Comparative research on visibility and light extinction of PM 2.5 components during 2014–17 in the North China plain

Xinrui Wu ,, Jinyun Xin ,,, Xioling Zhng , Ruirui Si , Gungjing Liu , An’n Li ,,Tinxue Wen , Zirui Liu , Shigong Wng , Gungzhou Fn ,, Yuesi Wng , Lili Wng ,Wenkng Go

a Plateau Atmosphere and Environment Key Laboratory of Sichuan Province, School of Atmospheric Sciences, Chengdu University of Information Technology, Chengdu,China

b State Key Laboratory of Atmospheric Boundary Layer Physics and Atmospheric Chemistry (LAPC), Institute of Atmospheric Physics, Chinese Academy of Sciences,Beijing, China

c Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters, Nanjing University of Information Science & Technology, Nanjing, China

Keywords:PM2.5 Chemical component Visibility IMPROVE algorithm Light extinction

ABSTRACT Severe air pollution with visibility deterioration has long been a focus in the North China Plain (NCP).In this study, concentration and light extinction analysis of PM 2.5 chemical components were carried out from 2014 to 2017 to study the pollution characteristics in Baoding, a case city of the NCP.The annual average concentration of total PM 2.5 components showed a declining trend, decreasing by 11 μg m ? 3 (water-soluble inorganic ions), 23 μg m ? 3 (carbonaceous aerosols), and 1796 ng m ? 3 (inorganic elements).Contributing 82.9%to the concentration of total ions, the dominant components, NH 4 +,NO 3 ?,and SO 4 2 ? became the main pollutants in PM 2.5 pollution.Based on the IMPROVE algorithm, the average reconstructed PM 2.5 mass concentration was 93 ± 69 μg m ? 3 during the observation period.Meanwhile, the light extinction coefficients were 373.8 ± 233.6 M m ? 1,405.3 ± 300.1 M m ? 1,554.3 ± 378.2 M m ? 1 and 1005.2 ± 750.3 M m ? 1,in spring,summer, autumn, and winter, respectively.Ammonium sulfate, ammonium nitrate, and organic matter were the largest contributors to light extinction, accounting for a total of 55%–77% in the four seasons.The b sca (light scattering by particles and gases) reconstructed from PM 2.5 components (R b sca ) and the b sca converted from visibility (V b sca ) were compared to evaluate the performance of the IMPROVE algorithm, revealing a high correlation coefficient of 0.84.The high values of V b sca were underestimated while the low values were overestimated, as determined through comparison with the one-to-ne line.Especially, when R b sca > 1123 M m ? 1 (corresponding to< 2.0 km, approximately), V b sca was underestimated by 17.6%.PM 2.5 mass concentration and relative humidity also had an impact on the estimation.

1.Introduction

Located in the North China Plain (NCP), the Beijing–Tianjin–Hebei(BTH) region is not only the political and economic center of China,but also a strategic area.In recent decades, the government has spared no effort in coping with the problem of atmospheric pollution in this region, which has worsened in line with its soaring economy( Cheng et al., 2017 ; Ji et al., 2012 ; Li et al., 2017 ; Wang et al., 2018 ).The launch of the Action Plan for Air Pollution Prevention and Control in 2013 marked the full start of the Blue Sky Protection Campaign( Li et al., 2019 ; Zong et al., 2018 ).After that, a series of toughest-ever actions have been implemented to reduce PMpollution.Industrial facilities have been upgraded or shut down, emission standards have strengthened, clean fuels in residential sectors have been promoted,and an “odd–even number ”vehicular ban has been put into practice( Chen et al., 2019 ; Liu et al., 2019 ).Accordingly, numerous studies have documented dramatic changes in the atmospheric environment during this period ( Li et al., 2019 ; Zhang et al., 2018 ).For instance,Zhang et al.(2019) studied the air quality improvement in China during 2013–17 and found that the aforementioned policies had reduced national emissions of SOand NOby 59% and 21%, respectively.Primary PM 2.5 had decreased by 33% nationally.Cheng et al.(2017) investigated the residential energy use in the BTH region during 2013–14 and found that Baoding was the largest coal contributor to regional residential coal consumption in 2013.An et al.(2019) researched the combined effects of multiple sources during severe haze pollution episodes in northern China.High levels of PM 2.5,gaseous precursors, and relative humidity aggravated the generation of secondary organic aerosol and secondary inorganic aerosol.

