Xingna Yu 1,2, Li Shen1, Sihan Xiao1, Jia Ma3, Rui Lü1, Bin Zhu1, Jianlin Hu4, Kui Chen1, Jun Zhu1

Key Laboratory of Meteorological Disaster, Ministry of Education (KLME)/Joint International Research Laboratory of Climate and Environment Change (ILCEC)/Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters (CIC-FEMD)/Key Laboratory for Aerosol-Cloud-Precipitation of China Meteorological Administration, Nanjing University of Information Science and Technology, Nanjing 210044, China
Shanghai Key Laboratory of Atmospheric Particle Pollution and Prevention (LAP3), Shanghai 200433, China
Guangzhou Hexin Analytical Instrument Company Limited, Guangzhou 510530, China
Collaborative Innovation Center of Atmospheric Environment and Equipment Technology, Jiangsu Key Laboratory of Atmospheric Environment Monitoring and Pollution Control, School of Environmental Science and Engineering, Nanjing University of Information Science and Technology, Nanjing 210044, China


Received: December 29, 2017
Revised: April 19, 2018
Accepted: April 19, 2018

Download Citation: ||https://doi.org/10.4209/aaqr.2017.12.0572  

  • Download: PDF


Cite this article:

Yu, X., Shen, L., Xiao, S., Ma, J., Lü, R., Zhu, B., Hu, J., Chen, K. and Zhu, J. (2019). Chemical and Optical Properties of Atmospheric Aerosols during the Polluted Periods in a Megacity in the Yangtze River Delta, China. Aerosol Air Qual. Res. 19: 103-117. https://doi.org/10.4209/aaqr.2017.12.0572


HIGHLIGHTS

  • Aerosol chemical and optical properties were observed in Nanjing.
  • Secondary formation dominated PM2.5 pollution were discussed.
  • Contributions of aerosol chemical components to light extinction were quantified.
 

ABSTRACT


The chemical composition and optical properties of particulate matter (PM) were characterized in an urban-industrial area of Nanjing, China, in January 2015, when heavily polluted periods frequently occurred. Only 30% of the days fulfilled the National Ambient Air Quality Standards of China. The average scattering and absorption coefficients at 532 nm during the polluted periods were 620 ± 320 Mm–1 and 102 ± 57 Mm–1. An increasing relative fraction of the large size PM during the polluted periods can be deduced from the variations of the scattering Ångström exponent, backscattering ratio, and asymmetry factor. The mean mass concentrations of NO3, SO42– and NH4+ in PM2.5 during the polluted periods were 32.87 ± 17.76 µg m–3, 23.6 ± 13.2 µg m–3, and 19.4 ± 10.1 µg m–3, respectively. NO3, SO42– and NH4+ were the dominant water-soluble ions (WSIs) and accounted for 87% of the total ion concentration. Nitrate and organic matter (OM) dominated the aerosol composition during the polluted periods. The averaged organic carbon/elementary carbon ratios during the polluted and the cleaner periods were 3.6 and 4.3, respectively, consistent with a mix of primary emissions and secondary organic aerosol formation. Organic matter and ammonium nitrate (AN) were the dominant species contributing to light extinction during the polluted periods, contributing values of 159 ± 63 Mm–1 and 156 ± 91 Mm–1, respectively.


Keywords: Air pollution; Aerosol chemical composition; Aerosol optical properties; Atmospheric extinction.


INTRODUCTION


Atmospheric aerosols play a significant role in radiation balance and radiative forcing by directly interacting with solar radiation and by indirectly influencing the formation, optical properties and lifetime of clouds by acting as cloud condensation nuclei (Li et al., 2011; Ramanathan et al., 2001). In addition, elevated atmospheric aerosols cause visibility impairment and adverse health effects (Zhou et al., 2014; Yu et al., 2016a). However, quantification of these effects still contains large uncertainties, mainly due to poor understanding of aerosol physicochemical properties and their heterogeneous temporal and spatial variation.

