The North China Plain (NCP) has been experiencing severe air
pollution problems with rapid economic growth and urbanisation. Many field
and model studies have examined the distribution of air pollutants in the
NCP, but convincing results have not been achieved, mainly due to a lack of
direct measurements of pollutants over large areas. Here, we employed a
mobile laboratory to observe the main air pollutants in a large part of the
NCP from 11 June to 15 July 2013. High median concentrations of sulfur
dioxide (SO
The North China Plain (NCP) is a geographically flat region in the northern
part of eastern China, which includes Beijing, Tianjin, most of Hebei, Henan
and Shandong provinces and the northern parts of Anhui and Jiangsu provinces.
This region is surrounded by the Yan Mountains to the north, the Taihang
Mountains to the west and the Bohai Sea to the east. The NCP covers an area
of 300 000 km
Over the past decade, there have been a number of investigations of air pollution in the NCP, taking advantage of observation sites, aircraft measurement platforms, mobile laboratories, satellite data and air quality models. In the NCP, a network of observation sites has been built for air pollution research, mostly located in and around large cities, particularly Beijing (Xu et al., 2011, 2014; Wang et al., 2013; Meng et al., 2009; Shen et al., 2011; Lin et al., 2011; Li et al., 2015). Variability, sources and meteorological and chemical impacts of air pollutants have been discussed by analysing these observational results. The concentrations of long-lived pollutants have been shown to be significantly influenced by wind, particularly the south and north winds, indicating that regional transport plays an important role in urban air pollution. In addition, model studies have yielded similar results in various areas of the NCP (An et al., 2007; Zhang et al., 2008; Liu et al., 2013). Satellite data have indicated that regional transport has a significant impact on the haze period in the NCP (Wang et al., 2014). Thus, it is necessary to understand regional transport to address air pollution problems in the NCP, which will require data on the distribution of air pollutants in this region.
However, observational data from a single or limited number of measurement
sites cannot present the whole picture of air pollution in the NCP. A number
of mobile laboratory measurements (Johansson et al., 2008; Li et al., 2009;
Wang et al., 2009, 2011) and aircraft measurements (Huang et al., 2010; Zhang
et al., 2011, 2009, 2014) have been used to determine pollution
distributions,
mainly within the megacity of Beijing. There have been several reports of
model and satellite studies on the air pollution distribution in the NCP, or
even the whole of China (Wei et al., 2011; Zhao et al., 2013; Ying et al.,
2014; Ding et al., 2015, 2009). However, there are disagreements between
these results, e.g. regarding the distributions of NO
In this study, we measured the concentrations of nitrogen oxides (NO
A mobile laboratory was built by our research group, details of which were previously described (Wang et al., 2009). Briefly, this mobile laboratory was constructed in 2006 on an IVECO Turin V diesel vehicle (length 6.6 m, width 2.4 m, height 2.8 m; payload 2.7 t). Instrumentation was powered by two sets of uninterruptible power systems (UPS), consisting of three series of 48 V/110 Ah lithium batteries, which could support all of the equipment operations without interruption for up to 5 h. The inlet systems for our mobile laboratories were specifically configured to accommodate the type of measurement requirements and the instrument suite to be employed in specific field campaigns.
Instruments deployed on the mobile laboratory included those for studying
NO
Each time before an experiment, we did a calibration to obtain calibration
curves, e.g. on 16 June in 2013 (Fig. S2 in the Supplement), and after the experiment we did
another calibration and recorded the span drifts. The span drifts were less
than 10 %. For example, according to the calibration on 23 June in 2013,
the span drifts of NO, SO
Ultrafine particles were measured using a fast mobility particle sizer (FMPS,
TSI 3090; TSI, Shoreview, MN), which covers particle sizes from 5.6 to
560 nm in 32 channels with a time resolution of 0.1 s. The data were
recorded on a designated computer. Other auxiliary data including temperature, relative
humidity, barometric pressure and GPS coordinates were also measured. The
driving speed was kept stable at around 100
Precisions and uncertainties of air pollutants analysers used in our experiments.
