subdistributions, called modes, which includes the processes of coagulation, particle growth by the addition of mass, and new particle formation.Following the work of Kulmala et al. (1998), the new particle production rate presented here is calculated as a parameterized function of temperature, relative humidity, and the vapor-phase H 2 SO 4 concentration due to binary nucleation of H 2 SO 4 and H 2 O vapor, and the new particles are assumed to be 2.0 nm diameter.A number of recent studies have shown that organic compounds can play an important role in the nucleation process (Zhang et al., 2009(Zhang et al., , 2012;;R. Y. Zhang, 2015).The contribution from organic acids likely explains the high levels of aerosol, especially in polluted urban areas, where large amount of organic acids can be emitted directly and produced by photochemical oxidation of hydrocarbons (Fan et al., 2006), which needs to be considered in further studies.The wet deposition follows the method used in the CMAQ and the surface deposition of chemical species is parameterized following Wesely (1989).The photolysis rates are calculated using the FTUV (fast radiation transfer model) (Li et al., 2005(Li et al., , 2011a)), in which the effects of aerosols and clouds on photolysis are considered.
The inorganic aerosols are predicted in the WRF-CHEM model using ISORROPIA Version 1.7 (Nenes et al., 1998).The efficient and rapid secondary species formation in Beijing has been found during the severe haze formation process in the previous study (Guo et al., 2014).The secondary organic aerosol (SOA) formation is calculated using a nontraditional SOA module.The volatility basis set (VBS) modeling method is used in the module, assuming that primary organic components are semi-volatile and photochemically reactive and are distributed in logarithmically spaced volatility bins.Detailed information about the VBS approach can be found in Li et al. (2011b).Recent studies have shown that small di-carbonyls (glyoxal and methylglyoxal) are impor-tant for the aerosol formation due to their traffic origin (Zhao et al., 2006;Gomez et al., 2015).Li et al. (2011a) have indicated that glyoxal and methylglyoxal can contribute about 10 % of the SOA in the urban area of Mexico City.The SOA formation from glyoxal and methylglyoxal in this study is parameterized as a first-order irreversible uptake by aerosol particles and cloud droplets, with a reactive uptake coefficient of 3.7 × 10 −3 for glyoxal and methylglyoxal (Zhao et al., 2006;Volkamer et al., 2007;Gomez et al., 2015).