The Yangtze River Delta (YRD) is regarded as one of the major haze regions in China, and has been experiencing high concentrations of fine particulate matter (PM2.5) and low visual ranges (Che et al., 2007; Gao et al., 2011; Shen et al., 2015; Zhang et al., 2017). Nanjing, the provincial capital of Jiangsu Province with over 8 million inhabitants, has been suffering heavy haze pollution which is caused by large emissions from complex sources including automobiles, coal-fired power plants, electronics, petrochemical and steel industries, etc. The concentration levels of PM2.5and the numbers of haze days have been increasing. For example, the number of haze days in Nanjing was over 130 days during 2001–2006 (Dong et al., 2007; Qian et al., 2008), and increased to more than 161 days between 2008 and 2010 (Lin et al., 2012). The average mass concentration of PM2.5 reached 113 ± 69 µg m–3 and average visibility was 4.8 ± 3.2 km during haze pollution in the winter of 2014 (Kong et al., 2015). In recent years, many studies have been carried out in this area mainly focusing on aerosol chemical composition, source apportionment, new particle formation and optical properties based on long-term or short-term ground-based or satellite remote sensing data (Duan et al., 2006; Shen et al., 2009; Tan et al., 2009; An et al., 2015; Wang et al., 2015; Xu et al., 2015; Qi et al., 2016; Yu et al., 2016; Wang et al., 2016b; Du et al., 2017). However, the comprehensive understanding of aerosol optical and chemical properties during high pollution days remains unknown for this area. Specifically, a quantitative relationship between visibility impairment and aerosol chemical components, which are very valuable for designing effective pollutant control policies to improve impaired visibility condition, is far from complete. In this study, we measured the water-soluble ions (WSIs) and carbonaceous compounds in PM2.5 during polluted and cleaner periods in a winter season. We also analyzed the aerosol scattering andabsorption coefficients, and derived backscattering ratio, asymmetry factor, scattering Ångström exponent and mass scattering efficiency. Finally, we examined the relative contribution of chemical components in PM2.5 to visibility degradation using the IMPROVE (Interagency Monitoring of Protected Visual Environments) algorithm (Malm et al., 1994).


INSTRUMENTATION AND METHODOLOGY



Site and Sampling

All the surface observations of aerosol chemical and optical properties presented in this study were conducted at an urban-industrial site of Nanjing in Luhe District, which is about 5 km from the observation site located in the north of Nanjing, and is also an important production base of the steel industry. The pollution sources in this area are mainly emissions from industries, traffic and urban construction activities (Wang et al., 2014). The measurements were performed on the roof of the Meteorology Building about 30 m above ground level on the Nanjing University of Information Science and Technology (NUIST) campus (lat: 32.2°N, long: 118.7°E). The meteorological data were recorded by different weather sensors mounted on the automatic weather station (AWS) and the visibility data were obtained from a forward scattering visibility meter (Model: CJY-1A) installed in the campus, which is ~800 m from the experimental site. The pollution sources surrounding the observation site include the Nanjing chemical industry factories situated ~3 km to the southwest, iron and steel plants within 2 km vicinity of site to the northwest and main traffic roads within 500 m (see Fig. 1).


Fig. 1. Geographical location of Nanjing (NJ) over China represented with a solid star symbol along with its bordering countries and oceans. Also shown is the location of NJ urban-industrial area indicated with a triangle in the Yangtze River Delta (YRD) with its major sources in the vicinity of the sampling site.Fig. 1. 
Geographical location of Nanjing (NJ) over China represented with a solid star symbol along with its bordering countries and oceans. Also shown is the location of NJ urban-industrial area indicated with a triangle in the Yangtze River Delta (YRD) with its major sources in the vicinity of the sampling site.

Sampling was normally conducted once every 6 hours starting at 3:00 AM Beijing Time. PM2.5 samples were collected at a flow rate of 100 Lmin–1 with a KC-120H QingDaoLaoshan sampler (Laoshan Electronic Instrument Factory Co., LTD., QingDao, China). Samples were collected on 90 mm Whatman quartz fiber filters (QM-A™, Whatman, Clifton, NJ, USA) from 8 January to 31 January 2015. The filters were pre-baked by heating at 800°C for 5 h to remove residual carbon. After sampling, the filters were transported in a portable cooler to the aerosol laboratory to minimize loss of volatile compounds. Field blank filters were also collected periodically by exposing filters in the sampler without drawing air through them; these were used to account for any artifacts introduced during the sample handling process.

Measurements of half-hourly PM1 and PM2.5 were conducted using two Thermo Fisher Scientific FH62C14 beta attenuation instruments equipped with 1.0 µm and 2.5 µm cut-points, respectively. 


Chemical and Optical Analysis


Organic Carbon (OC) and Elemental Carbon (EC) Analysis

A 0.5 cm2 punch from each quartz filter was analyzed for OC and EC by using a Desert Research Institute Model 2001 thermal/optical carbon analyzer (Atmoslytic Inc., Calabasas, CA, USA). Detailed information of the method, including quality assurance/quality control (QA/QC) procedures, is described by Wang et al. (2014).


Water-
Soluble Inorganic Ions Analysis

A quarter portion of filter was used to determine the WSIs. WSIs were analyzed using a Model 850 professional Ion Chromatography (IC) system (Metrohm, Switzerland). The chromatography system includes a column oven, conductivity detector, a model 858 auto-injector and MagIC Net chromatography workstation (Metrohm, Switzerland); Column; Metrosep C 4150/4.0 separation column and Metrosep A Supp 5150/4.0 separation column; eluent: 3.2 mmol L–1 Na2CO3 + 1.0 mmol L–1 NaHCO3 (anions), 1.7 mmol L–1 HNO3 + 0.7 mmol L–1 pyridine carboxylic acid (cations); column temperature: 30°C; flow-rate: 1.0 mL min–1; injection volume: 20 µL. Detailed information is available from Li et al. (2014).


