Investigation of the Air Pollution Event in Beijing-Tianjin-Hebei Region in December 2016 Using WRF-Chem

/e online coupled weather research and forecasting model with chemistry (WRF-Chem) was used to investigate an air pollution event during December 2016 in Beijing-Tianjin-Hebei urban agglomeration. Evaluation indicates that WRF-Chem captured the main weather conditions and pollutant distribution in this event. /e primary meteorological drivers of air pollution formation were stationary atmospheric flows in both vertical and horizontal directions. High relative humidity and a strong temperature inversion accelerated event formation. In the shallow temperature inversion layer, aerosol particles were strongly confined near the surface, producing high surface contaminant concentrations. In addition, based on a normal experiment, three sensitivity experiments were constructed by adding hypothetical terrain (HT) of 400, 300, and 200 meters, over the region 115°E, 38.8°N to 117.54°E, 38.8°N. /e results indicate that pollutants were diffused and transported below 400 meters, and the pollutant amounts concentrated south of the HT because of the HT blocking effect. Nevertheless, because there were less total contaminants north of the HT in the normal run, there was a slight decrease in pollutants north of the HT./ere were some increases in pollution north of the HT because of local emissions, which were obstructed by the HT. /e higher the HT, the stronger the blocking effect.


Introduction
Beijing-Tianjin-Hebei (BTH) is a super urban agglomeration on the North China Plain, with an area of 21,600 kilometers and more than 100 million people [1].BTH has become one of the most heavily polluted areas in the world because of extensive industrialization and urbanization in the past few decades [2,3].High concentrations of fine particulates (PM 2.5 , with aerodynamic diameters not larger than 2.5 µm) substantially reduce visibility through scattering, and they penetrate the human respiratory tract [4][5][6][7].
e convergence of pollutants is favored by calm or weak winds [26][27][28][29].e formation of a temperature inversion is adverse to the dispersion of pollutants and exacerbates air pollution in BTH [23,[26][27][28].ere is a significant negative correlation between the concentration of pollutants and the height of the planetary boundary layer (PBL).A relatively low PBL height may aggravate air pollution [23,26,29].Further, aerosols can alter the local spatial temperature distribution through a direct effect [30][31][32], thus modifying local atmospheric circulations.is in turn changes the distribution of aerosols, complicating the development of a pollution event [24,26,[33][34][35][36]. From the perspective of dynamic mechanisms, Wu et al. [37] showed that strong and persistent sinking motion in the midlower troposphere over the BTH region can reduce the PBL height and generate a temperature inversion, allowing air pollutants to accumulate in the lower PBL.
Regional transport also affects air pollution incidents.In addition to local emissions, the contribution of regional transport mainly depends on wind direction [19].Taking Beijing as an example, southerly winds are more likely associated with severe air pollution than other wind directions because a portion of the pollutants and warm humid air can be carried from upwind areas to the city [18,20,23,24,[27][28][29][37][38][39].With prevailing northerly winds, pollutants are exported from Beijing to its south [19,38,39].On the contrary, in order to reduce pollution incidents, the Chinese government urges the BTH region and its neighboring area to adjust the energy structure in the way of closing down factories with high energy consumption and high pollution, promote the modification of desulphurization and denitrification, eliminate backward production capacity, and so on (http://www.hebqg.com/newsv.asp?nid�2025).Such kind of policy gives huge pressure to Hebei Province which relies much of its economic on the coal industry.Most regions in Hebei Province (especially the middle-south region) have to sacrifice a plenty of money in order to finish such kind of reformation of energy.A possible consequence is the stagnation of the economic development, as a result of which, a large number of workers will lose their jobs and local residents could hardly pass the cold winter without cheap heating system.For this reason, we hope to find the way which could guarantee the economic development and improve the air quality around Beijing-Tianjin with this hypothetical experiment.We conducted a hypothetical experiment using a WRF-Chem model to create a virtual terrain in Hebei Province.And trying to understand the impact of hypothetical topography on the distribution of pollutants in the BTH region when pollutants are based on the routine emissions.We investigated the heavy pollution event of 19 December 2016 to see whether virtual terrain can effectively block the transport of air pollutants.
e formation mechanism of the events was analyzed.
e spatial and temporal distribution and transmission regularity of air pollutants in the PBL were explored by setting up multiple virtual terrain experiments.