Pollution episode simulation
A persistent air pollution episode from 5 to 14 July 2015 in Beijing-Tianjin-Hebei (BTH) is simulated using the WRF-CHEM model.During the episode, the observed mean daily PM 2.5 concentration is 73.8 µg m −3 and the average O 3 concentration in the afternoon reaches 237.0 µg m −3 in Beijing.The maximum of O 3 concentration is higher than 350 µg m −3 , and the maximum of PM 2.5 concentration can reach a high level exceeding 150 µg m −3 .Supplement Fig. S1a-c show the daily averages of the temperature, relative humidity, and wind speed in Beijing during the summer of 2015.The minimum air temperature is 18.7 • , and the maximum air temperature is 40 • during the summer, with an average of 25.7 • .The average relative humidity is 63.8 %.The southeast or southwest wind is prevailing over the NCP due to the influence of East Asian summer monsoon (Zhang et al., 2010), with the average wind speed of 5.6 m s −1 in the summer of 2015.During the study period, the average temperature, relative humidity, and wind speed are 28.4 • C, 51.7 %, and 6.3 m s −1 , respectively, indicating typical summertime meteorological conditions.During the summer of 2015, the average PM 2.5 concentration is 56.1 µg m −3 and the average O 3 concentration in the afternoon is 216.4 µg m −3 (Fig. S1de).The high O 3 and PM 2.5 event occurs frequently during the summertime of 2015, so the study period can well represent the summertime O 3 and PM 2.5 pollution in Beijing, and provide a suitable case for observation analyses and model simulations to investigate the effect of trans-boundary transport on the summertime air quality of Beijing.
The WRF-CHEM model adopts one grid with horizontal resolution of 6 km and 35 sigma levels in the vertical direction, and the grid cells used for the domain are 200 × 200 (Fig. 1).The physical parameterizations include the microphysics scheme of Hong and Lim (2006), the Mellor, Yamada, and Janjic (MYJ) turbulent kinetic energy (TKE) planetary boundary layer scheme (Janjić, 2002), the unified NOAH land-surface model (Chen and Dudhia, 2001), the rapid radiative transfer model (RRTM) long-wave radiation scheme (Mlawer et al., 1997), and the Goddard shortwave parameterization (Suarex and Chou, 1994;Chou and Suarez, 1999;Chou et al., 2001).The NCEP 1 • × 1 • reanalysis data are used to obtain the meteorological initial and boundary conditions, and the meteorological simulations are not nudged in the study.The chemical initial and boundary conditions are interpolated from the 6 h output of MOZART (Horowitz et al., 2003).The spin-up time of the WRF-CHEM model is 28 h.The SAPRC-99 (Statewide Air Pollution Research Center, version 1999) chemical mechanism is used in the present study.
The anthropogenic emissions are developed by Zhang et al. (2009), whose work is based on the 2013 emission inventory, including contributions from agriculture, industry, power generation, residential, and transportation sources.The SO 2 , NO x , and CO emissions have been adjusted according to their observed trends from 2013 to 2015 in the present study, but the VOC emissions are not changed considering that the VOC emissions are still not fully considered in the current air pollutant control strategy.The major pollutant emissions used in the model simulation for Beijing, Tianjin, and the neighboring provinces (Hebei, Shanxi, and Shandong) are summarized in Table 1.Obviously, high anthropogenic emissions are distributed outside of Beijing, especially in Hebei and Shandong provinces.Figure 2 presents distributions of the emission rates of VOCs, NO x , OC, and SO 2 in the simulation domain, showing that the anthropogenic emissions are generally concentrated in urban areas.As shown in Fig. 2, the total emissions from neighboring regions are much more than those in Beijing, and the emission rates in Tianjin, the south of Hebei and Shandong are also higher than those in Beijing, particularly with regard to SO 2 emissions.Therefore, when the south or east wind is prevailing in the NCP, the severe air pollution can be formed in Beijing when precursor emissions in highly industrialized areas chemically react as they are carried toward Beijing, blocked by mountains and further accumulated and interacted with  those in Beijing.It is worth noting that uncertainties of the emission inventory used in the study are still rather large considering the rapid changes in anthropogenic emissions that are not fully reflected in the current emission inventories, particularly since implementation of the APPCAP, and the complexity of pollutants precursors.For example, different VOC types exhibit distinct kinetic behaviors, and as an important fraction of total VOCs in the urban atmosphere, aromatics are responsible for the photochemical ozone production and secondary organic aerosol formation (Suh et al., 2003;Fan et al., 2004).In the SAPRC99, aromatics are lumped into ARO1 and ARO2.ARO1 mainly includes toluene, benzene, ethylbenzene, and other aromatics with reaction rate with OH (kOH) less than 2 × 10 4 ppm −1 min −1 .ARO2 includes xylene, trimethylbenzene, and other aromatics with kOH greater than 2 × 10 4 ppm −1 min −1 .Additionally, bio-genic VOCs also play a considerable role in the ozone production (Li et al., 2007), and monoterpenes and isoprene are the main biogenic VOCs in the SAPRC99 chemical mechanism.The biogenic emissions are calculated online using the MEGAN (Model of Emissions of Gases and Aerosol from Nature) model developed by Guenther et al. (2006).

Factor separation approach
The formation of the secondary atmospheric pollutant, such as O 3 , secondary organic aerosol, and nitrate, is a complicated nonlinear process in which its precursors from various emission sources and transport react chemically or reach equilibrium thermodynamically.Nevertheless, it is not straightforward to evaluate the contributions from different factors in a nonlinear process.The factor separation approach (FSA) proposed by Stein and Alpert (1993) can be used to isolate the effect of one single factor from a nonlinear process and has been widely used to evaluate source effects (Gabusi et al., 2008;Weinroth et al., 2008;Carnevale et al., 2010;Li et al., 2014).The total effect of one factor in the presence of others can be decomposed into contributions from the factor and that from the interactions of all those factors.Suppose that field f depends on a factor ϕ: The FSA decomposes function f (ϕ) into a constant part that does not depend on ϕ(f (0)) and a ϕ-depending component (f (ϕ)), as follows: considering that there are two factors X and Y that influence the formation of secondary pollutants in the atmosphere and also interact with each other.Denoting f XY , f X , f Y , and f 0 as the simulations including both of the two factors, factor X only, factor Y only, and neither of the two factors, respectively.The contributions of factor X and Y can be isolated as follows: (5) Note that term f X(Y ) represents the impacts of factor X(Y ), while f 0 is the term independent of factors X and Y .The simulation including both factors X and Y is given by the following: The mutual interaction between X and Y can be expressed as follows: The above equation shows that the study needs four simulations, f XY , f X , f Y , and f 0 , to evaluate the contributions of two factors and their synergistic interactions.

