Beijing has suffered from heavy local emissions as well as regional
transport of air pollutants, resulting in severe atmospheric fine-particle
(PM
Presently, haze in China has the characteristics of high frequency and long
duration on a regional scale and has influenced public life and human
health (Xie et al., 2016). High concentrations of fine particulates, which
can significantly reduce atmospheric visibility, are one of the main factors
in the formation of haze episodes (Y. Sun et al., 2016; Watson et al., 2002;
Yang et al., 2015). Previous studies have found that PM
As the capital of China, Beijing has suffered from heavy emissions from
various sources, resulting in severe PM
Previous studies have indicated that PM
Due to the importance of the regional transport contribution to PM
Online sampling of PM
Online PM
An in situ Gas and Aerosol Compositions monitor (IGAC, Model S-611,
Fortelice International Co. Ltd.), which collects both gases and
particles simultaneously, was applied to measure water-soluble ions online
with 1 h time resolution in this study. A detailed description of IGAC can
be found in Young et al. (2016). Briefly, IGAC was composed of three major
units, including a wet annular denuder (WAD), to collect gases into aqueous
solution, a scrub and impact aerosol collector (SCI) to collect particles
into solution and a sample analysis unit comprised of two ion chromatographs
(DionexICS-1000) for analyzing anions and cations (IC). Ambient air was
drawn through a PM
Twenty-three trace elements in PM
Strict quality assurance (QA) and quality control (QC) protocols for online
instruments were performed during the whole sampling period. For IGAC, the
internal standard (LiBr) was added continuously to each sample and analyzed
by the IC system during the analysis to check the stability of the IGAC
instrument. During the sampling period, the mean concentrations of
Chemical closure has been done between the measured and reconstructed
PM
To qualitatively and quantitatively identify sources of PM
The footprint model developed by Peking University was used to simulate the
potential source region of air pollution. The footprint is a transfer
function in a diffusion problem linking the source and the measurement
result at a point (receptor) (Pasquill and Smith, 1977). That is,
The meteorological data used to drive the footprint model were provided by the
Weather Research and Forecasting model (WRF-ARWv3.6.1)
(
In this study, the NAQPMS model was applied to analyze the contribution of
local emissions and regional transport to PM
Three nested model domains were used in this study. The coarsest domain (D1)
covered most of China and East Asia with a 27 km resolution. The second
domain (D2) included most anthropogenic emissions within the North China
Plain with a 9 km resolution. The innermost domain (D3) covered the
Beijing–Tianjin–Hebei region at a 3 km resolution. The first level of model
above the surface is 30 m in height, and the average vertical layer spacing
between 30 m and 1 km is around 100 m. The MIX
(
The footprint model was used to provide the direction of source regions, while the NAQPMS model was run to calculate the contribution of local emission and regional transport. To verify the consistency between the two models, the footprint with a time resolution of 6 h was divided into four types (local, south, north and east) according to the direction of potential source regions, and average local contributions of different types obtained from NAQPMS were calculated (See Table S3). Based on the input data availability, the footprint simulation was performed from 1 to 31 December, while the NAQPMS model analysis was carried out from 10 November to 15 December. Therefore, we use the data from 1 to 15 December for the consistency test of NAQPMS and the footprint model. A typical example of different types of footprint can be seen in Fig. S5. The average local contribution estimated by NAQPMS was highest for the local footprint (85 %) and lower for southern (68 %), northern (63 %) and eastern footprints (66 %). The results of the two models correlated well with each other.
Based on online measurements of PM
Chemical composition of PM
Temporal variation in the chemical composition of PM
Variation in
Figure 2 shows the large differences in chemical composition of PM
To conduct high-time-resolution source apportionment in Beijing, a PMF model
was applied to 1 h online measurement data. The six-factor solution gave the
best performance. The profile for each factor is shown in Fig. S7.
Contribution of different factors to PM
Source contribution of PM
The source apportionment result of PMF in winter in Beijing is shown in Fig. 3. During the campaign, the source contribution in Beijing ranked as secondary sources (44 %) > traffic sources (18 %) > coal combustion (16 %) > biomass burning (9 %) > industrial sources (8 %) > dust (5 %). The high contribution of secondary sources in winter was similar to previous studies (Gao et al., 2016; Peng et al., 2016; Zhang et al., 2013), which might be attributed to regional transport and heterogeneous reactions (Ma et al., 2017).
Meteorological conditions during pollution episodes and nonhaze periods.
Considering data integrity and representativeness, four typical pollution
episodes (EP1-4) and two nonhaze periods (NH1 and NH2) were selected. The
average PM
Generally, secondary sources were predominant (
The high-time-resolution source apportionment result by PMF was combined with the NAQPMS and footprint modeling outcomes to investigate the variation in source types and contributions with source regions in different haze episodes in Beijing. EP1 and EP4, with the longest duration and significantly different source compositions, were selected as two case episodes for further analysis.
Variation in sources and local contribution during EP1. The above pie charts show the daily local (Beijing as BJ) and regional contributions (labeled “others”). The pie charts below show the daily source type and contribution.
Figure 4 shows the variation in sources and local contribution and Fig. 5
shows the footprint regions and daily source apportionment results by PMF in
EP1. The spatial mass concentrations of PM
The relationship between source apportionment and the footprint model results can also be found in the daily variation of 17 November (see Fig. 5). From 01:00 to 12:00 of the day, the footprint remained in local areas, while primary sources were predominant. However, when the footprint changed to southwestern areas to Beijing from 13:00 to 18:00, the contribution of secondary sources increased significantly to 42 %. After the footprint changed back to a local type from 19:00 to 24:00, the secondary source contribution decreased to the previous level (19 %).
