Source apportionment of PM10 by using positive matrix factorization at an urban site of Delhi, India

Urban Climate xxx (2013) xxx–xxx

Contents lists available at ScienceDirect

Urban Climate journal homepage: www.elsevier.com/locate/uclim

Source apportionment of PM10 by using positive matrix factorization at an urban site of Delhi, India S.K. Sharma ⇑, T.K. Mandal, Mohit Saxena, Rashmi, Rohtash, A. Sharma, R. Gautam CSIR-National Physical Laboratory, Dr. K. S. Krishnan Road, New Delhi 110 012, India

a r t i c l e

i n f o

Article history: Available online xxxx Keywords: PM10 Organic carbon Elemental carbon Positive matrix factorization Enrichment factor analysis

a b s t r a c t In the present study, source apportionment of PM10 has been done using positive matrix factorization at an urban site of Delhi, India based on the chemical compositions of PM10 collected during January 2010 to December 2011. The concentration of PM10 and its chemical components including organic carbon (OC), elemental carbon (EC), water soluble inorganic ionic components (WSIC) and major and trace elements showed strong seasonal cycle with maxima during winter (PM10: 241.4 ± 50.5 lg m3; OC: 34.7 ± 10.2 lg m3; EC: 10.9 ± 3.0 lg m3) and minima during monsoon (PM10: 140.1 ± 43.9 lg m3; OC: 15.5 ± 7.5 lg m3; EC: 4.9 ± 2.3 lg m3). In this process, chemical composition of the PM10 mass was reconstructed using IMPROVE equation from the observed elemental composition. The highest contribution comes from particulate organic matter (24%) to the estimated average values of PM10 apart from other components e.g., soil/crustal matter (16%), ammonium sulphate (7%), ammonium nitrate (6%), aged sea salt (5%) and light absorbing carbon (4%). Positive Matrix Factorization (PMF) analysis quantified the sector wise contribution from the secondary aerosols (21.7%), soil dust (20.7%), fossil fuel combustion (17.4%), vehicle emissions (16.8%), and biomass burning (13.4%) to PM10 mass at the observational site of Delhi. Ó 2013 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Address: Radio and Atmospheric Sciences Division, CSIR-National Physical Laboratory, Dr. K. S. Krishnan Road, New Delhi 110 012, India. Tel.: +91 11 45609448; fax: +91 11 45609310. E-mail addresses: [email protected], [email protected] (S.K. Sharma). 2212-0955/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.uclim.2013.11.002

Please cite this article in press as: Sharma, S.K., et al. Source apportionment of PM10 by using positive matrix factorization at an urban site of Delhi, India. Urban Climate (2013), http://dx.doi.org/10.1016/j.uclim.2013.11.002

2

S.K. Sharma et al. / Urban Climate xxx (2013) xxx–xxx

1. Introduction Over the continents, particulate matters (PM) are produced either by various natural processes or due to several anthropogenic activities and it is now well documented (Langner and Rodhe, 1991; Andreae and Merlet, 2001; Lawrence and Lelieveld, 2010) that anthropogenic activities have increased atmospheric aerosol concentrations. Both fossil fuel combustion and biomass burning (Andreae and Crutzen, 1997; Saud et al., 2012) contributes to aerosol production by emitting primary aerosol particles (flyash, dust and black carbon, etc.) and aerosol precursor gases (SO2, NOx and volatile organic compounds) which form secondary aerosol particles through gas-to-particle conversion (Sharma et al., 2012). Particulate matter has been extensively studied in recent years due to its potential impacts on health and air quality (Schwartz et al., 1996; Li et al., 2009). The particulate air mass has shown significant seasonal variability at various locations in India (Sharma et al., 1995; Sharma et al., 2007b). The variation in PM levels over Delhi may be due to influences of several anthropogenic activities e.g., soil dust emission, biomass burning, vehicular emission, industries, formation of inorganic secondary particles, etc. (Khillare et al., 2004; Tiwari et al., 2009; Shridhar et al., 2010). Several researchers have reported about carbonaceous aerosols and chemical composition of total suspended particulates (TSP) in the Indian region (Kulshrestha et al., 1998; Venkataraman and Rao, 2002; Tare et al., 2006; Parashar et al., 2005; Tiwari et al., 2009; Ram and Sarin, 2011). Ram and Sarin (2010, references therein) suggested that carbonaceous aerosol contribute 30–35% of the TSP mass over Indo Gangetic Plain (IGP) during winter whereas, contribution of WSIC is of the order of 15–20% (Tare et al., 2006). Recently, Mandal et al. (2013) had reported very higher annual average concentrations of OC (93.0 ± 44.7 lg m3), EC (27.3 ± 13.4 lg m3) and total carbonaceous aerosols (176.1 ± 84.7 lg m3) in PM10 (280.7 ± 126.1 lg m3) at an industrial area of Delhi, India. Ambient aerosols consist of mineral dust, metals, sea salts as well as organic and inorganic pollutants. The relative abundance of these components is highly variable both temporally and spatially. Traditionally, most studies were carried out using inorganic trace elements like, Fe, Zn, Pb, Cr, Al and Ni. However, since many of the trace elements are emitted from a range of sources (e.g., Zn is emitted from tyre wear as well as refuse burning) and it may be difficult to apportion the PM to its sources with a high degree of confidence. The contribution of different sectors to the ambient concentration of particulate matter is to be quantified for possible mitigation and reduction their burden in future. Hence, the development and application of improved tools are required for the identification and quantification of different sources of the atmospheric aerosols at a particular location. Different receptor models have been used for the identification of sources and their respective contributions to airborne PM across the world (Kumar et al., 2001; Begum et al., 2004; Lai et al., 2005; Gu et al., 2011). Receptor modeling offers a method to complete the process by measurements of the pollutant concentrations at a sampling site (Hopke, 1991). During last few years, receptor models have been used effectively for developing air quality management plans in various cities. Different models including principal component analysis/absolute principal components scores (PCA–APCS), edge analysis (UNMIX), chemical mass balance (CMB) and PMF have been applied by several researchers to identify and establish the sources contributing to ambient air. Rizzo and Scheff (2007) compared the magnitude of source contributions resolved by each model and examined correlations between PMF and CMB resolved contributions. They observed that the major factors correlated well and were similar in magnitude. Additionally, PMF resolved source profiles were generally similar to measured source profiles. Recently, Callén et al. (2009) carried out source apportionment of PM10 in Zaragoza, Spain by three multivariate receptor models based on factor analysis: PCA–APCS, UNMIX and PMF. Special attention was paid to the models comparison in order to determine which were more adequate for the apportionment. They concluded that greater requirements of measure of uncertainty in PMF permitted to obtain better results than with the other two models: PCA–APCS and UNMIX. Therefore, in the present study, source apportionment of PM10 has been carried out using PMF model. PMF is a multivariate receptor model developed by Paatero and Tapper (1994), Paatero (1997), using a least squares approach. PMF generally solves the problem arising in the factor analysis by integrating non-negativity constraints in the optimization process and utilizing the error estimates for each data value as a point-by-point weight (Begum et al., 2004) and has been applied

Please cite this article in press as: Sharma, S.K., et al. Source apportionment of PM10 by using positive matrix factorization at an urban site of Delhi, India. Urban Climate (2013), http://dx.doi.org/10.1016/j.uclim.2013.11.002