Statistical metrics for observation-model comparisons
In the present study, the mean bias (MB), root mean square error (RMSE), and index of agreement (IOA) are used as indicators to evaluate the performance of WRF-CEHM model in simulation against measurements.IOA describes the relative difference between the model and observation, ranging from 0 to 1, with 1 indicating perfect agreement.
where P i and O i are the predicted and observed pollutant concentrations, respectively.N is the total number of the predictions used for comparisons, and P and Ō represent the average of the prediction and observation, respectively.

Pollutant measurements
The hourly measurements of O 3 , NO 2 , and PM 2.5 used in the study are downloaded from the website http://www.aqistudy.cn/.The submicron sulfate, nitrate, ammonium, and organic aerosols are observed by the Aerodyne Aerosol Chemical Speciation Monitor (ACSM), which is deployed at the National Center for Nanoscience and Technology (NCNST), Chinese Academy of Sciences, Beijing (Fig. 1).The mass spectra of organic aerosols are analyzed using the positive matrix factorization (PMF) technique to separate them into four components: hydrocarbon-like organic aerosol (HOA), cooking organic aerosol (COA), coal combustion organic aerosol (CCOA), and oxygenated organic aerosol (OOA).HOA, COA, and CCOA are interpreted as surrogates of primary organic aerosol (POA), and OOA is a surrogate of SOA.
The APPCAP has been implemented since 2013 September, so comparisons of summertime pollutants between 2013 and 2015 can show the mitigation effects on the air quality.Considering that high O 3 concentrations generally take place in the afternoon during summertime, Table 2 presents the summertime concentrations of pollutants in the afternoon (12:00-18:00 Beijing Time (BJT)) averaged at 12 monitoring sites in Beijing in 2013 and 2015.The rainy days during summertime in Beijing are 43 and 46 days in 2013 and 2015, respectively, showing the similar meteorological conditions between the 2 years.Therefore, in general, the air pollutant variations between 2013 and 2015 can be mainly attributed to implementation of the APPCAP.Apparently, implementation of the APPCAP has considerably decreased the concentrations of primary species of CO and SO 2 , particularly with regard to SO 2 , which was reduced by more than 40 % from 2013 to 2015.Most NO x exists in the form of NO 2 in the afternoon during summertime due to active photochemical processes.Therefore, a 25.1 % decrease of NO 2 in the afternoon from 2013 to 2015 shows that the NO x emission mitigation is also effective in Beijing.The PM 2.5 concen-  Therefore, if the simulated winds cause the O 3 plume formed in the urban area to leave early or deviate the O 3 plume transported from outside of Beijing from the urban area, the model is subject to underestimating the O 3 concentration in Beijing (Bei et al., 2010).The WRF-CHEM model also reasonably yields the NO 2 diurnal profiles, but frequently overestimates the NO 2 concentrations during nighttime, which is likely caused by the biased boundary layer simulations.