The Interagency Monitoring of Protected Visual Environments (IMPROVE) algorithm was initiated in the United States in 1988 to identify the impact of chemical species and emission sources on visibility impairment ( Malm et al., 1994 ).In the revised IMPROVE algorithm,visibility deterioration was quantified based on the calculated light extinction coefficient b,which included light scattering and absorption by atmospheric particles and gases ( Pitchford et al., 2007 ).The calculation of b ext is important in quantifying the effects of different chemical components on light extinction ( Deng et al., 2011 ; Lin et al., 2014 ;Qu et al., 2015 ).Deng et al.(2016) found that ammonium sulfate and organic matter (OM) contributed most to light extinction in Xiamen,China, accounting for 36.4% and 30.6% of total b,respectively.In another study, meanwhile, sulfate, nitrate, and total carbon accounted for 29%, 28%, and 22% of the total bin Kaohsiung, China ( Yuan et al.,2006 ).Liu et al.(2019) compared the pollution characteristics and light extinction of three major cities and a background station in the BTH region.It was found that the bhad decreased by 51%–58% from before to after the establishment of a coal ban.Reductions in ammonium sulfate were particularly pronounced at the three cities sites, ranging from 59%–68%, indicating that the coal ban policy had improved light extinction significantly.

2.Data and methods

2.1.Sample collection and laboratory analysis

In this study, one of the most polluted cities in the BTH Region, Baoding, was selected to represent typical atmospheric pollution in the NCP.PM 2.5 component data were obtained from a field campaign carried out at Hebei University (38.8°N, 115.4°E, 16 m elevation) of Baoding.The sampling station was surrounded by university and residential areas,with no pollution sources nearby.The sampling period started from the summer of 2014 and lasted to the winter of 2017 (451 days observed),and a typical month of each season was selected to sample.Samples were collected in two periods in each period of 24 hours: from 0800 LST to 1930 LST (daytime), and from 2000 LST to 0730 LST the next day (nighttime), respectively.Chemical component analysis included eight water-soluble inorganic ions (Na,NH 4,K,Mg,Ca,Cl,NO 3,and SO 4? ), carbonaceous aerosols(elemental carbon (EC) and organic carbon (OC)) and 20 inorganic elements (Na, Mg, Al, K, Ca, V,Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Ag, Cd, Ba, Tl, and Pb).More detailed information about the experiment apparatus and laboratory operations can be found in previous studies ( Tan et al., 2017 ; Tian et al., 2014 ;Wang et al., 2015 ).

The online measured data of PMand PMmass concentrations were provided by the Ministry of Ecological Environment of China.Meteorological data, including visibility and relative humidity, were derived from the China Meteorological Administration.

2.2.Reconstruction of b ext with IMPROVE

After revision of the IMPROVE algorithm in 2007, the reconstructed fine mass (RCFM) concentration is now computed based on the following six major components:

where AS is ammonium sulfate ((NH)SO,AS = 1.375 ? [SO]) and AN is ammonium nitrate (NH 4 NO 3,AN = 1.29 ? [NO 3]).Organic matter was computed according to the organic carbon (OM = 1.6 × OC).Sea salt (SS) could be calculated based on the water-soluble ion Na(SS = 2.54 ? [Na]).Generally, fine soil mass concentration is estimated by summing the oxides of some inorganic elements (A1O,SiO,CaO,K 2 O, FeO, Fe 2 O 3,and TiO 2 ) ( Malm et al., 1994 ).Here, considering data availability, fine soil was calculated as follows:

According to Pitchford et al.(2007),light extinction bis the sum of b sca and b abs,representing the contribution of light scattering by particles and gases ( b= b+ b) and light absorption by particles and gases ( b sca = b a,p + b a,g ):