S.K. Sharma et al. / Urban Climate xxx (2013) xxx–xxx

3

successfully worldwide for such studies (Polissar, 1998; Kim et al., 2004; Lee and Hopke, 2006; Karanasiou et al., 2009). This method requires a substantial number (at least 50) of separate samples and works best with a large dataset in which the number of samples far exceeds the number of analytical variables. A minimum variable to case ratio of 1:3 should be mentioned in order to obtain accurate results (Thurston and Spengler, 1985; Pant and Harrison, 2012). For a clearer distinction, it is better to have short sampling time so that overlap of multiple point source contributions to a given sample in minimized. Past knowledge of source chemical profiles is used to assign factors to sources, and typically identifications of six to seven different sources are a good outcome (Pant and Harrison, 2012). In the present study, first, we have characterized the chemical components of PM10 using different analytical methods, then, PMF model has been used to identify source profiles and apportionment of PM10 at the observational site of Delhi. The study includes the concentrations of air mass of PM10, OC, EC, WSIC and major (Na, Mg, Al, Ca, K and Fe) and trace elements (P, S, Cl, Cr, Zn and Mn) at an urban location of Delhi for the period of January 2010 to December 2011. 2. Methodology 2.1. Description of site PM10 samples (n = 102) were collected (on every Wednesday; 4–5 samples in a month) periodically at sampling site of the National Physical Laboratory (NPL), New Delhi (28°380 N, 77°100 E; 218 m amsl), India (Fig. 1) during January 2010 to December 2011. The sampling site is amenable to free wind flow from all the directions. The sampling location represents a typical urban atmosphere, surrounded by huge roadside traffic (100 m) and agricultural fields in the southwest direction (500 m). There are different small, medium and large scale industries in and around Delhi (4–5 km). The total number of registered vehicles in the city was of the order of 6.35 million in 2010–11 (Delhi Statistical Handbook 2011). This area is under the influence of air mass flow from Northeast to North-west in winter and from Southeast to South-west in the summer (Goyal and Sidhartha, 2002). In addition, Delhi experiences severe fog and haze weather conditions and poor visibility during wintertime. Roadside vehicle, industrial emission and biomass burning, etc. could be the major sources of carbonaceous aerosols and several other pollutants. The occasional occurrence of dust storm may contribute the presence of mineral dust significantly to the aerosol loading in summer time at Delhi (Ram et al., 2010). The temperature of Delhi varies from minimum (monthly Ave: 12.9 °C) in winter (November to February) to maximum (monthly Ave: 34.8 °C) in summer (March to June). The average rainfall in Delhi during monsoon (July to October) is of the order of 730 mm. Monthly variation of ambient temperature; relative humidity (RH), wind direction and wind speed at the observational site has given in Fig. S-1 (in supplementary information). 2.2. Sampling method PM10 samples (n = 102) were collected (weekly basis on every Wednesday; 4–5 samples in each month) on prebaked (at 550 °C for 6 h before sample collection) quartz fiber filters (QM-A) at 10 m height (above ground level) at sampling site of NPL, New Delhi during January 2010 to December 2011. Respirable Dust Sampler (Model: PEM-RDS 8NL, S/No.:1709; Make: M/s. Polltech Instruments, Mumbai, India) was used to collect the PM10 samples. The blower flow rate of the sampler was calibrated (with the accuracy of ±1% of F.S.) with Top Loading Orifice Calibrator traceable to National Standard. The flow meter of the sampler was calibrated (with the accuracy of ±2% of F.S.) with Air Flow Calibrator traceable to National Standard. Ambient air was passed through a QM-A filter paper (20  25 cm2) at a flow rate of 1.13 m3 min1 (accuracy ±2%) for 8 h during the sampling period (1000–1800 h). In general, rush hour started from 900 to 1100 h and 1700 to 1900 h in Delhi (Sharma et al., 2010) hence, the sampling had been started at 1000 h. The filter papers were weighed before and after the sampling during the experiment in order to determine the mass of the PM10 collected. The concentration of PM10 (lg m3) was calculated on the basis of the difference between initial and final Please cite this article in press as: Sharma, S.K., et al. Source apportionment of PM10 by using positive matrix factorization at an urban site of Delhi, India. Urban Climate (2013), http://dx.doi.org/10.1016/j.uclim.2013.11.002

4

S.K. Sharma et al. / Urban Climate xxx (2013) xxx–xxx

Fig. 1. Map of sampling location.

weights of the QM-A filter papers measured by a micro balance (resolution: ±1 lg) was determined by dividing the amount of total volume passed during the sampling. After collecting samples, filter papers were stored under dry condition at 20 °C in de-freezer prior to analysis. Please cite this article in press as: Sharma, S.K., et al. Source apportionment of PM10 by using positive matrix factorization at an urban site of Delhi, India. Urban Climate (2013), http://dx.doi.org/10.1016/j.uclim.2013.11.002

S.K. Sharma et al. / Urban Climate xxx (2013) xxx–xxx

5

The meteorological parameters (temperature and RH, wind speed, wind direction and pressure, etc.) were measured by using sensors of meteorological tower (5 stages tower of 30 m height), which is 100 m away from the observational site in the same campus. Tower measures above mentioned parameters at five different layers. We use the meteorological data available at 10 m height (i.e., temperature (accuracy ±1 °C), relative humidity (RH) (accuracy ±2%), wind speed (accuracy ±2%) and wind direction (accuracy ±2°)) during the study period. 2.3. Chemical analysis Analysis of OC and EC in ambient PM10 samples (n = 102) has been carried out by OC/EC carbon analyzer (Model: DRI 2001A; Make: Atmoslytic Inc., Calabasas, CA, USA) following the USEPA Method ‘Improve Protocol’ with negative pyrolysis areas zeroed. The principle of the OC/EC carbon analyzer is based on the preferential oxidation of OC and EC at different temperatures in which the sample is heated to four temperature plateaus (140, 280, 480 and 580 °C) in pure helium and three temperature plateaus (580, 740 and 840 °C) in 98% helium and 2% oxygen (Chow et al., 2004). Approximately 0.5 cm2 area of QM-A filter paper was cut using the proper punch and the values are reported as lg cm2 as given by the instrumental analysis software. Details of OC and EC analysis of PM10 are described in Saud et al. (2012). Each filter paper was analyzed triplicate with several blank run to get the representative estimation of OC and EC mass in PM10. The collected PM10 samples were extracted in de-ionized water having conductivity > 18.2 MO for 90 min in ultrasonic extractor for the determination of WSIC. The WSIC of PM10 has been analyzed by 2 Ion Chromatograph (Model: DIONEX-ICS-3000, USA). Concentrations of F, Cl, NO 3 and SO4 were determined by Ion Chromatograph (IC) using an Ion Pac-AS11-HC analytical column with a guard column, ASRS-300 4 mm anion micro-membrane suppressor, 20 mM NaOH (50% w/w) as eluent and tri+ 2+ ple-distilled water as regenerator. Li+, Na+, NHþ and Mg2+ were determined by using a 4 , K , Ca separation column with a guard column, suppressor CSRS-300 and 5 mM MSA (methane sulphonic acid) as eluent. Calibration standards have been prepared by National Institute of Standards and Technology (NIST, USA) traceable certified standards for calibration of IC. Several blank filters were also 2 + 2+ analyzed for cations (Li+, Na+, NHþ and Mg2+) and anions (F, Cl, NO 4 , K , Ca 3 and SO4 ). The analytical error (repeatability) was estimated to be 3% based on triplicate (n = 3) analysis. Details of WSIC analysis of PM10 are discussed in Sharma et al. (2012). The quantitative elemental analysis of PM10 samples was carried out using Rigaku ZSX Primus Wavelength Dispersive X-ray Fluorescence Spectrometer (WD-XRF). The spectrometer has an Rh-target, end window, 4 kW, sealed X-ray tube as the excitation source and scintillation counter (SC) for heavy elements and flow proportional counter (F-PC) for light elements as detectors. The measurement conditions for the Ka X-spectral lines of the identified elements (Mg, Al, P, S, Si, Cl, K, Ca, Ti, Cr, Mn, Fe and Zn) are: RX25 analyzer crystal and F-PC detector for Mg, PET analyzer crystal and FPC detector for Al, Ge analyzer crystal and F-PC detector for P, S and Cl, LiF(200) analyzer crystal and F-PC detector for K and Ca, LiF(200) analyzer crystal and SC detector for Ti, Cr, Mn, Fe and Zn. Data acquisition and quantitative analysis were carried out by using ZSX software (Rigaku Corporation, Japan). Fundamental Parameter method was used for the quantitative analysis. 2.4. Enrichment factor (EF) In order to have a better understanding about the origin (anthropogenic or natural) of the elements and their abundance in the atmospheric aerosols, the crustal enrichment factors (EF) have been estimated in PM10. The EF of the elements in PM10 mass concentration was calculated (Taylor and McLennan, 1985) as:

EF ¼

Elsample =Xsample Elcrust =Xcrust

ð1Þ

where Elsample and Xsample are the element (El) and the reference element (X) mass concentration in the sample respectively, and Elcrust and Xcrust are the element (El) and the reference element (X) mass Please cite this article in press as: Sharma, S.K., et al. Source apportionment of PM10 by using positive matrix factorization at an urban site of Delhi, India. Urban Climate (2013), http://dx.doi.org/10.1016/j.uclim.2013.11.002

6

S.K. Sharma et al. / Urban Climate xxx (2013) xxx–xxx

concentrations in the upper continental crust, respectively and Al is used for reference element in this case. 2.5. Positive matrix factorization (PMF) In the present study, PMF (PMF v3.0; USEPA2008) was used to quantify the contribution of various emission sources to PM10 mass. PMF is a multivariate factor analysis tool that decomposes a matrix of speciated sample data into two matrices: factor contributions and factor profiles. The PMF v3.0 model requires two input files: one of the measured concentrations of the species and another for the estimated uncertainty of the concentration. The PMF model in the details has been described by Paatero and Tapper (1994) and Paatero (1997). A speciated data set can be viewed as a data matrix X of i by j dimensions, in which i number of samples and j chemical species are measured. The aim of multivariate receptor modeling, for example with PMF, is to identify a number of factors p, the species profile f of each source, and the amount of mass g contributed by each factor to each individual sample which is given as:

X ij ¼

p X fkj g ik þ eij

ð2Þ

k¼1

where eij is the residual for each sample/species. Results are constrained so that no sample can have a negative source contribution. PMF allows each data point to be individually weighed. This feature allows the analyst to adjust the influence of each data point, depending on the confidence in the measurement. For example, data below detection limit can be retained for use in the model, with the associated uncertainty adjusted so these data points have less influence on the solution than measurements above the detection limit. The PMF solution minimizes the object function Q, based upon these uncertainties (u) as follows.

Q¼

P 2 n X m  X X ij  pk¼1 g ik fkj uij i¼1 j¼1

ð3Þ

where Xij are the measured concentration (in lg m3), uij are the estimated uncertainty (in lg m3), n is the number of samples, m is the number of species and p is the number of sources including in the analysis. The detail descriptions of EPA PMF v3.0 are described in the EPA PMF User Guide (2008). In this study, information on chemical properties of 102 PM10 samples has been used as input to the PMF model for total 25 parameters. Categorization of quality of data was based on the signal to noise ratio (S/N) and the percentage of sample method detection limit (MDL). Those species which have S/N P 2 were categorized as strong in data quality. Those with S/N between 0.2 and 2 were categorized as weak in quality. These species are not likely to provide enough variation in concentration and therefore contribute to the noise in the results. Those species with an S/N ratio below 0.2 are classified as bad values and were thus excluded from further analysis. In the present case S/N ratio of the individual species is more than 2 except PM10. 3. Results and discussions 3.1. Mass concentration of PM10 and reconstructed PM10 The average mass concentrations of PM10, OC, EC, WSIC and major and minor elements are summarized in Table 1 (The concentration of ionic form of Na+, K+, Mg2+, Ca2+ and Cl were included in concentration of their respective elemental form (Na, K, Mg, Ca and Cl) as major and minor elements which are analyzed by non-destructive method using X-ray Fluorescence Spectrometer). The mass concentration of PM10 has varied from 93.4 to 386.2 lg m3 with an average value of 191.4 ± 45.5 lg m3 (±values are standard deviation; n = 102) during the study period. Significant seasonal variation was noticed in the mass concentration of PM10 and its chemical composition with maxima during winter (PM10: 241.4 ± 50.5 lg m3; OC: 34.7 ± 10.2 lg m3; EC: 10.9 ± 3.0 lg m3) Please cite this article in press as: Sharma, S.K., et al. Source apportionment of PM10 by using positive matrix factorization at an urban site of Delhi, India. Urban Climate (2013), http://dx.doi.org/10.1016/j.uclim.2013.11.002

7

S.K. Sharma et al. / Urban Climate xxx (2013) xxx–xxx Table 1 The annual average concentrations of OC, EC, WSIC and major and trace elements of PM10 (lg m-3) at Delhi. Species

Annual (n = 10)

Range

Mass OC EC Cl

191.4 ± 45.5 25.8 ± 8.3 7.8 ± 3.4 3.36 ± 1.27 9.86 ± 3.27

93.4–386.2 9.4–69.0 1.8–29.5 1.65–6.59 2.08–18.70

8.89 ± 3.45 4.93 ± 1.03 3.24 ± 1.71 1.64 ± 0.64 0.64 ± 0.18 4.88 ± 1.17 4.31 ± 1.62 1.80 ± 0.57 2.95 ± 0.93 2.06 ± 0.51 0.48 ± 0.23 4.40 ± 1.62 3.53 ± 1.13 2.04 ± 0.91 8.11 ± 2.81 0.28 ± 0.19 0.16 ± 0.14 1.00 ± 0.69 0.51 ± 0.49 0.02 ± 0.01

0.67–17.52 0.37–14.60 1.33–5.13 1.62–2.38 0.19–0.88 1.95–6.85 2.12–6.98 0.71–3.01 1.53–4.59 1.01–4.22 0.07–0.92 1.76–6.89 1.86–5.69 0.87–3.55 2.97–10.95 0.13–0.63 0.03–0.52 0.30–2.43 0.01–1.42 0.003–0.05

SO2 4 NO 3 NHþ 4 + Na + K Mg2+ Ca2+ Na Mg Al Si P S Cl K Ca Cr Ti Fe Zn Mn a

Seasons Winter (W) (n = 36

Summer (S) (n = 36

Monsoon (M) (n = 30

241.4 ± 50.5 a 34.7 ± 10.2 a 10.9 ± 3.0 a 5.0 ± 3.5 a 11.6 ± 2.7

192.7 ± 28.4 25.2 ± 12.2 a 7.4 ± 3.1 3.5 ± 1.4 a 9.2 ± 6.2

a

a

14.1 ± 5.1 a 9.6 ± 3.6 2.3 ± 1.1 1.7 ± 0.6 0.7 ± 0.2 4.8 ± 09 6.1 ± 2.1 1.6 ± 0.7 3.5 ± 1.1 3.1 ± 1.3 0.4 ± 0.1 3.9 ± 1.7 4.4 ± 1.8 2.0 ± 0.9 7.6 ± 3.7 0.37 ± 0.08 0.23 ± 0.02 0.54 ± 0.16 0.55 ± 0.12 0.014 ± 0.002

a

5.1 ± 3.4 2.6 ± 1.5 3.3 ± 0.7 1.5 ± 07 0.6 ± 0.3 4.7 ± 1.8 3.3 ± 1.1 1.8 ± 0.6 2.9 ± 1.1 1.9 ± 0.6 0.6 ± 0.2 5.2 ± 2.1 2.9 ± 1.0 2.5 ± 0.9 7.6 ± 2.7 0.14 ± 0.02 0.06 ± 0.01 0.96 ± 0.62 0.15 ± 0.02 0.03 ± 0.004 a