Model Configuration.
e weather research and forecasting model coupled with chemistry (WRF-Chem, version 3.5.1)was used in the study, which simulates transport, diffusion, wet and dry deposition, emissions, aerosol gasphase chemical reaction, liquid chemical reaction, photochemical reaction and radiation processes, and aerosols simultaneously with meteorological fields [40].As shown in Figure 1, two nested domains (D01 and D02) were introduced to the model with horizontal resolutions 18 and 6 km, respectively.e D01 area covers most of mainland China and Mongolia and Korea, which could be used to analyze the large-scale circulation background of the pollution event.e D02 area covers BTH and its vicinity and was used to study air pollutant diffusion and transport in that region.To simulate the diffusion of air pollutants in the PBL more accurately, we increased the vertical resolution so that each domain had 42 [41], Lin microphysics scheme [42,43], and rapid radiative transfer model for general circulation model longwave and shortwave radiation scheme [44] with aerosol direct effects.
e gas-phase chemistry model was based on the Regional Acid Deposition Model version 2 [45], and the Modal Aerosol Dynamics Model for Europe/Secondary Organic Aerosol Model [46,47] was used as the driver module for aerosol.Detailed model configurations and parameter options for physical and chemical processes are listed in Table 1.Meteorological initial and boundary conditions were from the 1 °× 1 °National Centers for Environmental Prediction Final Analysis dataset (https://rda.ucar.edu/datasets/ds083.2). e monthly 2010 Multiresolution Emission Inventory for China (http:// www.meicmodel.org/) was used for anthropogenic emissions.
is emission inventory contains emissions of SO 2 , NO X , CO, NH 3 , PM 2.5 , PM 10 , BC, OC, and nonmethane volatile organic compounds from five sectors (agriculture, industry, power plants, residential, and transportation), the diffusion (advection and convection), wet and dry deposition, and chemical reactions of which may produce secondary aerosols in WRF-Chem.Advances in Meteorology southerly winds over BTH, Beijing naturally becomes a pool for incoming pollutants because of its semibasin topography.We simulated the heavy air pollution episode in BTH from 00:00 UTC on 16 December 2016 to 20 December (all subsequent time references are in UTC, which is late by Beijing time for 8 hours).We integrated over 96 h, the first 24 h of which were treated as the spin-up period.We mainly used a hypothetical terrain (HT) experiment to evaluate the transport of pollutants across the BTH region.First, a control experiment (CTL) was conducted using normal terrain to explore the spatiotemporal distribution of pollutants.

Experimental Design.
en, on the basis of the CTL, three groups of sensitivity experiments were run within the domain 115 °E, 38.8 °N to 117.54 °E, 38.8 °N, adding a hypothetical 400 meters (HGT400), 300 meters (HGT300), and 200 meters (HGT200) to the terrain.Corresponding locations are shown in Figure 2 by a black dotted line.Four experiments altering only terrain height were run to study the effects of the HT.Furthermore, we performed an experiment adding a hypothetical 500 meters to the terrain at the same location.However, its results revealed features similar to the HGT400 experiment.One possible reason is that contaminants are mainly diffused and transported below about 400 meters.erefore, we only analyze the results of the CTL, HGT400, HGT300, and HGT200 simulations in this paper.