Scattering and Absorption Coefficients

The total scattering (σsp; 7–170° angular integration) and hemispheric back scattering (σbsp; 90–170°) coefficients at the wavelengths of 450, 550 and 700 nm were measured with an Integrating Nephelometer (Model 3563, TSI, USA). An inverted funnel with screws was fitted at the entrance of the instrument to avoid dust, rainwater, and insects entering into the system. This instrument draws the ambient air through stainless steel tubing at a fixed flow rate of 20 L min1 without aerosol size cut-off. The samples were illuminated with a halogen lamp and measures scattered light at the above three wavelengths using three photomultiplier tubes. Calibration of the nephelometer was carried out before the experiments with filtered air as a low span gas and carbon dioxide (CO2, purity: 99.99%) as the high span gas. The total scattering and backscattering coefficients were corrected on a systematic basis considering the angular truncation correction method proposed by Anderson and Ogren (Anderson et al., 1998) and Sherman et al. (2015). The instrument calibration, error estimation and uncertainty have been presented elsewhere (Fan et al., 2010; Esteve et al., 2012; Jing et al., 2015).

The scattering coefficient is greatly affected by relative humidity (RH), which shows a negligible variation when RH < 50% and a steep rise when RH > 80% (Anderson et al., 1998). In this study, sample air enters the inlet through a protective cover to eliminate rain and insects then passes through a cyclone. After passing through the cyclone, the sample enters the top section of the inlet where its humidity is reduced with Perma Pure tubes. The Perma Pure diffusion drying tube, used to control sample RH, consists of an inner tube of Nafion through which the aerosol sample flows and an outer tube of stainless steel through which purge air flows. The inlet was always operated dry (RH < 30%) with nephelometer measuring the dry scattering coefficient.

Aerosol absorption coefficient was measured using a 7-channel aethalometer (Model AE-33, Magee Scientific, USA) at wavelengths of 370, 470, 520, 590, 660, 880, and 950 nm with a PM2.5 size cut. This instrument yielded a change in optical attenuation by measuring the intensity of the light beam passing through a filter tape. The air inlet is located ~2 m above the roof. Routine flow calibration and blank sample test were implemented before sampling. The aethalometer was operated at a flow rate of 5.0 L min1 in an automated mode and a sampling interval of 1 min, under which the filter tape was forward when the attenuation at wavelength of 370 nm reached 75.


Data Treatment


Calculation of Aerosol Absorption Coefficient (σabs) and the Aethalometer Model

Coen et al. (2010) and Weingartner et al. (2003) correction algorithms were selected for this study and the σabs has been estimated using the aethalometer data at all wavelengths.

where λ is the wavelength (in m), C and R are correction factors for minimizing the inherent uncertainty associated with the aethalometer, resulting from multiple scattering of light in the filter matrix and the change in the optical path length due to successive aerosol loadings. The corrections for these uncertainties were completed following the results reported in Drinovec et al. (2015) by incorporating values of R = 1 and C =1.57 for the multiple scattering corrections and loading effect, respectively.

The aethalometer model is designed to quantify the contribution of fossil fuel (BCff) and biomass burning (BCbb) to the total BC concentration. It is based on the assumptions that the fossil fuel and biomass burning are the primary contributors of carbonaceous aerosols and the light absorption of aerosol by these two sources can be modelled (Herich et al., 2011). The dependence of aerosol light absorption on wavelength is parameterized using a power law relationship: 

where σabs is the spectrally dependent mass absorption coefficient, λ the light wavelength, K is a constant and αap is the absorption Ångström exponent. Higher αap values were observed when aerosols are originating from biomass burning, as the spectral absorption of these species increases more rapidly at shorter wavelengths (Sandradewi et al., 2008). According to the study by Kirchstetter et al. (2004), αff = 1 and αbb = 2 are used in this study. αff and αbb are absorption Ångström exponents for fossil fuel and biomass burning emissions, respectively. The values for αff and αbbare just estimates. σabs for the two component aerosol system can be represented as:

σabs(λ)ff and σabs(λ)bb are the aerosol absorption coefficient of BC from fossil fuel and biomass burning emissions at wavelength λ, respectively.