Here, coarse matter (CM) is the difference between the PM 10 and PMmass concentration (CM = PM? PM).Rayleigh scattering displays the scattering of gases ( b) and is defined as a constant value of 10 M m1 ( Liu et al., 2019 ; Shen et al., 2014 ).In this paper, a performance evaluation of the IMPROVE algorithm is given by comparing the b sca reconstructed from PM 2.5 components (R b sca ) and that converted from visibility (V b).V bwas calculated by the Koschmieder equation (Vbsca = 3.912 / visibility).Because the visibility was measured at 550 nm by a Belfort Model 6000 instrument, it needed to be converted to 880 nm via the following equation so as to fit with R b( Jung et al.,2009 ):

Here, the scattering ?ngstrom exponent α is defined as 1.18 in accordance with previous research.A detailed illustration of the above calculation method can be found in Liu et al.(2019) .

3.Results

3.1.Fine particle pollution and meteorological conditions

Fig.1 presents the concentration variation of PMcomponents, including 8 water-soluble inorganic ions (WSII), OC, EC, and the sum of 20 inorganic elements (IE).The annual average concentration of total PMin the years 2014–17 was 110 ± 72

μ

g m,72 ± 44

μ

g m,91 ± 69 μg m3,and 65 ± 41

μ

g m3,respectively.Generally one of decline, the trend was consistent with the weighing result of PMsampling membrane and the online monitored PMmass concentration.Visibility and relative humidity were negatively related.The annual average visibility was 9.4 ± 6.9 km, 10.4 ± 7.6 km, 14.5 ± 9.8 km,13.0 ± 7.6 km during the four years, respectively.The improvement of visibility has resulted from a series of actions and pollution control policies made by the central and local government;in particular, the establishment of a coal-ban area in the NCP, which has contributed significantly to PM pollution control ( Liu et al., 2019 ).As shown in Fig.1,the total PM 2.5 concentration increased slightly by 10

μ

g min 2016.However, in the summer of 2016, visibility increased greatly to 19.4 km, having been 10.7 km in 2015.This may result from the contribution of WSII in summer, which decreased by 16

μ

g m(71.4%) from 2015, 34

μ

g m(152.4%) from 2014, and was 13

μ

g m3 lower than in 2017.However, as more biomass burning took place during the heating period (from 15 September 2016 to 15 March 2017)( An et al., 2019 ), carbonaceous aerosols, NH 4,and SO 4? increased significantly.Meanwhile, in the spring seasons of 2015–17, the total PMconcentration showed a rising trend, indicating that SNA (NH,NO 3,and SO 4? ) and OC pollution were high on the agenda of atmospheric pollution control.

Fig.1.Concentration variation of PM 2.5 components during (a) daytime and (b) nighttime.(c) Meteorological conditions, including visibility and relative humidity.

3.2. Variation characteristics of PM 2.5 components

Fig.2 shows the concentrations and percentage variations of the eight water-soluble inorganic ions from different perspectives.Distinct seasonal variation was observed, with the highest average concentration of total ions in winter (62 ± 37 μg m), followed by autumn (54 ± 36 μg m3 ), spring (44 ± 27 μg m3 ), and summer(42 ± 27 μg m3 ).The annual mean mass concentration of total ions varied from 44 ± 29 μg mto 65 ± 41 μg mduring 2014–17.SNA was the dominant component ( Lin et al., 2014 ), accounting for an average of 82.8% of total ions.NHand NOoriginated from their gaseous precursors NH 3 and NO

x

,and were 6.0% and 8.3% higher during the night, respectively.Because the high relative humidity and low temperature at night better kept ammonium salt and nitrate as particulates.Resulting from SO 2,SO 4? was 10.8% lower at night, the concentration of which varied from 9 to 18 μg min different seasons.The largest contribution of SO 4? to total ions was in summer in the daytime owing to the photochemical reaction.Kand Clin PMwere higher at night than during the daytime, with the increase in Cleven reaching 41.2%, suggesting that more coal combustion and biomass burning were happening at night.