Model Evaluation
3.1.Surface Pressure Simulations.Figure 3 shows a comparison of observations and simulations for the large-scale background field.
e simulated data are from the D01 region in the CTL run, and the observed data are from the Hong Kong Observatory surface weather map (http://www.envf.ust.hk/dataview/hkowc/current/).According to the observations, there was a high-pressure center of 1030 hPa over the Yangtze River Delta (YRD) region at 18:00 on 16 December 2016.Inner Mongolia was dominated by a cold high-pressure system with a central pressure of 1028 hPa.Pressure in the BTH region was lower than that in the two high-pressure areas.With the passage of time, both pressure systems over the YRD and Inner Mongolia moved eastward.By 00:00 on 18 December 2016, the YRD surface pressure system reached Japan and continued to move eastward.In spite of surface pressure in the BTH region increasing slightly, it was still intermediate to the two high-pressure systems.
is caused a prevailing weak southerly wind southeast of BTH and a weak northerly wind to its northwest.erefore, warm, moist air and pollutants may have been advected southeast of BTH, and cold, dry air transported to the northwest of BTH.By 00:00 on 19 December, pressure at BTH increased gradually, which is attributed to the cold high pressure over Inner Mongolia moving eastward slowly.At 18:00 on 16 December, the simulated highpressure center of the YRD was similar to the observation.Although both the location of the low-pressure center around BTH and cold high-pressure center over Inner Mongolia are somewhat westward, simulated and observed values are nearly in agreement.At 00:00 on 18 December, simulated highpressure system over South Korea was slightly west of the observed system, which may have been caused by the difference in size of the D01 area and synoptic chart.Fortunately, the performance for the BTH area between the two highpressure systems was good.At 00:00 on 19 December, the simulated pressure over BTH was ∼1026 hPa, 2 hPa lower than the observation.However, the simulations generally reproduced the large-scale background of the air pollution event.

Comparing Simulations with Observations.
In this section, a comparison between the model results and observations for the time series of hourly relative humidity and temperature 2 m above the surface (RH2 and T2), surface PM 2.5 , NO 2 , and SO 2 concentration during from 04:00 on    4

Advances in Meteorology
Environmental Protection and can be downloaded from the website https://www.zq12369.com/index.php.As illustrated in Figure 4 and Table 2, the model basically reproduces time and spatial distribution of these meteorological and chemical variables.For the RH2, the model can depict its temporal variation but slightly overestimates the values at the Beijing, Tianjin, and Shijiazhuang sites, and the corresponding mean bias (MB) are 13.99%, 4.04%, and 12.64%, respectively.Meanwhile, the observed RH2 are also high at the four sites, with observation mean (C OBS ) � 70.51 to 90.16%.e high RH2 may be conducive to secondary growth of air pollutants [23][24][25].e observed T2 at the four sites are also reasonably reproduced by the model (rootmean-square error (RMSE) � 2.02 to 4.44 and normalized mean error (NME) � 0.01 to 0.03).However, the model overestimates the T2 values at the four sites, especially at the Baoding site where the MB is 3.38 °C. is may be attributed to the coarse resolution of the NCEP-FNL initial field.e initial and boundary meteorological conditions may also influence on the results of the simulations.In addition, the magnitudes and trends over time of the simulated PM 2.5 , NO 2 , and SO 2 are basically consistent with measurements (Figure 4).From Table 2, the model overestimates surface PM 2.5 concentration at the Beijing and Tianjin site (MB � 54.73 and 48.11 µg/m 3 , respectively) and underestimates it at the Shijiazhuang and Baoding site (MB � −68.62 and −8.62 µg/m 3 , respectively).In this air pollution period, surface PM 2.5 concentration is extremely high at the Shijiazhuang site, with model mean C MOD and C OBS as 521.95 and 581.57µg/m 3 , respectively.
is may suggest more pollutants over the southern part of Hebei and transport of pollution to Beijing.In general, the model also reasonably reproduces the hourly variation of surface SO 2 concentration during from 04:00 on December 18 to 16:00 on December 19 2016 (RMSE � 20.63 to 50.70 and NME � 0.48 to 0.69).NO 2 is simulated very well in these three chemical variations, with MB � −5.95 to 27.76 µg/m 3 and normalized mean bias (NMB) � −0.05 to 0.16 at the four sites.Despite good simulation for these chemical variables, some deviations still exist.For example, the model underestimates the surface PM 2.5 concentration at the Shijiazhuang site during from 01:00 to 16:00 on December 19. is may be attributed to the uncertainties of emission inventory, coarse horizontal and vertical resolutions, inexact  Advances in Meteorology meteorological variables, and undiscovered processes of atmospheric chemistry [24].Overall, however, the simulated results generally captured the heavy air pollution event that provides some reliability for the later sensitive experiments.