140.1 ± 43.9 a 15.5 ± 7.5 a 4.9 ± 2.3 2.5 ± 1.1 a 8.8 ± 3.6 a

4.9 ± 3.1 2.5 ± 1.5 4.2 ± 1.8 1.7 ± 08 0.6 ± 0.2 5.2 ± 1.4 3.6 ± 1.7 1.9 ± 0.6 2.4 ± 0.9 1.1 ± 0.5 0.4 ± 0.1 4.1 ± 1.1 3.2 ± 1.0 1.6 ± 0.7 9.1 ± 2.6 0.32 ± 0.05 0.18 ± 0.02 1.49 ± 0.82 0.84 ± 0.23 0.02 ± 0.001 a

Significantly (intra seasonal) different at P < 0.05; (±standard deviation).

and minima during monsoon (PM10: 140.1 ± 43.9 lg m3; OC: 15.5 ± 7.5 lg m3; EC: 4.9 ± 2.3 lg m3) (Table 1). Average mass concentrations of OC, EC and WSIC, major and minor elements of PM10 during winter, summer and monsoon seasons are also summarized in Table 1. It has been observed that the average of total carbon (TC = OC + EC) concentration contribute 18% of PM10 mass, whereas, WSIC (WSIC: sum of the concentrations of the cations and anions) account for 23% of PM10 mass (Table 1). The unidentified mass (UN) and major and trace elements (Na, Mg, Al, P, S, Si, Cl, K, Ca, Cr, Ti, Fe, Zn and Mn) accounts for 42% and 17%, respectively of total PM10 mass (Rengarajan et al., 2007; Ram and Sarin, 2011). The unidentified mass (UM) is obtained by subtracting TC, WSIC and major and trace element concentrations from the PM10 mass concentration. Figs. S-2 and S-3 (in supplementary information) shows the monthly variation in concentrations of PM10, OC, EC and WSIC during January 2010 to December 2011 at Delhi. In order to obtain mass closure, constituents of PM10 mass were reconstructed (RCPM10) using IMPROVE equation (Chan et al., 1997; Malm et al., 1994; Malm et al., 2007) by the sum of concentrations of ammonium sulphate (AS), ammonium nitrate (AN), particulate organic matter (POM), light absorbing carbon (LAC), sea salts (SS) and soil (RCPM10 = [AS] + [AN] + [POM] + [LAC] + [SS] + [Soil]). The mass difference (dM = PM10  RCPM10) of PM10 was calculated by difference between PM10 and RCPM10. The majority of AS in the atmosphere produced through chemical reactions of SO2. Anthropogenic SO2 is emitted through industrial activities such as coal and diesel fuel combustion. The degree of acidity of sulphate dependent on the availability of ammonia to neutralize the H2SO4 formed from SO2. Sulphate acidity can vary spatially and temporally (Zhang et al., 2011). The average AS was estimated of the order of 13.6 lg m3 (1.375[SO2 4 ]), which is contributed for 7% of RCPM10. Ammonium nitrate forms from the reversible reaction of gas-phase NH3 and HNO3 (Sharma et al., 2013). Sources of oxidized nitrogen include combustion of fossil fuels such as from coal fired power plants, on road mobile sources and non road mobile sources. Reconstructed AN (1.29[NO 3 ]) was estimated of the order of 11.5 lg m3 which is contributed for 6% of RCPM10 (Fig. 2). Please cite this article in press as: Sharma, S.K., et al. Source apportionment of PM10 by using positive matrix factorization at an urban site of Delhi, India. Urban Climate (2013), http://dx.doi.org/10.1016/j.uclim.2013.11.002

8

S.K. Sharma et al. / Urban Climate xxx (2013) xxx–xxx

Fig. 2. Reconstructed PM10 mass (AS: ammonium sulphate; AN: ammonium nitrate; POM: particulate organic matter; LAC: light absorbing carbon, SS: sea salts and dM: mass difference) by using IMPROVE equation.

The OC multiplier (1.8[OC]) used to estimate POM is an estimate of average molecular weight per carbon weight for OC aerosol and takes account contributions from other elements associated with organic matter. During the study POM was estimated of the order of 46.4 lg m3 (24% of RCPM10). The sources of POM in the atmosphere are both primary emissions and secondary formation. Primary emissions include particle mass emitted directly from combustions of fossil fuels or biomass. Secondary organic aerosols form in the atmosphere from the oxidation of gas-phase precursors from both anthropogenic and biogenic sources. LAC is referred as black carbon, elemental carbon or graphite carbon, is produced directly through emission from incomplete combustion of fossil fuels or biomass burning. Replacing EC with LAC avoids potential confusion regarding the type of carbon particles responsible for light absorption. In the present case, the annual average LAC was estimated as 7.8 lg m3 (4% of RCPM10) (Table S-1 in supplementary information). Sources of soil dust in the atmosphere include entrainment from deserts, paved and unpaved roads, agricultural activity, construction and fire (Seinfeld, 1986). Soil sources at the observational site are expected to be higher due to the impact of local as well as transboundary transport from the Thar Desert and other Asian region. Reconstructed soil mass 1.16{(1.90[Al] + 2.15[Si] + 1.41[Ca] + 2.09[Fe] + 1.67[Ti])} concentrations are estimated by the sums of the oxides of elements that are typically associated with soil (Al2O3, SiO2, CaO, K2O, FeO, Fe2O3, TiO2), with correction for other compounds such as MgO, Na2O, H2O and carbonates (Chan et al., 1997; Malm et al., 2007). The average reconstructed concentration of soil at the observational site was estimated of the order of 30.3 lg m3 (16% of RCPM10) (Fig. 2). The average mass difference between PM10 and RCPM10 was recorded as 72.6 lg m3 (38% of PM10 mass) during the study period (Table S-1 in supplementary information). The mass difference between observed PM10 and RCPM10 may be attributed to unidentified mass of PM10. 3.2. Major and trace elements Out of all major and trace elements, Ca has recorded higher concentration with an average value of 8.11 ± 2.81 lg m3 followed by Na (4.31 ± 1.62 lg m3), S (4.40 ± 1.62 lg m3), Cl (3.53 ± 1.13 lg m3), Al (2.95 ± 0.93 lg m3), K (2.04 ± 0.91 lg m3), and Mg (1.80 ± 0.57 lg m3) during January 2010 to December 2011 (Table 1). Perrino et al. (2011) have reported same order of values of major and trace metals during pre-Diwali and post-Diwali festival at Delhi. The EF of the various species (Na, Mg, Al, P, S, Cl, K, Ca, Cr, Ti, Fe, Zn and Mn) of PM10 has been estimated from the present study (Fig. 3). Al, Fe, Ti, Mn, K, Mg, Ca, and Na had a low EF value ranging from <1 to 5 in the average EF, which indicates that, elements mostly derived from soil. The average EF values of Cr, Zn and P have ranged from 5 to 150 indicating that they are likely to be affected by both soil and non-soil emission sources. The higher EF of Cr and Zn of PM10 mass were also attributed to industrial sources as metal manufacturing plants and storage are located near the sampling site. Generally Zn, Cu, Mn, S, Ni, Cd, Fe, Mo and Cr are used as a tracer for IE in India (Shridhar et al., 2010). Please cite this article in press as: Sharma, S.K., et al. Source apportionment of PM10 by using positive matrix factorization at an urban site of Delhi, India. Urban Climate (2013), http://dx.doi.org/10.1016/j.uclim.2013.11.002

9

S.K. Sharma et al. / Urban Climate xxx (2013) xxx–xxx

Fig. 3. Enrichment factors of the trace elements in PM10 mass at Delhi.