To establish the spatial distribution and characterise the regional transport of air pollutants in the NCP, the routes for the mobile measurements were specially designed to cover important emissions hotspots (Fig. 1) and to map large areas of the NCP. The routes included the municipalities of Beijing and Tianjin, most of Hebei province, and part of Shandong province, which is about 300 km wide from the west to the east and 400 km long from the north to the south, covering most of the NCP. To avoid traffic jams and rough roads, only motorways were chosen for all routes. Limited by the duration of battery power and the variability of boundary layer height, we could not cover all routes in one trip. Instead, we divided the routes into five parts. Route 1 was along the Taihang Mountains from Beijing to Shijiazhuang, located in the western part of the NCP. Routes 2 and 3 were from Shijiazhuang to Dezhou and Cangzhou to Baoding, respectively, which were generally located in the central NCP. Routes 4 and 5 were from Tianjin to Beijing and around the south of Beijing, located in the northern NCP. In addition, we ran each route in 1 day. Two days were also needed for calibration and maintenance of instruments. Therefore, it took 1 whole week to conduct a single experiment. In total, six experiments, including one pre-test study, were designed from 1 June to 15 July 2013. The pre-test study was conducted between 1 and 7 June, and five formal repeated experiments were conducted between 11 June and 15 July (Experiment 1 (E1), 11–15 June; E2, 17, 18 and 20 June; E3, 24–25 June; E4, 2–7 July; E5, 11–15 July). All trips were started at about 09:55–10:50 LT (local time) and ended at about 13:05–15:05 LT to ensure that the boundary layer was relatively stable during the observation period in 1 day.
The major reasons for data lacking were the computer crashing and rain. Rain caused the missing data on routes 3, 4 and 5 in experiment 3, and on route 5 in experiment 5. The computer crashing caused the missing data on routes 4 and 5 in experiment 2 (Table S1). The computer crashing also caused missing data during every trip (Table S2).
The study area in NCP. The red track shows route 1, from Beijing to Shijiazhuang. The purple track shows route 2, from Shijiazhuang to Dezhou. The black track shows route 3, from Dezhou to Baoding. The blue track shows route 4, from Cangzhou to Zhuozhou. The yellow track shows route 5, from Zhuozhou to Beijing. The red dots on the map present the major cities near the routes. The yellow stars present the monitoring sites.
A Lagrangian particle dispersion model, FLEXPART-WRF version 3.1 (Brioude et
al., 2013; Stohl, 1998; Stohl et al., 2005; Fast and Easter, 2006), was used
to determine the origin and transport pathways of the air mass arriving at
the vehicle-based mobile measurement laboratory. The wind field used to drive
FLEXPART was the time-averaged wind provided by the WRF (Weather Research and
Forecasting) model, with temporal intervals of 10 min and
horizontally spatial resolution of 2 km. (The details of the mesoscale
meteorological model are described in Sect. S2.1 in the Supplement.) FLEXPART
simulates the transport and dispersion of tracers by calculating the backward
trajectories of multitudinous particles, which are termed plume back
trajectories. In this model, turbulence in the planetary boundary
layer is parameterised by solving
the Langevin equation, and convection is parameterised using the
Živković-Rothman scheme (Stohl et al., 2005). To improve the accuracy
of the trajectory calculation, we used high-resolution WRF simulation
domain 4 outputs as the input meteorological conditions for the FLEXPART
model. The turbulence, convection and boundary layer height were computed
along the trajectories of tracer particles using the WRF output data.
Backward integration was performed every 5 min during the mobile observation
period in June 2013. For each integration, 2000 stochastic particles were
released initially from within a box
The footprints of backward trajectories were calculated to present plume
trajectories. Footprints in this context refer to the total residence times
of released particles, which were calculated following Ashbaugh et al. (1985) and de Foy et al. (2009) by counting the accumulated number of
particles during the integration within each cell of a
Concentrations of air pollutants, including NO
The main pollutants at these sites were measured using commercial
instruments. At the QZ site, gas analysers were used to measure NO
Fire data were obtained from the Moderate Resolution Imaging
Spectroradiometer (MODIS) installed in Terra and Aura. The territory passing
times were 10:30 and 13:30 LT for Terra and Aura,
respectively. Fire images were obtained from EOSDIS Worldview (NASA,
The concentrations of SO
The spatial distributions of the measured concentrations in our
study and the emission maps of SO
BC, NO
The levels of CO, NO
The concentrations of NO
As shown in Fig. 3, the concentrations of BC, NO
During the five experiments, no clear temporal distributions of air
pollutant concentrations in the NCP were seen, except for the significantly
low levels of NO
In summary, our mobile laboratory measurements indicated spatial
distributions of the pollutants SO
The levels and distributions of air pollutants in the NCP are mainly attributable to three sources, i.e. regional transport, local emissions and traffic emissions. On-road measurements, however, could be greatly affected by traffic emissions (Wang et al., 2009). The influence of traffic emissions on our mobile laboratory measurements is discussed below.