Analysis of Meteorological Elements
Figure 5(a) shows that RH2 exceeded 95% southeast of BTH but was low in the Taihang and Yanshan Mountains, which is attributable to the fact that water vapor content is less at higher altitudes and that a strong northwest wind was advecting dry and cold air over the two mountain ranges.Surface wind speed was <2 m/s on the BTH plain, except for the eastern Shijiazhuang area, where there was a strong southerly wind band that converged with the weak north wind over Baoding.erefore, air pollution built southeast of BTH because of this meteorological situation (Figure 6(a)).
e RH2 declined by 07:00 on 19 December, which can be attributed to a warming surface temperature caused by increasing solar radiation (Figure 5(b)).An anticyclone formed southeast of BTH, which reduced the water vapor content and air pollution through divergence, particularly to the south of Tianjin, Cangzhou, Hengshui, and Langfang  .By 13:00 on 19 December (Figure 5(c)), there was a large area of calm or weak wind in the BTH region, and surface temperature was reduced by surface effective radiation, Hence, surface RH2 increased.Air pollution had not increased (Figure 6(c)), despite the high humidity and weak wind at BTH at this time.However, an hour later, concentrations of air pollutants gradually increased.is indicates that the increase of air pollution slightly lagged the weather conditions (figure not shown).Overall, a weak wind field is conducive to the persistence of water vapor and air pollutants.
An accelerated formation of secondary aerosols may be attributed to hygroscopic contaminants in aqueous-phase chemical reactions under high humidity, consistent with other measurement and simulation studies [23-25, 27, 48].Advances in Meteorology surface) as the horizontal distribution of near-surface PM 2.5 .Figure 6(d) shows an important phenomenon in which pollutant concentrations in southeastern Beijing, western Tianjin, north central Tangshan, northern Langfang, and southeastern HT had all increased, whereas there were decreases in southern Tangshan, eastern Tianjin, and over the Bohai Sea. e increases were mainly associated with the HT blocking effect, which reduced the spread of local pollutants.Decreases for the Tangshan and Tianjin coastal areas were also because of that effect, which prevented air pollutant transport from the southern HT to its north.However, the Bohai Sea area had reduced air pollutant concentrations.is may be attributed to the reduced pollutants in the Tangshan and Tianjin coastal areas, such that fewer pollutants were transported over the Bohai Sea, even with northwesterly and westerly winds.In addition, there was an extreme reduction of air pollution in the Baoding area, which may be attributed to the greatly enhanced convection near the HT region (Figure 7(a)).

Sensitivity Experiment Analysis
us, a large amount of contaminants was diffused at a high altitude, and they decreased dramatically at the surface.At 07:00 on 19 December (afternoon in Beijing) (Figure 6(b)), the contaminants decreased because of stronger wind speeds, warmer temperatures, and lower relative humidity.e pollutants concentrated south of the HT under the blocking effect, but north of the HT, there was not much decrease (Figure 6(e)).
is may be explained by the few contaminants north of the HT at that time.e causes of the change in contaminants over the Bohai Sea are similar to 22:00 on 18 December.e Bohai Sea breeze direction gradually changed from south to southwest between 11:00 and 13:00 on 19 December, reducing pollutants over Cangzhou because they were transported over the sea (Figures 6(c)).Furthermore, the air pollutant concentration in southern HT increased, and the northern concentration decreased gradually with increase of HT height (Figures 6(e), 6(f ), 6(h), 6(i), 6(k), and 6(l)).In addition, the HT clearly reduced the diffusion of local air pollutants, such as those in southern Beijing and western Langfang as seen in comparison to Figures 6(d), 6(g), and 6(j).