Aerosol species simulations in Beijing
Figure 4 shows the temporal variations of simulated and observed aerosol species at NCNST site in Beijing from 5 to 14 July 2015.The WRF-CHEM model generally performs reasonably in simulating the aerosol species variations compared with ACSM measurements.As a primary aerosol species, the POA in Beijing is determined by direct emissions from various sources and transport from outside of Beijing, so uncertainties from emissions and meteorological fields have a remarkable effect on the model simulations (Bei et al., 2012(Bei et al., , 2013)).Although the MB and RMSE for POA are 0.0 and 3.1 µg m −3 , respectively, the IOA is less than 0.60, indicating the considerable biases in POA simulations.The WRF-CHEM model has difficulties in simulating well the sulfate aerosol, with an IOA lower than 0.60.The model cannot produce the observed high peaks of sulfate aerosols around noontime on 8, 11, and 12 July 2015.The sulfate aerosol in the atmosphere is produced from multiple sources, including SO 2 gas-phase oxidations by hydroxyl radicals (OH) and stabilized Criegee intermediates (sCI), aqueous reactions in cloud or fog droplets, and heterogeneous reactions on aerosol surfaces, as well as direct emissions from power plants and industries (Li et al., 2016).The model reproduces reasonably well the observed temporal variations of SOA, nitrate, and ammonium, with IOAs exceeding 0.75.The model simulates well the peak concentration of SOA, nitrate and ammonium at the rush hour, but the model also underestimates the SOA, nitrate, and ammonium as well, with MB of −1.1, −0.7, and −0.5 µg m −3 , respectively.For nitrate and ammonium, the underestimates occur mainly on 8 July 2015, possibly due to wind fields, which will be further analyzed in the Supplement (Fig. S2).
3.1.3O 3 , NO 2 , and PM 2.5 simulations in BTH Figure 5 shows the diurnal profiles of observed and simulated near-surface O 3 , NO 2 , and PM 2.5 concentrations averaged over monitoring sites in BTH from 5 to 14 July 2015.
The WRF-CHEM model exhibits good performance in predicting the temporal variations of O 3 , NO 2 , and PM 2.5 concentrations compared with measurements in BTH, with IOAs higher than 0.80.In addition, O 3 and NO 2 simulations are also improved in BTH compared to those in Beijing, indicating better model performance for regional simulations on a large scale.Figure 6 presents the distributions of calculated and observed near-surface PM 2.5 concentrations along with the simulated wind fields at 10:00 BJT on the 6 selected representative days with high O 3 and PM 2.5 concentrations.The calculated PM 2.5 spatial patterns generally agree well with the observations at the monitoring sites.The observed PM 2.5 concentrations in BTH are still high even after implementation of the APPCAP, frequently exceeding 75 µg m −3 on the selected 6 days.The PM 2.5 concentrations in Beijing are higher than 115 µg m −3 at 10:00 BJT on 8, 11, and 12 July 2015, causing moderate air pollution.
The O 3 concentration during summertime reaches its peak during the period from 14:00 to 16:00 BJT in Beijing (Tang et al., 2012).Figure 7 presents the spatial distribution of calculated and measured near-surface O 3 concentration at 15:00 BJT on the selected 6 days, along with the simulated wind fields.In general, the simulated O 3 spatial patterns are consistent with the measurements, but model biases still exist.High O 3 concentrations at 15:00 BJT in Beijing are observed and also simulated by the model, frequently exceeding 250 µg m −3 .The O 3 transport to Beijing from its surrounding areas is also obvious when the winds are easterly or southerly.Figure 8 provides the spatial distribution of simulated and observed near-surface NO 2 concentration on the selected 6 days at 08:00 BJT when the NO 2 concentration reaches it peak due to rush hour NO x emissions and a low planetary boundary layer (PBL).The simulated near-surface NO 2 concentrations highlight the dominant impact of the anthropogenic emissions, primarily concentrated in cities or their downwind areas, which generally agree well with the measurements.Beijing is surrounded from south to east by cities with high NO 2 concentrations, which can influence the O 3 formation in Beijing when south or east winds are prevalent.
The good agreements between predicted PM 2.5 , O 3 , NO x , and aerosol species and the corresponding measurements show that the modeled meteorological fields and emissions used in simulations are generally reasonable.

Contributions of trans-boundary transport to the
O 3 and PM 2.5 levels in Beijing 3.2.1 Analysis of horizontal transport of O 3 and PM 2.5 The analysis in Sect.3.1.3has shown the strong correlation between the airflow and the high level pollutants in Beijing during the study episode.It is essential to confirm whether the continuous air pollution in Beijing is directly related to the airflow transport from outside of Beijing (An et al., 2007;Yang et al., 2010).In the present study, the horizontal transport flux intensity is defined as the horizontal wind speed on the grid border multiplied by the pollutant concentration of the corresponding grid from which the airflows come (Jiang et al., 2008).Considering that transboundary transport mainly occurs within the PBL, the study  68.2 and 68.5 g s −1 , respectively, showing important contributions of non-Beijing emissions to the air quality in Beijing.