Fig.3 (a) illustrates the mass concentration and diurnal variation of carbonaceous aerosols.The annual mean OC and EC decreased by 20 μg m3 and 2 μg m3,respectively, during 2014–17.The annual nighttime concentrations of OC and EC were 21.4% and 29.3% higher than in the daytime.OC and EC concentrations were highest in winter(50 ± 36 μg m3,12 ± 9 μg m3 ), followed by autumn (14 ± 11 μg m3 ,6 ± 4 μg m), spring (10 ± 7 μg m,4 ± 2 μg m) and summer(7 ± 5 μg m3,3 ± 2 μg m3 ).The ratio of OC to EC (OC/EC) is influenced by different emission sources, the transformation of primary organic carbon to secondary organic carbon, and the elimination of pollutants ( Wang et al., 2015 ).Here, OC/EC varied from 2.2%–4.1% during the observation period.It peaked in winter and was at a minimum in summer, because domestic heating contributed to coal combustion and biomass burning primary emissions.The unfavorable diffusion conditions and stable atmospheric stratification also fostered the gaseous volatile organic compounds converting to a particulate state.These two processes elevated the total carbon concentration and the OC/EC ratio.

Fig.2.Concentration variation of WSII during 2014–17: (a) diurnal difference; (b) average concentrations and percentages of eight ions; (c, d) concentrations and percentages of eight ions in (c-1, d-1) spring, (c-2, c-3) summer, (c-3, d-3) autumn, and (c-4, d-4) winter during (c-1 to c-4) daytime and (d-1 to d-4) nighttime.

Fig.3.Mass concentration and diurnal variation of (a) carbonaceous aerosols in different seasons and (b) inorganic elements in PM 2.5 during 2014–17.Note that the data of Ba and Tl started from 2016.

Fig.4.Seasonal and diurnal changes of reconstructed PM 2.5 chemical compositions: (a) mass concentration; (b) percentage in light extinction.

Fig.5.Comparison of reconstructed b sca (R b sca ) and visibility-calculated b sca (V b sca ) in (a) annual situation, (b) spring, (c) summer, (d) autumn, and (e) winter.

In Fig.3 (b), the total concentration of inorganic elements ranked as follows: spring (5305 ng m)

>

winter (4515 ng m)

>

autumn(3775 ng m3 )

>

summer (3409 ng m3 ).Originated from natural sources, Na, Ca, K, Fe, Al and Mg were the major elements in fine particles, accounting for 84.6%–98.0% of the total mass concentration in different seasons.Inorganic elements accounted for the largest proportion (9.0%) of PMin spring, and lowest (3.8%) in winter.Frequent sandstorms in spring elevated the concentrations of Ca, Mg, Al and Fe owing to their high abundance in dust.The peak value of K in winter resulted from biomass burning.

3.3. Seasonal variation and light extinction of reconstructed PM 2.5 chemical compositions

As was shown in Fig.4 (a), noticeable differences can be seen in the seasonal variation of reconstructed PM 2.5 chemical compositions and their light extinction results.The reconstructed PMmass concentration averages were 71 ± 33 μg m3,61 ± 34 μg m3,83 ± 53 μg m3 ,and 154 ± 91 μg m3 in the four seasons, respectively, and followed the same trend as WSII and carbonaceous aerosols.The daily annual average at night was 90 ± 67 μg m3 and 98 ± 72 μg m3,respectively.AS,AN, and OM were the three dominant species, accounting for 67.9%,75.7%, 82.0%, and 84.3% in the four seasons, respectively, and 79.8%of the annual total PM 2.5 mass concentration.

The average bcalculated by RCFM was 660.3 ± 526.1 M m,and in the four seasons it was 373.8 ± 233.6 M m1,405.3 ± 300.1 M m1 ,554.3 ± 378.2 M m1,and 1005.2 ± 750.3 M m1,respectively.AS, AN, and OM were still the largest contributors to light extinction( Deng et al., 2016 ), accounting overall for 71.5%, 21.1%, 20.3%, and 30.0%, respectively.During the study period, the PM,SO,NO,and OC mass concentrations reduced by 39.3%, 42.9%, 28.6%, and 60.6%.Meanwhile, the light extinction of their corresponding species’total b,AS, AN and OM declined by 50.5%, 47.1%, 48.9%, and 64.3%.Generally, the pollution control measures aimed at improving the visibility in the NCP during 2014–17 have achieved an initial success, but further analyses and accurate controls should be performed.