Vertical Profile Analysis.
Figure 7 shows a vertical cross section of air pollutants.e black solid line in Figure 2 gives the horizontal location of the intercepted vertical cross section.For better comparison, all the intercepted times were the same at 22:00 on 18 December.To more clearly show variation of the vertical wind field, the vertical wind was exaggerated by 100 times.Figure 7(a) reveals that the wind direction was essentially horizontal.e maximum horizontal wind speed is 4.7 m/s and the minimum 0.002 m/s, which indicates that the stratification throughout the column was very stable.is was also a key indicator of air pollution formation.Temperature gradually decreased from south to north.In addition, a strong temperature inversion made it difficult for contaminants to disperse to high altitudes, so they mainly accumulated below 300 m. ere was a relatively small amount of pollutants atop the inversion layer, which remained between 300 and 400 meters in height.From the HGT400 experiment (Figure 7(b)), substantial air pollution concentrated south of the HT and decreased steeply to its north because of the blocking effect.At the same time, the height of the inversion layer rose because of the HT, but pollutants remained below the top of temperature inversion.
ere was a large amount of contaminants diffused near the south side of the HT because of much stronger wind speeds and reverse flow.One can see that with the less weak wind speeds and reverse flow there with lower HT height (as compared with Figures 7(b)-7(d)), contaminants increased gradually there.ere were two large-value centers on the north side of the HT, which were mainly responsible for local emissions.However, the lower HT height caused more pollutants from the south to cross the HT to the north side, increasing the two large-value centers.

Vertical Integration Concentration Analysis.
e air pollutants were not only near the surface but also aloft, so we examined variation of the vertically integrated concentration.e equation for this concentration is where P * is the pressure difference between the surface and model top, C(i, j, k) is the pollution concentration, and Δσ is the sigma level difference.e distribution of PM 2.5 vertical concentration is basically consistent with that of the PM 2.5 concentration near the ground, as seen by comparing Figures 8(a)-8(c) with Figures 6(a)-6(c).is means that the pollutants were mainly near the ground and the concentration of pollutants at high altitude was low.At 22:00 on 18 December, southeastern Beijing, northwestern Tianjin, and northern Langfang air pollution decreased at high altitude for a HT of 400 meters, as seen by comparing Figure 8(d) with Figure 6(d).However, there were areas of strong air pollution at high altitude around Baoding, owing to strong horizontal and vertical wind speeds near the HT region, which may be attributed to the low resolution or model drawbacks as mentioned in Sections 5.1 and 5.2.In addition, from Figures 8(e), 8(f ), 6(e), and 6(f ), most of the pollutants concentrated near the surface during afternoon and night because of the blocking effect and stagnant meteorological conditions.Moreover, in contrast to Figures 8(d)-8(l), the lower the HT height, the more the contaminants were transported from south to north by crossing the HT.