Each on-road measurement trip started from a car park in a motorway service station and ended at the car park of another service station. The car parks are about 150 m away from the motorway, such as those in service stations Dezhou (DZ) and Xizhaotong (XZT) (Fig. S3). Using the difference of the concentrations of air pollutants measured in car parks and on motorways, we can estimate the level of the enhancement of the concentrations of air pollutants on motorways for backgrounds greater than on a regional level.
Table 2 shows the 5 min averaged concentrations of NO
The concentrations of NO
The 175 % enhancement of SO
Apparently, vehicular emission is the major source that led to the 82 to
1658 % enhancement of NO
Figure 4 shows the concentrations of NO
Based on the reported vehicular emission factors (Shen et al., 2015; Cai and
Xie, 2007, 2010; Lei et al., 2011) and the vehicle composition (Chinese
Automotive Technology & Research Centre, 2015) in Hebei province, where
we conducted the most of the mobile measurements, we estimated the weighted
vehicular emission factors on the motorways (Table 3). The factors for
NO
The time series of the concentrations of NO
Estimated weighted vehicular emission factors of CO, NO
The regression curve of the means of concentrations of NO
A strong correlation (
Each on-road measurement trip started from a car park in a motorway
service station and ended at the car park of another service station. The
driven speed of the mobile platform was never lower than 80 km h
During our measurements, as shown in Fig. 3, we always measured high
concentrations of air pollutants near large cities. As the time series of the
concentrations of NO
In summary, both the comparison of on- and off-road measured concentrations
and the vehicular emission factors provided evidence as follows.
Vehicular emission is the main source of the enhancement of the on-road
concentrations of NO The high enhancement of NO CO and SO The difference in the enhancement is the difference of level of
background contribution. NO
The time series of the concentrations of NO
Thus, the CO and SO
Trip-average concentrations of SO
Local emissions and regional transport are the two main sources of
pollutants in the NCP (Xu et al., 2011). As stated above, local emissions in
large cities had a major impact on the air quality in their adjacent areas.
We found that all air pollutants measured near large cities were at a high level.
According to the analysis in Sect. 3.2, they were mainly from local
emissions of cities. In addition, according to the estimated emission inventories,
these cities were major sources of air pollution that caused the
concentrations of air pollution to be high around them. Regional transport also
plays a major role. Our study demonstrated that the contribution of regional
transport could vary both spatially and temporally, depending on a number of
parameters, such as prevalent wind, terrain and vertical mixing. We also
roughly divided the NCP into two parts according to these parameters, i.e. the northern border area and the central area, to discuss the influence of
regional transport on air quality. All trip-average concentrations of
SO
The northern border area of the NCP included major parts of routes 4 and 5, and the western border areas of the NCP included a major part of route 1. The area is surrounded by the Taihang Mountains to the west and the Yan Mountains to the north. The north wind prevailed in the winter and the south wind prevailed in the summer in this area.
During the measurements, the three routes experienced both north and south
winds. Specifically, northwest winds and east winds brought air
masses from outside the NCP from Northeast China and the Bohai Sea to the northern border area on
2 July (route 1) and 7 July (route 5), respectively (Fig. 7). On both trips,
the concentrations of SO
It is worth noting that the BC concentration was not lowest on 2 July, which
was the opposite of the observations for the gas pollutants, SO
On 24 June (route 1), 14 June (route 4) and 15 June (route 5), the air
masses were transported inside the NCP from the southern NCP to the northern
border areas by south winds (Fig. 7). Under these wind conditions, the
concentrations of SO
The back trajectories of observed air masses in the borders of the NCP on 14, 15 and 24 June and 2, 6 and 7 July.
Although route 4 on 6 July was under the influence of south winds, as the
same route on 14 June (Fig. 7), the concentrations of SO
In conclusion, for the northern border area, local emissions and regional
transport from other NCP areas due to south winds were two main sources of
long-lived pollutants; both north and east winds had significant dilution
effects on the concentrations of gas pollutants. This is based on a general
situation that the air quality in the north border of the North China Plain
was clean because the emissions of air pollutants were low in that area
(Fig. 3). The wind dependency scatter plots for SO
Wind dependency scatter plots of concentrations of SO
The central NCP consisted of routes 2 and 3, where numerous heavily polluted cities are located. The area was surrounded by the Taihang Mountains to the west or emissions hotspots in other directions. While the north wind prevailed in the winter, as for the northern border areas, low pressure prevailed in the summer with south and northeast winds in this area.