Then, BCff and BCbb can be calculated by Eqs. (3)–(7):


Calculation of Scattering Ångström Exponent (α)

The nephelometer has three working wavelengths, but none of them matches the 532 nm working wavelength of the AE33. So, we selected the scattering coefficients measured at 550 nm (σsp (550 nm)) and converted them to 532 nm (σsp (532 nm)) to match with the σap (532 nm) following the method adopted by Jung et al. (2010): 

where ‘α’ is the scattering Ångström exponent which is determined by the formula given below.


Calculation of
 Mass Scattering Efficiency (MSE)

The mass scattering efficiency can be estimated by dividing the scattering coefficient by aerosol mass concentration. In this study, MSE was estimated as the ratio of σsp (550 nm) to PM2.5 mass concentration. 


Calculation of Backscattering Ratio and Asymmetry Parameter (g
λ)

The backscattering ratio (bλ) is the ratio of the hemispheric backscattering coefficient to the total scattering coefficient at a given wavelength. 

The asymmetry parameter (gλ) is calculated from the backscattering ratio. 

where gλ and bλ obtained at the same wavelength. The above equation was suggested by Andrews et al. (2006) based on the plot of Wiscombe and Grams (1976) following the Henyey–Greenstein approximation for the asymmetry parameter. gλ measures the preferred scattering direction (forward or backward) for the light encountering the aerosol particles. 


Estimation of Chemical Extinction Coefficient (bext)

Visibility degradation occurs as a result of the scattering and absorption of light by particles and gases in the atmosphere. bext is estimated according to the IMPROVE program formula as described by Malm et al. (1994)


where the unit of bext is Mm1; [X] represents the individual PM2.5 species concentrations in µg m3. f(RH) is the humidification factor accountingfor the impact of relative humidity on the growth of the hygroscopic aerosol.

The concentrations of ammonium sulfate and ammonium nitrate are calculated by multiplying [SO42] and [NO3] by factors of 1.38 and 1.29, respectively, to account for paired ammonium ions. Organic matter is estimated from 1.6 × [OC], this multiplier is suitable for urban aerosol (Turpin et al., 2001; Ye et al., 2017). The extinction effect of fine particulate matter on visibility impairment was studied, so we excluded the contributions of coarse particles and soil in bext estimation since they only accounted for a small fraction of PM2.5 mass (Malm et al., 1994; Wang et al., 2003). So, the reconstructed equation used in this research is as follows: 



RESULTS AND DISCUSSION



Meteorological Parameters and Aerosol Mass Concentration

The new NAAQS of China set PM2.5 concentration limits for the 24-hour average of 75 µg m–3 for Grade II zones (MEP, 2012). The 24-hour average PM2.5 measured in this study in Nanjing ranged from 40.5 µg m–3 to 241.4 µg m–3 with an average of 104.0 µg m–3, severely exceeding the NAAQS of China and other standards recommended by the WHO (25 µg m–3) and the United States (35 µg m–3) (MEP, 2012; US EPA, 2013; WHO, 2006). According to the new NAAQS (Grade II), about 70% of days exceeded the standard during the measurement period. Fig. 2 displays the mass concentrations of PM1, PM1-2.5 and mass concentration ratios of PM1/PM2.5 in January 2015. The polluted period in this study is defined as the 24-hour average PM2.5 exceeds 75 µg m–3. For example, the averaged PM2.5 mass concentration during the polluted periods reached 124.95 ± 57.84 µg m3, which was 2.3 times higher than that during the clean periods. The PM1 and PM2.5 hourly concentrations simultaneously reached peaks of 202.4 and 370.5 µg m3, respectively, at 8:00 PM on 24 January. Then the mass concentrations decreased rapidly due to the occurrence of precipitation. The averaged mass ratios of PM1/PM2.5 during the polluted and clean periods were 0.62 ± 0.14 and 0.66 ± 0.12. The meteorological parameters during the measurement period are presented in Fig. 3. The averaged wind speed (WS) was 1.57 m s1 and the wind was mainly northeasterly and northwesterly during this period. Average RH and visibility during the polluted periods were 63.31% and 4.1 km; the lowest visibility of 0.94 km occurred in the evening of 25 January.


Fig. 2. Time series of (a) mass concentrations of PM1 and PM1-2.5; (b) mass ratios of PM1/PM2.5. The clean periods with low aerosol mass concentration are marked as gray shaded areas.Fig. 2. Time series of (a) mass concentrations of PM1 and PM1-2.5; (b) mass ratios of PM1/PM2.5. The clean periods with low aerosol mass concentration are marked as gray shaded areas.

Fig. 3. Time series of meteorological factors: temperature, relative humidity, visibility, wind vector and rainfall during the measurement period. The clean periods are marked as gray shaded areas.Fig. 3. Time series of meteorological factors: temperature, relative humidity, visibility, wind vector and rainfall during the measurement period. The clean periods are marked as gray shaded areas.