3.3. Source apportionment To quantify the contribution of possible sources to PM10 at Delhi, we have performed PMF analysis. The goodness of the model fit parameter ‘Q’ was evaluated to identify the optimal number of factors and the optimal solution should lie in Fpeak range. The PMF was applied to the data set consisting 25 species and 102 samples collected during January 2010 to December 2011 at Delhi. The mass fraction distribution of species was used to identify the sources were soil dust (SD), vehicle emission (VE), aged sea salt (SS), industrial emissions (IE), secondary aerosol (SA), biomass burning (BB) and fossil fuel combustion (FFC) for PM10 mass. Using PMF analysis, we could identify source profiles of PM10 mass concentrations (Fig. 4) and source contributions for all concentration (Fig. 5) are respectively.

SD

IE

BB

VE

SA

% of Species

Mass of Species (µg/µg)

FFC

SS

Fig. 4. PMF source profile of fossil fuel combustion (FFC), soil dust (SD), industrial emissions (IE), biomass burning (BB), vehicle emissions (VE), secondary aerosols (SA) and aged sea salt (SS) in Delhi for PM10 mass.

Please cite this article in press as: Sharma, S.K., et al. Source apportionment of PM10 by using positive matrix factorization at an urban site of Delhi, India. Urban Climate (2013), http://dx.doi.org/10.1016/j.uclim.2013.11.002

10

S.K. Sharma et al. / Urban Climate xxx (2013) xxx–xxx

Fig. 5. Percentage source apportionment of PM10 mass (FFC, fossil fuel combustion; SD, soil dust; IE, industrial emissions; BB, biomass burning; VE, vehicle emissions; SA, secondary aerosol and SS, sea salt) over Delhi estimated by PMF.

Source 1: Soil dust includes most of the crustal elements and has high concentration of Fe, Ca, Na, Mg, Al and K. These elements are major constituents of airborne soil and road dust and usually contribute to coarse aerosol (Lough et al., 2005). The concentration of Ca of the PM10 is associated with the re-suspension from agricultural fields or bare soils by local winds. In the present study, PMF analysis showed that SD has contributed 20.7% of aerosol mass in PM10 at observational site. Crustal elements typically used as tracers for soil and/or crustal re-suspension include Al, Si, Ca, Mg, Fe and Na (Begum et al., 2010). A whole array of element tracers has been used in India for identification of this source type include Al, Si, Ca, Ti, Fe, Pb, Cu Cr, Ni, Co and Mg (Khillare et al., 2004; Chelani et al., 2008; Gupta et al., 2007). Source 2: A vehicle exhaust is generally dominated by elemental carbon, Cu, Zn, Ba, Sb, Pb, Mn, Mo and Ni and widely used as markers of vehicle sources. In the present study, Zn, Mn and EC have been considered as an indicator of vehicle emission. Furusjo et al. (2007) suggested that the VE are associated with high concentration of Cu, Zn and Sb. Cu, Zn, Mn, Sb, Sn, Mo, Ba and Fe are markers of brake wear and can serve as indicators of traffic re-suspension (Querol et al., 2008). PMF analysis indicates that VE has contributed 16.8% in PM10 mass at Delhi. Internationally, EC (Lee et al., 2008) is used extensively as a marker for diesel exhaust. In India, V, Mn, Co, Pb and Zn are used tracer elements for identification of VE (Chelani et al., 2008). Vehicular emissions are a major source of PM and research indicates that they contribute between 10% and 80% to particulate matters in cities across India. Comparison of such estimates is made difficult by the fact that the various studies have quantified different vehicular sources (exhausts, re-suspension, abrasion, etc.). Source 3: Higher concentrations of Na, K and Cl in PM10 mass indicate the possible contribution of aged sea salt, which is supported by PMF analysis (SS 4.6%). Sievering et al. (1991) suggested that SO2 could react on the SS particle to produce SO2 4 in addition to the direct reaction to the gas phase H2SO4 with NaCl. The use of K offers possible confusion with wood/biomass burning combustion and Cl with coal burning, but a combination of the four elements (Na, K, Cl and Mg) should provide a reliable signature. Source 4: Secondary aerosols are mainly composed of ammonium sulphate and nitrate deriving primarily from the gaseous precursors NH3, SO2 and NOx. The abundance of gaseous NH3, SO2 and NOx are 2 at Delhi (Sharma et al., 2010). Secondary aerosols of PM10 (NO 3 and SO4 ) are originally from anthro2 pogenic or natural sources being formed in the atmosphere. The key markers of SA are NO 3 , SO4 and NHþ and were present in PM mass. Present PMF analysis shows that SA has contributed 21.7% for 10 4 PM10 mass concentrations. SO2 has been used as a marker for coal combustion in some Indian studies 4 whereas NHþ 4 has been used as a marker as a BB. Source 5: Biomass burning, wood burning and vegetative burning have been characterized as having high concentrations of K+ and SO2 4 by various source studies (Wu et al., 2007). These sources also could be possible facilitated by regional sources or long range transport. Results show that OC and EC are contributed by traffic emission, biomass burning, and wood burning and crop residue burning in PM10. PMF analysis also shows that BB has contributed 13.4% for PM10 mass in the present study. In Please cite this article in press as: Sharma, S.K., et al. Source apportionment of PM10 by using positive matrix factorization at an urban site of Delhi, India. Urban Climate (2013), http://dx.doi.org/10.1016/j.uclim.2013.11.002

11

S.K. Sharma et al. / Urban Climate xxx (2013) xxx–xxx

India K+ has been used a key marker for biomass/wood combustion for TSP, PM10 and PM2.5 (Shridhar et al., 2010) whereas levoglucosan is the key organic marker (Chowdhury et al., 2007). Biomass burning has been estimated to contribute in the range of 7–20% depending upon season and location. It has been reported to be one of the major sources in Delhi, particularly in winter due to combustion of wood (Sharma et al., 2003). Source 6: The results of the PMF analysis show that IE accounted for about 5.5% of PM10 mass. A range of tracers has been used for identification of IE including Cu, Cr, Mn, Ni, Co, Zn, etc. The high concentration of Cr, Mn, Zn, and S in PM10 mass at observational site attributed to industrial sources as metal manufacturing plants and storage are located near the sampling site. Begum et al. (2006) used Ni, Pb and S as markers for IE, Song et al. (2006) used Ni, Cr, Fe and Mn and Tauler et al. (2009) used Zn, Fe, Mn and Cd as tracers for IE. Generally Zn, Cu, Mn, S, Ni, Cd, Fe, Mo and Cr are used as a tracer for IE in India (Shridhar et al., 2010). Source 7: The higher concentrations of Al, Cl, Fe, Zn, Cr and SO2 4 at the sampling site clearly indicate the source of FFC of PM10 mass. Cr and Cd are known to occur at high temperatures during the combustion of coal, oil, refuse and so on. Ni and V are widely used as markers for the combustion of heating fuel (Vallius et al., 2005), while Se and Zn are representative marker species for oil fired power plants and coal combustion, respectively (Lee et al., 2002). PMF analysis shows that FFC has contributed 17.4% for PM10 mass in the present study. In international studies, a key marker for coal combustion include As, Se, Te and SO2 4 and it has been contributed between 6% and 30% to PM in