3.4.Performance evaluation of the IMPROVE algorithm

Fig.5 compares V band R bto evaluate the performance of the revised IMPROVE algorithm for estimating light extinction.The bdescribes the aerosol scattering contribution of light extinction, which comprised a large part of, and could probably represent, the b ext.The color of the scatter points represents the relative humidity, and the influence of PMmass concentration is shown through the scatter sizes.The determination of PM 2.5 has considered the online measured and weighing results, to make sure it is as close as possible to the real situation.As shown in Fig.4 (a), strong correlation can be seen between V band R bin the overall situation with a correlation coefficient of 0.84.There were distinct seasonal characteristics in the light extinction distribution (see Fig.4 (b-e)).The correlation coefficient was lowest in spring (0.48), followed by summer (0.56), autumn (0.74), and winter(0.85).From Fig.4 (b, c), Vbsca was apparently overestimated by R b sca ,through comparison with the one-to-one line.In spring, the contribution of coarse particles to light extinction was elevated by frequent and typical dust weather in northern China, and the generation of secondary aerosols in summer increased the AN and AS concentrations, leading to overestimation of the actual particle scattering light extinction ( b).The V bhigh values were underestimated and the low values were overestimated in autumn, winter, and the whole year, which is consistent with another finding in the BTH region ( Liu et al., 2019 ); while in extremely polluted situations with high RH and high PM,reconstructed results were more likely to underestimate Vbsca.Particularly when R b sca> 1123 M m(corresponding to visibility < 2.0 km), V bwas underestimated by 17.6%.

4.Conclusions

PMpollution has long been a complex problem in the NCP region, involving local emissions, secondary reactions, regional transport,and meteorological conditions.In order to cast light on the impact of PMpollution on visibility degradation, this study analyzed the pollution characteristics of eight water-soluble inorganic ions, carbonaceous aerosols, and 20 kinds of inorganic elements in a case city of northern China.The b ext of reconstructed PM 2.5 chemical components based on the IMPROVE algorithm is presented to show the light extinction characteristics and variation during 2014–17.

With a series of government measures having been taken to prevent and control air pollution, effective reduction can be seen in all PMcomponents mass concentrations during the four years.However, inevitable biomass burning during the heating period caused significant increases in OC, EC, NH 4,and SO 4?.Meanwhile, slight rises in the total PMconcentration in spring also indicated that SNA and carbonaceous aerosol pollution were high on the agenda of atmospheric pollution control.

Noticeable seasonal variation was observed in the reconstructed fine mass concentration and light extinction.The light extinction reduction of AS, AN, and OM was greater than that of their corresponding chemical components, and the average total bdeclined more than the total PM 2.5 mass concentration.Meanwhile, a performance evaluation of the IMPROVE algorithm supported the credibility of the results.There were strong correlations between Vbsca and R b sca.The high values of V bwere underestimated, while the low values were overestimated.The PMmass concentration and relative humidity could cause an impact on the estimation.In particular, the underestimation aggravated as PMmass concentration increased.To conclude, pollution control for improving visibility in the NCP region has achieved an initial success,but a locally adapted IMPROVE algorithm should nonetheless be further analyzed, and more accurate control of PMchemical components is urgently need.

Declaration of Competing Interest

No potential conflict of interest was reported by the authors.

Funding

This study was supported by the National Key Research and Development Program of China [grant number 2016YFC0202001], the Chinese Academy of Sciences Strategic Priority Research Program [grant number XDA23020301], the National Natural Science Foundation of China [grant numbers 41375036 and 91644226], and the National Key Research and Development Program of China [grant number 2018YFC0214002].