Aerosol Optical Properties

Aerosol optical parameters were obtained from 22 to 31 January which covered the polluted periods and clean periods before and after.Measurements of aerosol optical parameters on other days were not available because of instrument failure. Time series of hourly average aerosol optical parameters measured at dry RH are shown in Fig. 4: total scattering coefficients, backscattering ratio, asymmetry parameter at three wavelengths (450, 550 and 700 nm), absorption coefficient at 532 nm, scattering Ångström exponent at 700–450 nm and mass scattering efficiencies. The scattering coefficient (σsp) decreased with increasing wavelength, while backscattering ratio increased with wavelength due to decreasing size parameters. They also showed an inverse variation trend, for example, high value of σsp occurred during 24–26 January corresponding with low bλ. The averaged σsp at 550 nm during the polluted periods was 521 ± 271 Mm1, which was almost 4.0 times higher than that observed during the clean periods. However, the average value for bλ during the polluted periods (0.096 ± 0.008 at 550 nm) was slightly lower than that of clean periods (0.107 ± 0.013), and those measured in Shijiazhang (0.175), Lanzhou (0.158), Shouxian (0.101) and a rural site of the Pearl River Delta (0.124), China (Zhang et al., 2004; Garland et al., 2008; Fan et al., 2010; Zhang et al., 2012). According to the calculation based on Mie theory, bλ will be larger than 0.10 if the particles’ diameters are under about 1.5 µm (Zhang et al., 2004). This result indicated that the content of fine particles with diameters less than 1.5 µm was relatively high during the clean periods in Nanjing.


Fig. 4. Time series of (a) scattering coefficients, (b) absorption coefficients, (c) backscattering ratios, (d) scattering Ångström exponent (450–700 nm), (e) asymmetry parameter at wavelengths of 450 nm, 550 nm and 700 nm, and (f) mass scattering efficiencies. The clean periods are marked as gray areas.Fig. 4. 
Time series of (a) scattering coefficients, (b) absorption coefficients, (c) backscattering ratios, (d) scattering Ångström exponent (450–700 nm), (e) asymmetry parameter at wavelengths of 450 nm, 550 nm and 700 nm, and (f) mass scattering efficiencies. The clean periods are marked as gray areas.

The scattering Ångström exponent (α) represents the wavelength dependence of scattering coefficient and is related to the slope of the number-size distribution or the mean size and relative concentrations of the accumulation- and coarse-mode aerosol. The value of α varied from 0.31 to 1.67, suggesting a mixture of fine- and coarse-mode particles during these pollution days. Compared to values during the clean periods, a lower value of α = 0.99 during the polluted periods indicated that the relatively large size particles were present, due to coagulation and hygroscopic growth at relatively high RH conditions. It is worth noting that the value of α was the lowest on 25 and 26 January, which is consistent with the pattern of PM1/PM2.5. The mean value of α during the polluted periods was smaller than 1.51 in a rural area near Guangzhou (Garland et al., 2008) and 1.1 measured in Beijing (Wang et al., 2015), but much higher than the 0.16 reported for dust storm days in Zhangye (Tian et al., 2010), which indicated a more dominant coarse-mode particle compared with the other locations. The MSE during the polluted periods in Nanjing varied from 1.40 m2 g1 to 5.36 m2 g1 with an average of 3.35 m2 g1. The averaged value was much lower than that observed during June 2008–May 2009 at Beijing with a value of 5.88 m2 g1, but larger than that observed in Taiwan with a range of 0.7–1.6 m2 g1 and the clean periods in Nanjing with an average of 3.10 m2 g1 (Chang et al., 2006; Zhao et al., 2011).

The absorption coefficients (σap) showed a similar variation trend with scattering coefficients. The maximum daily average σsp and σap at 532 nm occurred on 24 January with value of 1076 ± 246 Mm1 and 185 ± 33 Mm1 respectively. The low WS and high RH on these days encouragedthe hygroscopic growth and accumulation of aerosol particles. The averaged σap during the polluted periods (102 ± 58 Mm1) was about 3.5 times higher than those during the clean periods, indicating less absorbing aerosols in Nanjing. The value during the polluted periods in Nanjing was much greater than that observed in Lhasa (Zhu et al., 2017), but it was comparable with the result of agricultural field burning event in Shouxian (Fan et al., 2010). The average values of gλ during the polluted periods at 450, 550 and 700 nm were 0.67 ± 0.01, 0.68 ± 0.02 and 0.64 ± 0.03 respectively, which were slightly higher than those during the clean periods and observed during an intense haze episode in winter of Beijing (Wang et al., 2015). High value of gλ was possibly attributed to the higher contribution of large size particles, and indicated the particles were predominant in forward scattering.