Conclusions
We acquired encouraging results from the fully coupled WRF-Chem model, which was used to investigate the effects of HT height on air pollutant transport based on four experiments with differing HT heights.ese experiments were designated CTL, HGT400, HGT300, and HGT200.e results show that warm, moist air and air pollutants were transported southeast of BTH because of high pressure in the northwestern Pacific, which increased the near-surface Advances in Meteorology relative humidity and aggravated pollution.Stable weather conditions including a strong temperature inversion and weak wind are unfavorable for pollutant dispersion.
According to the four simulations, the pollutants were mainly diffused and transported below a height of 400 meters.
e results also show the strong effect of HT on pollutant transport, which caused substantial pollutant concentration south of the HT.However, there was a slight decrease on the north side of the HT, which may have been caused by fewer total pollutants to the north of the HT.ere were even some increases north of the HT because local emissions were obstructed by that terrain.Comparison of the simulations shows clearly that the greater the height of the HT below 400 meters, the stronger the blocking effect.
In essence, we still need to replace energy sources such as coal with clean energy to ensure the blue sky. is hypothetical experiment is impossible to achieve in real life, and we just hope that through hypothetical experiments to understand the distribution of pollutants in BTH area.e current experiments can be improved in several aspects.e model resolution was 6 km, which may be not adequate to resolve the PBL.e use of supercomputers should allow  10 Advances in Meteorology simulations of air pollution events at greater resolution, possibly showing even greater sensitivity to the PBL.In this research, the observation data are primarily obtained at surface station only, and this may reduce the credibility of the simulations.Implementation of detailed PBL processes, accurate descriptions of pollutant transport, and using more observation data such as the radiosondes and satellite products will be the next step toward improvement of simulation by WRF-Chem.

Data Availability
e National Centers for Environmental Prediction Final Global Tropospheric Analysis (NCEP-FNL) data were taken from https://rda.ucar.edu/datasets/ds083.2. e data used to support the findings of this study are available from the corresponding author upon request.

Figure 2 :
Figure 2: Topographic map of D02 (unit: m).Shading indicates terrain height, except for dark blue showing Bohai Sea.Black dotted line indicates location of hypothetical terrain, and solid black line indicates vertical cross section corresponding to Figure 7. Dotted circles indicate cities.

Figure 3 :
Figure 3: Surface weather chart from Hong Kong Observatory (a).For large-scale background field simulated in D01 (b), contour interval of sea-level pressure is 2 hPa, and black arrows show wind field 10 m above the surface; times are 18:00 on 16 December, 00:00 on 18 December, and 00:00 on 19 December 2016.

Figure 4 :
Figure 4: Simulated and observed hourly relative humidity and temperature 2 m above the surface (RH2, unit: %; T2, unit: °C), surface PM 2.5 , NO 2 , and SO 2 concentration (unit: µg/m 3 ) during from 04:00 on December 18 to 16:00 on December 19. e blue star indicates observed, and the red solid line indicates simulated.ese four columns indicate (a) Beijing site, (b) Tianjin site, (c) Shijiazhuang site, and (d) Baoding site from left to right, respectively.

Figure 7 :
Figure 7: (a) Vertical cross section of PM 2.5 diffusion from CTL, and (b) differences of HGT400 and CTL simulations.(c) Differences of HGT300 and CTL simulations.(d) Differences of HGT200 and CTL simulations.Shading indicates PM 2.5 concentrations (unit: µg/m 3 ), contours show temperature (unit: °C), black arrows depict wind field, and green solid circles on the horizontal axis indicate locations of Hengshui and Beijing.Time is 22:00 on 18 December 2016.
December 18 to 16:00 on December 19, 2016, is presented.To evaluate better the performance of the model in this study, Figure4is the time series of simulated and observed hourly averaged T2, RH2, PM 2.5 , SO 2 , and NO 2 at Beijing, Tianjin, Shijiazhuang, and Baoding, respectively.ecorrespondingstatisticalanalysis results of the comparisons between simulated and observed are presented in Table2.e observation data are from China's Ministry of 18:00 Dec 16 Hong Kong Observatory 00:00 Dec 18 Hong Kong Observatory 00:00 Dec 19 Hong Kong Observatory

Table 2 :
Performance statistics for meteorological and chemical variables from 04:00 on December 18 to 16:00 on December 18. a is the number of paired samples.C OBS and C MOD are the time series of the average of observation and model results.MB is the mean bias between the observation and model results.RMSE is the root-mean-square error of observation and model results.NMB is the normalized mean bias between the observation and model results.NME is the normalized mean error between the observation and model results. N