Table 2 Average contribution (%) from major source types, compiled from the results reported in recent source apportionment studies. No. of factors

Sea salt (%)

Crustal/ soil dust (%)

Vehicle emission (%)

Secondary aerosol (%)

Other combustion/ Industrial emission (%)

Reference

191.4

7

4.6

20.7

16.8

21.7

219.0

2

–

27.0

–

–

416.3

2

–

22.0

–

–

36.2 (FFC, BB, IE) 45.0 (BB and FFC) 60.0 (VE, IE)

545.8

5

–

25.0

–

–

Present study Tiwari et al. (2009) Khillare et al. (2004) Shridhar et al. (2010)

1176.0

5

15.0

41.0

15.0

–

114.0

5

15.0

17.7

23.0

–

–

5

–

37.0

–

–

Location

No. of Samples

Avg. concen. (lg m3)

Delhi, India

102 PM10 35 PM10 50 coarse 30 coarse

Delhi, India Delhi, India Delhi, India

Mumbai, India

50.0 (IE, BB, battery, refuse oil) 6.0 (coal combustion) 31.9 (FFC, IE) 35.0 (FFC, IE, Others) 8.35

Mumbai, India

45 coarse PM10

Kolkata, India

PM10

Shinjung, Taiwan Salamanca, Maxico Taiyuan, China Bandung, Indonesia

18 PM10

39.45

5

8.4

34.0

24.92

24.33

135 PM10 14 PM10

89.12

5

–

39.87

30.16

14.42

305.0

8

–

12.0

13.0

16.0

30.0

180 coarse

19.0

5

23.0

19.0

–

18.0

40.0

Barcelona, Spain Southwestern Oregon, US

243 coarse 493 fine

12.6

8

29.0

67.0

28.0

3.0

2.0

3.2

9

10.0

9.0

3.0

40.0

38.0

Eagle Farm, Brisbane, Austraila

28 fine

7.2

7

58.0

8.0

30.0

22.0

32.0

1.82

Kumar et al. (2001) Chelani et al. (2008) Gupta et al. (2007) Gugamsetty et al. (2012) Murillo et al. (2012) Zeng et al. (2010) Lestari and Mauliadi (2009) Amato et al. (2009) Hwang and Hopke (2007) Chan et al. (2011)

Please cite this article in press as: Sharma, S.K., et al. Source apportionment of PM10 by using positive matrix factorization at an urban site of Delhi, India. Urban Climate (2013), http://dx.doi.org/10.1016/j.uclim.2013.11.002

12

S.K. Sharma et al. / Urban Climate xxx (2013) xxx–xxx

different studies (Gupta et al., 2007; Sharma et al., 2007a). In Delhi, where three coal fired thermal power plants are situated within the city boundaries, Sharma et al. (2007a) attributed 17% of variance as per Principal Component Analysis (PCA) results of coal combustion while Srivastava and Jain (2007) attributed 15% of the variance of PM0.7 fraction of the source. The results of the PMF analysis show that the SA (21.7%), SD (20.7%), FFC (17.4%), VE (16.8%), BB (13.4%), IE (5.4%) and aged SS (4.6%) are the major sources of PM10 mass at the observational site of Delhi. Tiwari et al. (2009) reported that SD contributed 27.0% of PM10 mass at Delhi whereas Khillare et al. (2004) estimated as 22.0% SD of coarse particle at Delhi. Chelani et al. (2008) reported that SD and VE contributed 17.7% and 23.0% of PM10 mass respectively at Mumbai whereas SD contributed 37.0% of PM10 mass at Kolkata (Gupta et al., 2007). In the present study soil dust contributes to 20.7% of PM10 mass at Delhi which is more or less same to previous study at Delhi whereas the quantifications of other source types were missing in previous study. Table 2 summarizes the major source contributions in the present study with the recent source apportionment studies in India and other part of the world. Gugamsetty et al. (2012) used the PMF model and analyzed that the SD (34.0%), VE (24.92%), SA (24.44%) and SS (8.4%) are the major sources of PM10 mass at Shinjung Taiwan. More or less similar source types were also reported at Salamanca, Maxico (Murillo et al., 2012) and other part of the world (Zeng et al., 2010; Amato et al., 2009; Hwang and Hopke, 2007; Chan et al., 2011). The concentrtaions of sources were varying at different locations might be due to source strength of PM at observational sites, sampling duration, sampling time, meteorological conditions, number of samples (with sampling intervals) and number of chemical species, etc. However, from the above comparison it is clear that the number and type of source factors derived from the PMF analysis in this work are similar to those reported in other studies.

4. Conclusions In the present paper, we have collected particulate samples (PM10) for the period of January 2010 to December 2011 and characterized the chemical composition of PM10 mass using different analytical methods. The concentrations of PM10, OC, EC, WSIC and major and trace elements of PM10 showed strong seasonal cycle with maxima during winter and minima during monsoon. Using PMF, a widely used receptor model, we have quantified seven source profiles of PM10 mass at Delhi. PMF analysis quantifies the contribution of secondary aerosols (21.7%), soil dust (20.7%), fossil fuel combustion (17.4%), vehicle emissions (16.8%), biomass burning (13.4%) and aged sea salts (4.6%) to PM10 mass concentration at the observational site of Delhi. The reconstructed PM10 mass (using IMPROVE equation) on average were also observed as particulate organic matter (24%), soil/crustal matter (16%), ammonium sulphate (7%), ammonium nitrate (6%), aged sea salts (5%) and light absorbing carbon (4%) were major contributors of PM10 mass at Delhi.

Acknowledgements The authors are thankful to Director, NPL, New Delhi and Head, Radio and Atmospheric Sciences Division, CSIR-NPL, New Delhi, India for their encouragement. The authors also acknowledge Council of Scientific and Industrial Research (CSIR), New Delhi for providing financial support for this study (under CSIR-EMPOWER Project). The author (SKS) thanks Department of Science & Technology, New Delhi and CSIR for providing the funds to attend the conference. Authors are thankful to reviewers for constructive suggestions to improve the manuscript. Authors are thankful to Dr. Thomas John, Scientist, RASD, CSIR-NPL, New Delhi for providing meteorological data.