Trans-boundary transport contributions to O 3 in Beijing
The      (Sillman et al., 1990;Lei et al., 2007Lei et al., , 2008)).When the O 3 precursors emitted from outside of Beijing are transported to Beijing and mixed with local emissions, the concentrations of O 3 precursors are increased and the ratio of VOC s / NO x is also altered, causing the formed O 3 concentration unequal to the simple linear summation of O 3 contributions from the local and non-Beijing emissions.
The background O 3 in Beijing plays an important role in the O 3 level in the afternoon, accounting for 46.1 % of the O 3 concentration.The background O 3 contribution varies from 32.6 to 62.9 % during the episode, which is primarily determined by the prevailing wind direction.When the northerly wind is prevalent, the clean airflow from the north affects Beijing, enhancing the background O 3 contribution, such as on 5, 13, and 14 July 2015.However, when the polluted airflow from the south impacts Beijing, the background O 3 contribution is decreased.The O 3 contributions in Beijing induced by the trans-boundary transport of emissions outside of Beijing are about 31.5 % of the O 3 concentration during the study episodes, which is in agreement with previous studies (Streets et al., 2007;Wang et al., 2008), indicating that the Previous studies have proposed that the regional transport of O 3 precursors can play an important role in inducing the high O 3 concentration level in Beijing (Wang et al., 2009;Zhang et al., 2014).Table S2 provides the average NO 2 contributions in Beijing from local emissions, non-Beijing emissions, emission interactions, and background.Different from O 3 , the local emissions dominate the level of NO 2 in the Beijing area, with an average contribution of 70.3 % during the study episode.The average contribution of non-Beijing emissions, emission interactions, and background are 24.8,0.9, and 4.0 %, respectively.

Trans-boundary transport contributions to PM 2.5 in Beijing
When the Beijing local emissions are not considered in simulations, Beijing still experiences high PM 2.5 pollution, with an average PM 2.5 concentration of 48.3 µg m −3 during the simulation episode, and the PM 2.5 level in Beijing still exceeds 75 µg m −3 on several days.However, when only considering the Beijing local emissions, the average PM 2.5 concentration in Beijing is 19.6 µg m −3 during the episode, showing that Beijing's PM 2.5 pollution is dominated by the trans-boundary transport (Fig. 10b).Table 4 shows the average PM 2.5 contribution in Beijing from local emissions, non-Beijing emissions, emission interactions, and background.
During the study episode, the average PM 2.5 contribution from local emissions is 13.7 %, which is much lower than the contribution of 61.5 % from non-Beijing emissions, further showing the dominant role of the trans-boundary transport in the Beijing PM 2.5 pollution.The emission interactions enhance the PM 2.5 level in Beijing on average, with a contribution of 5.9 %.The background PM 2.5 contribution to Beijing is 18.9 % on average, lower than those for O 3 .The PM 2.5 contribution caused by the trans-boundary transport is about 67.4 % of PM 2.5 concentrations in Beijing, indicating that the cooperation with neighboring provinces to control the PM 2.5 level is key for Beijing to improve air quality.Previous studies have also demonstrated the dominant role of non-Beijing emissions in the PM 2.5 level in Beijing.Based on the CMAQ model, Streets et al. (2007) have reported that average contribution of regional transport to PM 2.5 at the Olympic Stadium can be 34 %, up to 50-70 % under prevailing south winds.Guo et al. (2010) have provided a rough estimation that the regional transport can contribute 69 % of the PM 10 and 87 % of the PM 1.8 in the Beijing local area using the short and low time resolution data in the summer.Combining the PM 2.5 observations and MM5-CMAQ model results, regional transport is estimated to contribute 54.6 % of the PM 2.5 concentration during the polluted period, with an annual average PM 2.5 contribution of 42.4 % (Lang et al., 2013).Using the longterm measurements of PM 2.5 mass concentrations from 2005 to 2010 at urban Beijing, and trajectory cluster and receptor models, the average contribution of long-distance transport to Beijing's PM 2.5 level can be approximately 75.2 % in the summer (L.L. Wang et al., 2015).