Figure 6 shows the variation of sources and Fig. 7 shows the footprint
regions and daily source apportionment results by PMF in EP4. In contrast to
EP1, the footprint in EP4 was mostly located in the southwestern area of
Beijing, where there were heavily polluted cities including Baoding and
Shijiazhuang (see Fig. 7). The daily local and regional contribution by
NAQPMS of this episode was not provided due to lack of data. From the
formation stage (16–17 December) to the peak (20 December) of EP4, the
contribution of secondary sources increased from 34 % to 58 %, while the
contributions of coal combustion and biomass burning were also significant
among primary sources (see Fig. 6). Figure 7 shows that the footprint on
December 17 was more concentrated in the local and eastern areas of Beijing,
while it gradually moved to southwestern areas along with the increase of
PM
Source contribution in EP4. The pie charts show the daily source type and contribution. Dates are mm/dd.
The above results confirmed that high-time-resolution source apportionment result can be integrated with footprint and NAQPMS model to identify the rapid evolution of different episodes – EP1 was mainly caused by local emission from transportation and coal combustion, while EP4 was typical of regional transport from southwestern areas of Beijing with increasing contributions of secondary sources.
Receptor models which are used for source apportionment have the limitation
that they cannot quantify the local or regional transport contribution.
Therefore, the receptor model was combined with the chemical transport model
NAQPMS to investigate the correlation of source contribution with
local or regional transport. As shown in Sect. 3.2, secondary and combustion
sources were predominant in haze episodes in Beijing. To better control
those major sources in winter, it is essential to determine the correlation of
source contribution with the contribution of local emission or regional
transport. Figure 8 shows the correlations of relative contribution of
secondary sources, coal combustion and biomass-burning sources by PMF with
local contribution by NAQPMS during the sampling period. The results showed
that, for PM
Correlations of local contribution by NAQPMS with the relative
contribution by PMF of
The combination of the PMF result with the footprint model was used to further
identify specific source type and contribution in different source regions.
As mentioned in Sect. 2.2.4, the footprint with the time resolution of 6 h was divided into four types (local, south, north and east) according
to the direction of potential source regions. The typical examples of
different types of footprint are shown in Fig. S5. The local footprint
referred to the cases with source region located within Beijing. The southern footprint mainly covered southwestern areas in Hebei province including
Baoding, Shijiazhuang and Xingtai. The northern footprint included Zhangjiakou
and Inner Mongolia. The eastern footprint covered the northern part of Hebei, such
as Tangshan and Qinhuangdao, and the southern part of the Liaoning province. The
local footprint was predominant in winter in Beijing (
The results of PMF and the footprint model showed that the source contribution in winter in Beijing was influenced by the potential source regions, and the predominant source could change specifically for different footprint types, which might suggest that source apportionment and footprint analysis need to be combined to better control specific sources from different source regions.
In this study, the high-time-resolution online measurements were conducted by
Xact, IGAC and the Sunset OCEC analyzer, which could measure inorganic
species including water-soluble ions, elemental components, OC and EC. As a
result, most of the tracers selected for PMF source apportionment were
inorganic species. In previous studies based on online measurements, organic
tracers are also not commonly used due to current technical difficulties in
carrying out online and quantitative measurements of organic species with
high-time resolution (Gao et al., 2016; Y. Li et al., 2017; Peng et al.,
2016). However, some organic tracers are believed to be more specific for
certain sources, such as levoglucosan for biomass burning, hopane and
sterane for traffic sources and cholesterol for cooking sources (Fraser et
al., 2000; Yin et al., 2010; Zhao et al., 2015). Therefore, future online
measurements of organic species could be conducted, which will be very
helpful in identifying sources. Besides, vertical measurements of PM
High-time-resolution online measurements of PM
The multiple models were combined to analyze the evolution of two typical
pollution episodes in Beijing. The high-time-resolution results indicated
that source contribution can vary rapidly and significantly with source
regions within different types of haze episodes. EP1, with a locally
concentrated footprint and high local emission, was characterized by coal
combustion and traffic sources, while EP4 with a more southwestern footprint was
typical of a high secondary source contribution. The relationship of
PM
The data in this study are available from the corresponding author upon request (mzheng@pku.edu.cn).
The supplement related to this article is available online at:
MZ, XC and JL designed the research. MZ organized the field campaign. YL, TZ and CY conducted the measurements. YL wrote the paper. YL, MY and HD analyzed the data. XW, ZS, RH, QZ and KH took part in data analysis and revised and commented on the paper. All authors contributed to the discussion of this paper.
The authors declare that they have no conflict of interest.
This article is part of the special issue “In-depth study of air pollution sources and processes within Beijing and its surrounding region (APHH-Beijing) (ACP/AMT inter-journal SI)”. It is not associated with a conference.
The authors gratefully thank for the assistance of Jinting Yu in Peking University for maintaining the online instruments in this work.
This research has been supported by the National Natural Science Foundation of China (grant nos. 41571130033, 41430646, 41571130035, 91744203, and 41571130034) and the UK Natural Environment Research Council (grant nos. NE/N006992/1 and NE/R005281/1).
This paper was edited by Leiming Zhang and reviewed by three anonymous referees.