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http:// dx.doi.org/10.1016/j.uclim.2013.11.002. Please cite this article in press as: Sharma, S.K., et al. Source apportionment of PM10 by using positive matrix factorization at an urban site of Delhi, India. Urban Climate (2013), http://dx.doi.org/10.1016/j.uclim.2013.11.002

S.K. Sharma et al. / Urban Climate xxx (2013) xxx–xxx

13

References Amato, F., Pandolfi, M., Escrig, A., Querol, X., Alastuey, A., Pey, J., Perez, N., Hopke, P.K., 2009. Quantifying road dust resuspension in urban environment by multilinear engine: a comparison with PMF2. Atmospheric Environment 43, 2770–2780. Andreae, M.O., Crutzen, P.J., 1997. Atmospheric aerosols: Biogeochemical sources and role in atmospheric chemistry. Science 276, 1052–1058. Andreae, M.O., Merlet, P., 2001. Emission of trace gases and aerosols from biomass burning. Global Biogeochemical Cycle 15, 955–966. Begum, B.A., Akhter, S., Sarker, L., Biswas, S.K., 2006. Gravimateric analysis of air filters and quality assurance in weighing. Nuclear Science Application 15, 36–41. Begum, B.A., Biswas, S.K., Hopke, P.K., 2010. Key issues in controlling air pollutants in Dhaka, Bangladesh. Atmospheric Environment. http://dx.doi.org/10.1016/j.atmosenv.2010.10.02. Begum, B.A., Kim, E., Biswas, S.K., Hopke, P.K., 2004. Investigation of sources of atmospheric aerosol at urban and semi-urban areas in Bangladesh. Atmospheric Environment 38, 3025–3038. Callén, M.S., de la Cruz, M.T., Lopez, J.M., Navarro, M.V., Mastral, A.M., 2009. Comparison of receptor models for source apportionment of the PM10 in Zaragoza (Spain). Chemosphere 76, 1120–1129. Chan, Y.C., Hawas, O., Hawker, D., Vowles, P., Cohen, D.D., Stelcer, E., Simpson, R., Golding, G., Christensen, E., 2011. Using multiple type composition data and wind data in PMF analysis to apportion and locate sources of air pollutants. Atmospheric Environment 45, 439–449. Chan, Y.C., Simpson, R.W., Mctainsh, G.H., Vowels, P.D., Cohen, D.D., Bailey, G.M., 1997. Characterisation of chemical species PM2.5 and PM10 aerosols in Brisbane. Australia. Atmospheric Environment 31, 3773–3785. Chelani, A.B., Gajghate, D.G., Devotta, S., 2008. Source apportionment of PM10 in Mumbai, India using CMB model. Bulletin of Environmental Contamination and Toxicology 81, 190–195. Chow, J.C., Watson, J.G., Chen, L.W.A., Arnott, W.P., Moosmuller, H., 2004. Equivalence of elemental carbon by thermal/optical reflectance and transmittance with different temperature protocols. Environmental Science and Technology 38, 4414–4422. Chowdhury, Z., Zheng, M., Schauer, J.J., Sheesley, R.J., Salmon, L.G., Cass, G.R., Russell, A.G., 2007. Speciation of fine organic carbon particles and source apportionment of PM2.5 in Indian cities. Journal of Geophysical Research 112, D15303. http:// dx.doi.org/10.1029/2007JD008386. Furusjo, E., Sternbeck, J., Cousins, A.P., 2007. PM10 source characterization at urban and highway roadside locations. Science of Total Environment 387, 206–219. Goyal, P., Sidhartha, 2002. Effect of winds on SO2 and SPM concentration in Delhi. Atmospheric Environment 36, 2925–2930. Gu, J., Pitz, M., Schenlle-Kreis, J., Diemer, J., Reller, A., Zimmermann, R., Soentgen, J., Stoelzel, M., Wichmann, H.E., Peters, A., Cyrys, J., 2011. Source apportionment of ambient particles: comparison of positive matrix factorization analysis applied to particle size distribution and chemical composition data. Atmospheric Environment 45, 1849–1857. Gugamsetty, B., Wei, H., Liu, C.N., Awasthi, A., Hsu, S.C., Tsai, C.J., Roam, G.D., Wu, Y.C., Chen, F.C., 2012. Source characterization and apportionment of PM10, PM2.5 and PM0.1 by using Positive Matrix Factorization. Aerosol and Air Quality Research 12, 476–491. Gupta, A.K., Karar, K., Srivastava, A., 2007. Chemical mass balance source apportionment of PM10 and TSP in residential and industrial sites of an urban region of Kolkata, India. Journal of Hazardous Materials 142, 279–287. Hopke, P.K., 1991. Receptor Modeling for Air Quality Management. Elsevier, Amsterdam. Hwang, I., Hopke, P.K., 2007. Estimation of source apportionment and potential source locations of PM2.5 at a west coastal IMPROVE site. Atmospheric Environment 41, 506–518. Karanasiou, A.A., Siskos, P.A., Eleftheriadis, K., 2009. Assessment of source apportionment by positive matrix factorization analysis on fine and coarse urban aerosol size fractions. Atmospheric Environment 43, 3385–3395. Khillare, P.K., Balachandran, S., Meena, B.R., 2004. Spatial and temporal variation of heavy metals in atmospheric aerosol in India. Environmental Monitoring and Assessment 90, 1–21. Kim, E., Hopke, P.K., 2004. Source apportionment of fine particles in Washington, DC, utilizing temperature-resolved carbon fractions. Journal of Air Waste Management Association 53, 773–785. Kulshrestha, U.C., Saxena, A., Kumar, N., Kumari, K.M., Srivastava, S.S., 1998. Chemical composition and size differentiated aerosols at a suburban site in a semi-arid tract of India. Journal of Atmospheric Chemistry 29, 109–118. Kumar, A.V., Patil, R.S., Nambi, K.S.V., 2001. Source apportionment of suspended particulate matter at two traffic junctions in Mumbai, India. Atmospheric Environment 35, 4245–4251. Lai, C.H., Chen, K.S., Ho, Y.T., Peng, Y.P., Chou, Y., 2005. Receptor modeling of source contributions to atmospheric hydrocarbons in urban Kaohsiung. Taiwan. Atmospheric Environment 39, 4543–4559. Langner, J., Rhode, H., 1991. A global three-dimensional model of the global sulfur cycle. Journal of Atmospheric Chemistry 13, 225–263. Lawrence, M.G., Lelieveld, J., 2010. Atmospheric pollutant outflow from southern Asia: a review. Atmospheric Chemistry and Physics 10, 11017–11096. Lee, J.H., Hopke, P.K., 2006. Apportioning sources of PM2.5 in St. Luis, MO using speciation trends network data. Atmospheric Environment 40, S360–S377. Lee, S., Liu, W., Wang, Y., Russell, A.G., Edgerton, E.S., 2008. Source apportionment of PM2.5: comparing PMF and CMB results for four ambient monitoring sites in the southeastern United States. Atmospheric Environment 42, 4126–4137. Lee, J.H., Yoshida, Y., Turpin, B.J., Hopke, P.K., Poirot, P.J., Lioy, P.J., Oxley, J.C., 2002. Identification of sources contributing to MidAtlantic Regional Aerosol. Journal of Air Waste Management Association 52, 1186–1205. Lestari, P., Mauliadi, Y.D., 2009. Source apportionment of particular matter at urban mixed sites in Indonesia using PMF. Atmospheric Environment 43, 1760–1770. Li, X., Wang, S., Duan, L., Hao, J., Nie, Y., 2009. Carbonaceous aerosol emissions from household biofuel combustion in China. Environmental Science and Technology 43, 6076–6081. Lough, G.C., Schauer, J.J., Park, J.S., Shafer, M.M., DeMinter, J.T., Weinstein, J.P., 2005. Emissions of metals associated with motor vehicle roadways. Environmental Science and Technology 39, 826–836.