Trans-boundary transport contributions to aerosol species in Beijing
Figure 11 shows the temporal variation of the averaged contributions to the near-surface aerosol constituents from total emissions (f BS ), local emissions (f B ), the trans-boundary transport of non-Beijing emissions (f S ), emission interactions (f BS ), and the background (f 0 ) during the simulation episode.The temporal variations of elemental carbon (EC) and POA from local emissions and trans-boundary transport exhibit obvious diurnal cycles, i.e., highest during nighttime and lowest in the afternoon, corresponding to the variations of PBL height and anthropogenic emissions.The SOA from local emissions reaches its peak in the afternoon when the O 3 concentration is high, but the trans-boundary transport causes the gradual accumulation process of SOA in Beijing from 5 to 9 July and from 9 to 13 July.The sulfate temporal profile from the trans-boundary transport is similar to that of SOA, also showing the accumulation process.In addition, the sulfate aerosols from local emissions do not vary remarkably.The nitrate aerosols from local emissions and the transboundary transport generally attain peaks in the morning when the air temperature is not high and the HNO 3 concentrations are not low.The ammonium aerosol variations are generally determined by those of sulfate and nitrate aerosols.For example, the variations of ammonium aerosols from the trans-boundary include not only the morning peaks, but also the accumulation processes from 5 to 9 July and from 9 to 13 July.Except for the sulfate aerosol, the temporal variations of aerosol species from background are not large.Table 5 presents the average aerosol constituent contributions from Beijing local emissions, non-Beijing emissions, emission interactions, and the background, and mass fractions in the total PM 2.5 in Beijing during the episode.Organic aerosols (POA+SOA) constitute the most important component of PM 2.5 , accounting for 34.8 % of PM 2.5 mass concentration, which is consistent with the ACSM measurement in Beijing (Sun et al., 2014).In addition, SOA contributes more than 70 % of organic aerosol mass concentrations.Although the SO 2 concentrations have been decreased by more than 40 % since implementation of the APPCAP, sulfate aerosols still play an important role in the PM 2.5 level in Beijing and make up 25.1 % of the PM 2.5 mass concentrations, showing high sulfate contributions from the trans-boundary transport and background.The ammonium, nitrate, EC, and unspecified species account for 13.7, 14.1, 5.8, and 6.5 % of the PM 2.5 mass concentrations, respectively.Secondary aerosol species dominate the PM 2.5 mass concentration in Beijing, with a contribution of 77.9 %.
The local emissions contribute more than 20 % of the mass concentrations for the primary aerosol species, but less than 15 % for the secondary aerosol species in Beijing (Table 5).The trans-boundary transport of non-Beijing emissions dominates all the aerosol species levels in Beijing, with contributions exceeding 50 %, particularly for SOA and nitrate.In addition, the POA and sulfate background contributions are also high, more than 20 %.Although the primary aerosol species of EC and unspecified constituents are not involved in the chemical process and also do not participate in the gas-particle partitioning, the emission interactions still enhance EC and unspecified constituents concentrations, with contributions of around 1.5 %, which is caused by the PBLpollution interaction.It is clear that the PBL-pollution interaction plays an important role in the pollutant accumulation in Beijing (Y.Wang et al., 2013;Peng et al., 2016).Mixing of Beijing local emissions with those outside of Beijing increases the aerosol concentrations in the PBL and decreases the incoming solar radiation down to the surface, cooling the temperature of the low-level atmosphere to suppress the development of the PBL and hinder the aerosol dispersion in the vertical direction.
The emission interactions increase the POA and SOA concentrations, with a POA contribution of 5.3 % and a SOA contribution of 5.9 %.In the VBS modeling approach, primary organic components are assumed to be semi-volatile and photochemically reactive.Mixing of Beijing local emissions with non-Beijing emissions enhances the organic condensable gases, and considering that the saturation concentrations of the organic condensable gases do not change, more organic condensable gases partition into the particle phase, increasing the POA and SOA concentrations.
The contributions of emission interactions to inorganic aerosols, including sulfate, nitrate, and ammonium, are more complicated, depending on their particle phase and precursor concentrations.In the present study, ISORROPIA (Version 1.7) is used to calculate the thermodynamic equilibrium between the sulfate-nitrate-ammonium-water aerosols and  their gas phase precursors H 2 SO 4 -HNO 3 -NH 3 -water vapor.Although mixing of Beijing local emissions with non-Beijing emissions increases inorganic aerosol precursors, the inorganic aerosol contributions from emission interactions are still uncertain due to the thermodynamic equilibrium between inorganic aerosols and their precursors.The nitrate contributions from emission interactions are 18.1 %, much more than those for other aerosol constituents.The sulfate contribution from emission interactions is not significant, only 3.4 %.The ammonium contributions from emission interactions are 1.5 %, similar to those of primary aerosol species.