During the observation period, the measurements along the two routes experienced different wind fields. Route 2 experienced southwest winds on 12 June and 3 July, northeast winds on 18 June and a low-pressure system with south and northeast winds on 25 June (Fig. S6). Unlike the northern border area of the NCP where strong north winds had a dilution effect, the concentrations of gas pollutants were mostly high regardless of the wind direction on route 2 in the central NCP, e.g. on 18 June and 3 July (Fig. S6). Generally, our observations were reasonable according to the unique terrain and emissions map in the central NCP. Due to the heavy emissions level in the central NCP and surrounding areas, pollutants readily accumulated to high levels on their way to the central NCP in air masses from all directions, such as the clean air masses from the Bohai Sea, western China and Northeast China, and polluted air masses from Southeast China.
The back trajectories of observed air masses in the central NCP on 13 June and 4 July.
The situation was slightly different in areas along route 3, particularly for those off the coast of the Bohai Sea. Route 3 experienced east winds on 14 July (Fig. S6), and the concentrations of pollutants were low (Fig. 2). This was not only because of the wet deposition from the rain on that day, but also because of the transport of clean air from the Bohai Sea. A featured peak of aerosol number density at around 20 nm (Fig. S5) further confirmed the incoming air from the Bohai Sea (Haaf and Jaenicke, 1980; Hoppel et al., 1986).
Vertical mixing can also affect the concentrations of pollutants. For example, while the wind fields were similar on route 3 on 13 June and 4 July (Fig. 9), the concentration of pollutants on 13 June was lower than that on 4 July (Fig. 2). This was because the boundary layer was much higher on 13 June (976 m) than on 4 July (626 m), and the strong vertical convection diluted the air pollutants.
Specifically, the relative contributions of emissions and regional transport
to the local air pollution levels were slightly different on different
routes. At the junction of routes 1 and 2 around Shijiazhuang area, the
concentrations of air pollutants were always high, except in the last
experiment when wet precipitation occurred. The local emissions contributed
significantly to the air pollution levels in this area. The city of
Shijiazhuang is known as an emissions hotspot with heavy coal consumption.
Previous model results showed that Shijiazhuang is an important emissions
hotspot of SO
Similar to the situation in the Shijiazhuang area, transport convergence would occasionally pass through other cities along routes 2 and 3. In a typical case on 13 June (Fig. 9), air masses were transported far from the southwest of the NCP along the transport convergence through Shijiazhuang and reached the area on route 3, with air pollutants accumulating during transport and showing high concentrations.
Overall, in most areas in the central NCP, regional transport would play
essential roles in determining the local air pollution levels, although the
underlying mechanisms were different for the transport convergence area,
central NCP area and coastal area. The wind dependence scatter plots for
SO
We discussed five impactors: local emission, precipitation, location, wind
direction and boundary layer height. The influence of local emission
was reflected in the spatial distribution of concentrations (Fig. 3). Hotspots
were found near cities. However, for route-average results (Fig. 2), local
emission plays a minor role in the distribution of concentrations because
the routes were mainly in suburbs. The large reduction in SO
A mobile laboratory was employed to obtain snapshots of the spatial distributions of air pollutants in the NCP. The concentrations observed were at the highest levels in the world and were distributed unevenly in the NCP. Most high concentrations of air pollutants, i.e. 95 percentile concentrations, were found near emissions hotspots, which suggested the influence of local emissions. However, regional transport of air pollutants was also considered significant in determining the air quality in the NCP. Back trajectory analysis, satellite data and tracer pollutants were combined to recognise various cases of regional transport in both the northern border and the central NCP. Where the border areas would occasionally be diluted by winds from outside the NCP, the central NCP was affected by regional transport of air pollutants with a few exceptions, such as when precipitation occurred. To achieve the aims of air quality locally, emissions control policies must consider the whole emissions budget in the NCP.
The data of mobile and stationary measurements are available upon request. We are also working on installing a website this year. We will provide data of GPSs, vehicle speed, meteorology and concentrations of air pollutants.
T. Zhu, Y. Zhu, Y. Han and W. Chen designed the experiments. T. Zhu secured the research grants. Y. Zhu, Y. Han and W. Chen carried out the experiments. J. Zhang developed the model code and performed the simulations. J. Wang managed the data in the program. J. Liu provided the emission maps. L. Zeng, Y. Wu, X. Wang, W. Wang and J. Chen provided the data of stationary measurements. Y. Zhu analysed the data with contributions from all co-authors. Y. Zhu prepared the manuscript with help from T. Zhu, C. Ye and Y. Li.
This study was supported by the National Natural Science Foundation Committee of China (21190051, 41121004, 41421064), the European Seventh Framework Programme Project PURGE (265325), the Collaborative Innovation Center for Regional Environmental Quality and the National Key Research and Development Program of China (2016YFC0209000). Edited by: D. Heard Reviewed by: two anonymous referees