Aerosol Chemical Properties

Time series of aerosol absorption coefficients at 370 nm, 880 nm and BCff and BCbb concentrations with 1-h time resolution are displayed in Fig. 5. Higher absorption coefficients were observed at 370 nm, with an average value of 89 Mm1, while the mean coefficient at 880 nm was about 26 Mm1 for the duration. This is because a considerable number of light absorbing organic carbon or brown-carbon significantly absorb light towards shorter wavelengths. The mean concentration of BCff and BCbb were 1.94 µg m3 and 0.57 µg m3 respectively during the whole period, indicated that the fuel fossil burning was main emission source of BC in the north of Nanjing. The highest concentrations of BCff and BCbb occurred from 24 to 25 January during the polluted episode with average values about 3.89 µg m3 and 1.25 µg m3 respectively, accounting for 75.5% and 24.5% of the total BC concentration respectively. The contribution of biomass burning to BC were slightly higher during the polluted days, suggested that biomass burning emission also has certain contribution to the heavy pollution formation.


Fig. 5. Time series of hourly mean absorption coefficient and the concentrations of BCff and BCbb.Fig. 5. 
Time series of hourly mean absorption coefficient and the concentrations of BCff and BCbb.

Temporal variations of concentrations of WSIs and carbonaceous compounds in PM2.5 in January 2015 are presented in Fig. 6. The 6-hour sum of WSIs concentrations ranged from 15.4 to 217.8 µg m3 with a monthly mean of 71.0 µg m3, accounting for 49.5 ± 20.7% of the total PM2.5 mass concentration (filter sampling). The highest 6-hour concentration of WSIs occurred on 24 January and reached 181.6 µg m3. In general, NO3,SO42 and NH4+ were the dominant WSIs and accounted for 85.7% of the total mass concentrations of measured ions. Tian et al. (2015) also reported that the secondary ions (NO3, SO42 and NH4+) dominated in PM2.5 and accounted for 86% of the total mass of water-soluble ions. The water-soluble ion average mass concentrations ranked in the order of NO3 > SO42 > NH4+ > Cl > K+ > NO2 > Ca2+ > Na+ > F > Mg2+. This order was consistent with that observed during the polluted periods, but it changed slightly during the clean periods. On average, the NO3 was the most abundant ion in January 2015 in Nanjing with a monthly mean concentration of 26.2 ± 18.6 µg m3, accounting for 35.8 ± 24.6% of the totalmeasured WSIs. Ions of SO42 and NH4+ were the other two major species, accounting for 28.6 ± 17.2% and 22.5 ± 13.5% of the total WSIsrespectively. Wei et al. (2017) also reported that the higher percentage of secondary species to PM2.5 occurred in Nanjing, and indicated that the aerosol was mainly caused by the secondary transformation of local air pollutants. The high concentration of secondary inorganic species implied the need of further control measures for their precursor gases, SO2, NOx and NH3, emitted from fossil fuel combustion, industrial emissions and agriculture sources over the local and surrounding area of Nanjing. The monthly concentration of Cl, K+, NO2, Ca2+, Na+, F and Mg2+ ranged from 0.1 to 5.9 µg m3, accounting for 13% of the total ion concentration.


Fig. 6. Temporal variations of (a) water-soluble ions and (b) organic carbon and elemental carbon in PM2.5 in January 2015. The clean periods with low mass concentration of water-soluble ions and carbonaceous aerosols are marked as gray shaded areas.Fig. 6. 
Temporal variations of (a) water-soluble ions and (b) organic carbon and elemental carbon in PM2.5 in January 2015. The clean periods with low mass concentration of water-soluble ions and carbonaceous aerosols are marked as gray shaded areas.

The 6-hour concentration of OC during the study period ranged from 4.4 to 63.2 µg m3 with a monthly mean of 21.0 µg m3, while the concentration of EC varied between 0.7 µg m3 and 14.3 µg m3 with a monthly mean of 5.7 µg m3. The monthly mean of OC and EC in winter of Nanjing was higher than that observed in summer of Beijing (Tian et al., 2015), but lower than that measured in summer of Xi’an (Cao et al., 2007). An average concentration of OC achieved 44 µg m3 during November 1999 in Linan, and OC accounted for ~50% of the PM2.5 mass during the field study (Xu et al., 2002).

Table 1 summarizes aerosol physical and chemical characteristics during the study period. The lower concentration of aerosol chemical species during the clean period was mainly due to precipitation. The averaged mass concentrations of OC and EC in PM2.5 during the polluted periodsreached 24.7 ± 9.8 µg m3 and 6.8 ± 2.9 µg m3. These values were 2.3 and 2.7 times higher than those during the clean periods (i.e., 10.7 ± 3.6 and 2.5 ± 0.9 µg m3 for OC and EC, respectively). Higher concentrations during the polluted periods are expected due to emissions from fossil fuel and biomass burning, secondary formation from gas-to-particle conversion processes and stable meteorological conditions. The measured ion and carbonaceous species accounted for 52% and 19% of the total PM2.5 for the polluted periods, and those of 36% and 13% for the clean periods, respectively.