Summary and conclusions
In the present study, a persistent air pollution episode with high concentrations of O 3 and PM 2.5 is simulated using the WRF-CHEM model during the period from 5 to 14 July In general, the predicted temporal variations of PM 2.5 , O 3 , and NO 2 concentrations agree well with observations in Beijing and BTH, but the model biases still exist, which are perhaps caused by the uncertainties of simulated meteorological conditions and the emission inventory.The model also successfully reproduces the spatial distributions of PM 2.5 , O 3 , and NO 2 concentrations compared with measurements.The model performs reasonably well in modeling the variations of aerosol constituents compared with ACSM measurement at the NCNST site in Beijing, but there are considerable biases in POA and sulfate simulations.
The FSA is used to investigate the contribution of transboundary transport of non-Beijing emissions to the air quality in Beijing.If the Beijing local emissions are not included in model simulations, the O 3 and PM 2.5 concentrations in Beijing still remain high, showing that the transboundary transport of emissions outside of Beijing plays a more important role in the air quality in Beijing than the Beijing local emissions.On average, the local emissions contribute 22.4 % of O 3 in the afternoon and 13.7 % of PM 2.5 mass concentrations in Beijing during the episode.The O 3 contribution in the afternoon and PM 2.5 contribution from the trans-boundary transport of non-Beijing emissions are 36.6 and 61.5 %, respectively, far exceeding those from local emissions.The interactions between local and non-Beijing emissions generally decrease the O 3 level in the afternoon and increase the PM 2.5 level in Beijing during the episode, with contributions of −5.1 and +4.4 %, respectively.In addition, the trans-boundary transport dominates all the aerosol species levels in Beijing, with contributions exceeding 50 % on average, particularly for SOA and nitrate.The emission interactions in general increase all the aerosol species levels due to the PBL-pollution interaction and the enhancement of precursors of secondary aerosols.Hence, the air quality in Beijing during summertime is generally determined by the trans-boundary transport of emissions outside of Beijing.
However, there is still controversy over whether local or non-local emissions play a dominant role in the air quality in Beijing (Guo et al., 2010(Guo et al., , 2014;;P. Li et al., 2015;R. Zhang et al., 2015).When only considering the local emissions, the summertime PM 2.5 level in Beijing is comparable to that in Mexico City.Mexico City has once been one of the most polluted cities in the world, but the air quality has been greatly improved in recent years after taking emission control strategies (Molina et al., 2002(Molina et al., , 2007(Molina et al., , 2010)).Therefore, a comprehensive model comparison of summertime pollution in Mexico City and Beijing would be illuminating for elucidation of the contributions of trans-boundary transport to the air quality in Beijing.
It is worth noting that, although the WRF-CHEM model captures the spatial distributions and temporal variations of pollutants well, the model biases still exist.The discrepancies between the predictions and observations are possibly caused by the uncertainties in the emission inventory and the meteorological field simulations (R. Zhang et al., 2015).BTH has been considered as a polluted air basin (Zhao et al., 2009;Parrish and Stockwell, 2015), which frequently experiences O 3 and PM 2.5 pollution during summertime.Future studies need to be conducted to improve the WRF-CHEM model simulations, and further to assess the contributions of trans-boundary transport of emissions outside of Beijing to the air quality in Beijing, considering the rapid changes in anthropogenic emissions since implementation of the APP-CAP.This study mainly aims at providing a quantification of the effect of trans-boundary transport on the air quality in Beijing.It demonstrates that the effective approach to improving air quality in Beijing is to reduce both local and non-Beijing emissions in BTH.Further sensitivity simulations of different emission reduction measures are needed to design the most efficient emission control strategies.
The Supplement related to this article is available online at doi:10.5194/acp-17-2035-2017-supplement.
Competing interests.The authors declare that they have no conflict of interest.

Figure 1 .
Figure 1.WRF-CHEM simulation domain.The blue circles represent centers of cities with ambient monitoring sites and the red circle denotes the NCNST site.The size of the blue circle denotes the number of ambient monitoring sites of cities.

Figure 2 .
Figure 2. Spatial distribution of anthropogenic (a) NO x , (b) VOC s , (c) OC, and (d) SO 2 emission rates (g month −1 ) in the simulation domain.

3
Figure3shows the temporal variations of observed and simulated near-surface O 3 , NO 2 , and PM 2.5 concentrations averaged over monitoring sites in Beijing from 5 to 14 July 2015.The WRF-CHEM model performs reasonably well in simulating the PM 2.5 variations compared with observations in Beijing.The MB and RMSE are −3.6 and 22.5 µg m −3 , respectively, and the IOA is 0.86.The model reproduces well the temporal variations of O 3 concentrations, with an IOA of 0.92.The model considerably underestimates the O 3 concentration during daytime on 5, 6, and 13 July.Most of the monitoring sites in Beijing are concentrated in urban areas.Therefore, if the simulated winds cause the O 3 plume formed in the urban area to leave early or deviate the O 3 plume transported from outside of Beijing from the urban area, the model is subject to underestimating the O 3 concentration in Beijing(Bei et al., 2010).The WRF-CHEM model also reasonably yields the NO 2 diurnal profiles, but frequently overestimates the NO 2 concentrations during nighttime, which is likely caused by the biased boundary layer simulations.

Figure 3 .
Figure 3.Comparison of measured (black dots) and predicted (blue line) diurnal profiles of near-surface hourly (a) PM 2.5 , (b) O 3 , and (c) NO 2 averaged over all ambient monitoring stations in Beijing from 5 to 14 July 2015.

Figure 4 .
Figure 4. Comparison of measured (black dots) and simulated (black line) diurnal profiles of submicron aerosol species of (a) POA, (b) SOA, (c) sulfate, (d) nitrate, and (e) ammonium at NC-NST site in Beijing from 5 to 14 July 2015.

Figure 5 .
Figure 5.Comparison of measured (black dots) and predicted (blue line) diurnal profiles of near-surface hourly (a) PM 2.5 , (b) O 3 , and (c) NO 2 averaged over all ambient monitoring stations in BTH from 5 to 14 July 2015.
FSA is used in the present study to evaluate the contributions and interactions of emissions from Beijing and outside of Beijing to the near-surface concentrations of O 3 and PM 2.5 in Beijing.Four model simulations are performed, including f BS with both the anthropogenic emissions from Beijing and outside of Beijing, f B with the emission from Beijing alone, f S with only emissions outside of Beijing, and f 0 without both the emissions from Beijing and outside of Beijing, representing background concentrations.Apparently, the air pollutant levels in Beijing are determined by the contribution from local emissions (f B , f B −f 0 ), the trans-boundary transport of non-Beijing emissions (f S , f S − f 0 ), emission interactions between local and non-Beijing emissions (f BS , f BS − f B − f S + f 0 ), and background (f 0 ).
Figure 10 provides the temporal variations of the average near-surface O 3 and PM 2.5 concentrations from f BS with all the emissions, f B with Beijing emissions alone, and f S with non-Beijing emissions alone in Beijing from 5 to 14 July 2015.Apparently, non-Beijing emissions generally play a more important role in the O 3 level of Beijing than local

Figure 9 .
Figure 9. Temporal variations of total net horizontal transport flux of PM 2.5 , O 3 , and NO 2 over the Beijing boundary (blue line) and the contribution of non-Beijing emission to the PM 2.5 , O 3 , and NO 2 concentrations in Beijing (black line) during the study episode.

Figure 10 .
Figure 10.Temporal variations of the average near-surface O 3 and PM 2.5 concentrations from f BS with all the emissions (black line), f B with Beijing emissions alone (blue line), and f S with non-Beijing emissions alone (red line) in Beijing from 5 to 14 July 2015.

Figure 11 .
Figure 11.Temporal variations of the average contributions to the near-surface aerosol species concentrations from total emissions (black line, defined as f BS ), local emissions (blue line, f B , defined as f B − f 0 ), non-Beijing emissions (red line, f S , defined as f S − f 0 ), the emission interactions (green line, f BS , defined as f BS − f B − f S + f 0 ), and background (black dashed line, defined as f 0 ) in Beijing from 5 to 14 July 2015.

Table 3 .
Average O 3 contributions (%) from 12:00 to 18:00 BJT in Beijing from local and non-Beijing emissions, as well as the background emissions and interactions of both emissions from 5 to 14 July 2015.
s / NO x

Table 4 .
Average PM 2.5 contributions (%) in Beijing from local and non-Beijing emissions, as well as the interactions of both emissions and background emissions from 5 to 14 July 2015.

Table 5 .
Aerosol species' contributions (%) from local emissions, non-Beijing emissions, interactions of both emissions, and background, as well as mass fraction in the total PM 2.5 (%) in Beijing averaged during the period from 5 to 14 July 2015.
Wu et al.: Contributions of trans-boundary transport to summertime air quality 2047 2015 in BTH, to evaluate the contributions of trans-boundary transport to the air quality in Beijing.Although the APPCAP has been implemented since 2013 September, the average O 3 concentration in the afternoon has increased by 22.8 % from 2013 to 2015 in Beijing, and Beijing still experienced high O 3 and/or PM 2.5 concentrations frequently during summertime of 2015.