Table.1 Aerosol physical and chemical characteristics during the study period in Najing. unit: PM and chemical species in µg m–3.

The average mass concentrations of NO3, SO42 and NH4+ during the polluted periods increased approximately 2–3 times compared to theclean periods, which might result from the accelerated formation of secondary aerosols under the humid conditions that existed during the polluted period. The NO3/SO42 average ratios were 0.64 and 1.39 during the clean and polluted periods, respectively. The higher ratio during the polluted periods indicated the important role that NOx and NH3 emissions play in contributing to severe haze episodes during stagnation periods. Low temperature and high humidity in winter favor a shift from the gas phase as nitric acid (HNO3) and NH3 to the particle phase as ammonium nitrate (NH4NO3) (US EPA, 1999). Ca, which can be regarded as a tracer for soil source, was 3.4 times higher during the polluted periods than during the clean periods. This suggests that dust emitted from roads and construction sites had a significant contribution to the pollution episode, even for PM2.5.

It is interesting to note that the concentration of PM2.5 remained high from 24 to 27 January and the daily mean value ranging between 76 µg m3 and 241 µg m3. Nevertheless, the ratio of PM1/PM2.5 decreased sharply started from 25 January, and a lower mass ratio of PM1/PM2.5 occurred from noon of 25 January to midnight 26 January with an average of 0.31 (see Fig. 2(b)). Compared with aerosol chemical and optical properties on 24 January, the concentrations of WSIs, scattering and absorption coefficients were all decreased on 25 January. The time discrepancies in aerosol chemical compositions and the PM1/PM2.5 ratio can be explained by the transformation of meteorological conditions. The wind direction was north to northeast on 24 January but veered to northwest and became stronger on 25 January. Besides, there were several rainfalls started from 25 January.


Contributions of Chemical Species to Light Extinction

Formula (11) assumes that SO42– and NO3 were completely neutralized by NH4+. Examination of a scatterplot of [NH4+] versus ([NO3] + 2[SO42–]) (in moles) indicated that this assumption is reasonable for the study site; measured NH4+ was sufficient to neutralize NO3 and SO42–during the observation period in Nanjing (Fig. 7).


Fig. 7. Scatter plot of [NH4+] versus [NO3–] + 2 × [SO42–].
Fig. 7. 
Scatter plot of [NH4+] versus [NO3] + 2 × [SO42].

Both calculated and estimated aerosol extinction coefficients are shown in Fig. 8(a) and their relationship is linearly fitted in Fig. 8(b). The hygroscopic growth of aerosol was estimated according to different values of f(RH), which were selected from Malm and Day (2001) and Yan et al.(2009). The average estimated chemical bext based on the values of f(RH)Malm were 920 Mm1 during 22–27 January (dense pollution episodes).However, this value decreased to 708 Mm1 when values of f(RH) were selected from studies conducted at a Chinese urban site in Beijing (f(RH)Yan) (Yan et al., 2009). As shown in Fig. 8(b), the estimated chemical bext correlated strongly with the calculated bext. Nevertheless, at high bext range (such as polluted periods), the discrepancies between calculated bext and estimated chemical bext became larger (see Fig. 8(a)), sometimes exceeding and sometimes falling below calculated values. The discrepancies could be associated with uncertainties in f(RH). It is interesting to see the light extinction was largely overestimated using f(RH)Malm on the days of 25 January 2015 and 26 January 2015, which was related with decreasedsratio of PM1.0/PM2.5. The particulates grew to large size and mass scattering efficiencies for sulfate and nitrate in IMPROVE formula was no longer suitable to estimate light extinction. The estimated bext based on f(RH)Yan significantly correlated with the calculated values than those based on f(RH)Malm with a regression slope of 0.75 and r = 0.90. This is attributed to more similar pollution aerosol types in Nanjing and Beijing vs. the Malm et al. f(RH) values which are derived from measurements in the United States. Differences between calculated extinction and reconstructed values derived from the IMPROVE equation might also be expected if aerosol size distributions differ between Nanjing and U.S. aerosols upon which the IMPROVE formula is based.


Fig. 8. (a) Temporal distribution of aerosol extinction coefficients according to the estimation of revised IMPROVE equation (f(RH) are from Malm and Day (2010) and Yan et al. (2009)) and calculation by instruments. (b) Comparison between the calculated and estimated extinction coefficients.Fig. 8. 
(a) Temporal distribution of aerosol extinction coefficients according to the estimation of revised IMPROVE equation (f(RH) are from Malm and Day (2010) and Yan et al. (2009)) and calculation by instruments. (b) Comparison between the calculated and estimated extinction coefficients.

Based on the estimation of f(RH)Yan, an average reconstructed bext was 370 Mm1 during the total measurement period. This value was greater than that measured in Guangzhou (346 Mm1) (Zhang et al., 2010), but lower than has been observed in Beijing (504 Mm1) (Huang et al., 2015), Chengdu (900 Mm1) (Wang et al., 2013), Xi’an (912 Mm1) (Cao et al., 2012) and Delhi (644 Mm1) (Singh et al., 2008). The total average light extinction contributions of aerosol components during polluted and clean periods in Nanjing are presented in Fig. 9. The apportionment contributions from NH4NO3, OM, (NH4)2SO4 and EC during polluted periods were 30.8%, 31.5%, 23.9% and 13.8% respectively, indicating the dominant contributions of nitrate and organic matter to light extinction during winter polluted periods in Nanjing. Organic matter was the largest contributor to visibility degradation during the polluted periods, while sulfate, which is expected to have more regional source contributions, was the largest contributor for clean periods.


Fig. 9. Relative contributions of aerosol chemical components to light extinction coefficient during (a) polluted periods and (b) clean periods in January 2015 in Nanjing.Fig. 9. 
Relative contributions of aerosol chemical components to light extinction coefficient during (a) polluted periods and (b) clean periods in January 2015 in Nanjing.


CONCLUSIONS


Based on in situ measurements, the physical, chemical, and optical properties of aerosols were determined in January 2015 at an urban-industrial site in Nanjing, China. The polluted periods showed a significant decrease in visibility and an increase in particle mass loading. The average mass concentrations of PM1 and PM2.5 during the polluted periods were 74.3 ± 32.1 µg m3 and 125.0 ± 57.8 µg m3, respectively. The average scattering and absorption coefficients at 532 nm during the polluted periods were 620 ± 320 Mm1 and 102 ± 58 Mm1—about 3.5 times higher than those during the clean periods. Higher values of the asymmetry factor and lower values of the backscattering ratio and scattering Ångström exponent indicated an increasing fraction of relatively large size particles during the polluted periods, which may reflect local dust generation, aerosol hygroscopic growth during the high humidity air stagnation periods, and/or coagulation and condensational growth of aerosols under extremely polluted conditions.

The average mass concentration of the dominant water-soluble ions (SO42, NO3 and NH4+) in PM2.5 was about 2–3 times higher during the polluted periods than during the clean periods following precipitation. Similar to the ions, the average mass concentrations of OC and EC in PM2.5 during the polluted periods were 2.3 and 2.7 times higher than those during clean periods. The average OC/EC ratio during the polluted period was 3.6, consistent with a mix likely of primary emissions and secondary organic aerosol formation.

The estimated average chemical extinction coefficient based on the revised IMPROVE equation was 506.5 Mm1 during the polluted periods—much greater than that observed in many areas of the world. Light extinction apportionment showed that ammonium nitrate, organic matter, ammonium sulfate, and elemental carbon contributed fractions of 30.8%, 31.5%, 23.9%, and 13.8% respectively. The dominant contribution of organic matter and nitrate to extinction during haze episodes, in contrast to sulfate being the largest contributor during cleaner periods, points to the importance of local sources of organic components, NOx, and NH3 to fine particle formation in winter stagnation episodes.


ACKNOWLEDGMENTS


 This work was supported by the National Key Research and Development Program of China (2016YFC0203500, Task #1), the National Natural Science Foundation of China (Grant Nos. 91544229, 41475142 and 41575132), the Natural Science Foundation of Jiangsu Province of China(BK20170943), the Opening Project of Shanghai Key Laboratory of Atmospheric Particle Pollution and Prevention (LAP3) (Grant No. FDLAP16004), Natural Science Foundation of Guangdong Province of China-Major Basic Research and Cultivation Projects (2015A030308014), Qing Lan Project and the Priority Academic Program Development (PAPD) of Jiangsu Higher Education Institutions. We acknowledge work team of the China Meteorological Administration (CMA) meteorology observational field for maintenance in meteorological observation instrument.



Don't forget to share this article 

 

Subscribe to our Newsletter 

Aerosol and Air Quality Research has published over 2,000 peer-reviewed articles. Enter your email address to receive latest updates and research articles to your inbox every second week.

Latest coronavirus research from Aerosol and Air Quality Research

2018 Impact Factor: 2.735

5-Year Impact Factor: 2.827


SCImago Journal & Country Rank

Aerosol and Air Quality Research (AAQR) is an independently-run non-profit journal, promotes submissions of high-quality research, and strives to be one of the leading aerosol and air quality open-access journals in the world.