Please cite this article in press as: Sharma, S.K., et al. Source apportionment of PM10 by using positive matrix factorization at an urban site of Delhi, India. Urban Climate (2013), http://dx.doi.org/10.1016/j.uclim.2013.11.002

14

S.K. Sharma et al. / Urban Climate xxx (2013) xxx–xxx

Malm, W.C., Pitchford, M.L., McDade, C., Ashbaugh, L.L., 2007. Coarse particle speciation at selected locations in the rural continental United States. Atmospheric Environment 41, 2225–2239. Malm, W.C., Sisler, J.F., Huffman, D., Eldred, R.A., Cahill, T.A., 1994. Spatial and seasonal trends in particle concentration and optical extinction in the United States. Journal of Geophysical Research 99 (D1), 1347–1370. Mandal, P., Saud, T., Sarkar, R., Mandal, A., Sharma, S.K., Mandal, T.K., Bassin, J.K., 2013. High seasonal variation of atmospheric C and particulate concentrations in Delhi, India. Environmental Chemistry Letters. http://dx.doi.org/10.1007/s10311-0130438-y. Murillo, J.H., Ramos, A.C., Carcia, F.A., Jimenez, S.B., Cardenas, B., Mizohata, A., 2012. Chemical composition of PM2.5 particles in Salamanca, Guanajuato Mexico: source apportionment with receptor models. Atmospheric Research 107, 31–41. Paatero, P., 1997. Least squares formulation of robust nonnegative factor analysis. Atmospheric Environment 37, 23–35. Paatero, P., Tapper, U., 1994. Positive matrix factorization: a non-negative factor model with optimal utilization of error estimates of data values. Environmetrics 5, 111–126. Pant, P., Harrison, R.M., 2012. Critical review of receptor modelling for particulate matter: a case study of India. Atmospheric Environment 49, 1–12. Parashar, D.C., Gadi, R., Mandal, T.K., Mitra, A.P., 2005. Carbonaceous aerosol emissions from India. Atmospheric Environment 39, 7861–7871. Perrino, C., Tiwari, S., Catrambone, M., Torre, S.D., Rantica, E., Canepari, S., 2011. Chemical characterization of atmospheric PM in Delhi, India during different periods of the year including Diwali festival. Atmospheric Pollution Research 2, 418–427. Polissar, A.V., 1998. Atmospheric aerosol over Alaska: 2. Elemental composition and sources. Journal of Geophysical Research 10, 19045–19057. Querol, X., Alastuey, A., Moreno, T., Viana, M.M., Castillo, S., Pey, J., Rodriguez, S., Artinano, B., Salvador, P., Sanchez, M., Garcia Dos Santos, S., Herce Garraleta, M.D., Fernandez-Patier, R., Moreno-Grau, S., Negral, L., Minguillon, M.C., Monfort, E., Sanz, M.J., Palomo-Marin, R., Pinilla-Gil, E., Cuevas, E., de la Rosa, J., Sanchez de la Campa, A., 2008. Spatial and temporal variations in airborne particulate matter (PM10 and PM2.5) across Spain 1999–2005. Atmospheric Environment 42, 3964–3979. Ram, K., Sarin, M.M., 2010. Spatio-temporal variability in atmospheric abundances of EC, and WSOC over northern India. Journal of Aerosol Science 41 (1), 88–98. Ram, K., Sarin, M.M., 2011. Day–night variability of EC, OC, WSOC and inorganic ions in urban environment of Indo-Gangetic Plain: implications to secondary aerosol formation. Atmospheric Environment 45, 460–468. Ram, K., Sarin, M.M., Tripathi, S.N., 2010. One-year record of carbonaceous aerosols from an urban location (Kanpur) in the IndoGangetic Plain: characterization, sources and temporal variability. Journal of Geophysical Research. http://dx.doi.org/ 10.1029/2010JD014188. Rengarajan, R., Sarin, M.M., Sudheer, A.K., 2007. Carbonaceous and inorganic species in atmospheric aerosols during wintertime over urban and high-altitude sites in North India. Journal of Geophysical Research 112, D21307. Rizzo, M.J., Scheff, P.A., 2007. Utilizing the chemical mass balance and positive matrix factorization models to determine influencial species and examine possible retations in receptor modelling results. Atmospheric Environment 41, 6987–6998. Saud, T., Gautam, R., Mandal, T.K., Gadi, R., Singh, D.P., Sharma, S.K., Dahiya, M., Saxena, M., 2012. Emission estimates of organic and elemental carbon from household biomass fuel used over the Indo Gangetic Plain (IGP), India. Atmospheric Environment 61, 212–220. Schwartz, J., Dockery, D.W., Neas, L.M., 1996. Is daily mortality associated specifically with fine particle? JAPCA 46, 927–939. Seinfeld, J.H., 1986. Air pollution. New York: John Wiley. 738 p. Sharma, S.K., Datta, A., Saud, T., Mandal, T.K., Ahammed, Y.N., Arya, B.C., Tiwari, M.K., 2010. Study on concentration of ambient NH3 and interactions with some other ambient trace gases. Environmental Monitoring and Assessment 162, 225–235. Sharma, M., Kishore, S., Tripathi, S.N., Behra, S.N., 2007a. Role of atmospheric ammonia in the formation of inorganic secondary particulate matter: a study at Kanpur, India. Journal of Atmospheric Chemistry 58, 1–17. Sharma, H., Jain, V.K., Khan, Z.H., 2007b. Characterization and source identification of polycyclic aromatic hydrocarbons (PAHs) in the urban environment of Delhi. Chemosphere 66, 302–310. Sharma, S.K., Kumar, M., Rohtash, Gupta, N.C., Saraswati, Saxena, M., Mandal, T.K., 2013. Characteristics of ambient ammonia over Delhi, India. Meteorology and Atmospheric Physics. http://dx.doi.org/10.1007/s00703-013-0299-8. Sharma, M., McBean, E.A., Ghosh, U., 1995. Prediction of atmospheric sulphate deposition at sensitive receptors in Northern India. Atmospheric Chemistry 29, 2157–2162. Sharma, D.N., Sawant, A.A., Uma, R., Cocker III, D.R., 2003. Preliminary chemical characterization of particle-phase organic compounds in New Delhi, India. Atmospheric Environment 37, 4317–4323. Sharma, S.K., Singh, A.K., Saud, T., Mandal, T.K., Saxena, M., Singh, S., Ghosh, S., Raha, S., 2012. Study on water soluble ionic composition of PM10 and trace gases over Bay of Bengal during WICARB campaign. Meteorology and Atmospheric Physics 118, 37–51. Shridhar, V., Khillare, P.S., Agarwal, T., Ray, S., 2010. Metallic species in ambient particulate matter at rural and urban location of Delhi. Journal of Hazardous Materials 175, 600–607. Sievering, H., Boatman, J., Galloway, J., Keene, W., Kim, Y., Luria, M., Ray, J., 1991. Heterogeneous sulfur conversion in sea-salt aerosol particles: the role of aerosol water content and size distribution. Atmospheric Environment 25, 1479–1487. Song, Y., Zhang, Y., Xie, S., Zeng, L., Zheng, M., Salmon, L.G., Shao, M., Slanina, S., 2006. Source apportionment of PM2.5 in Beijing by positive matrix factorization. Atmospheric Environment 40 (1), 526–1537. Srivastava, A., Jain, V.K., 2007. Size distribution and source identification of total suspended particulate matter and associated heavy metals in the urban atmosphere of Delhi. Chemosphere 68, 579–589. Tare, V., Tripathi, S.N., Chinnam, N., Srivastava, A.K., Dey, S., Manar, M., Kanawade, V.P., Aggarwal, A., Kishore, L., Lal, R.B., Sharma, M., 2006. Measurement of atmospheric parameters during Indian Space Research Organization Geosphere Biosphere Program Land Campaign II at a Typical Location in the Ganga Basin: 2. Chemical Properties. Journal Geophysics Research 111, D23210. http://dx.doi.org/10.1029/2006/JD007279. Tauler, R., Viana, M., Querol, X., Alastuey, A., Flight, R.M., Wentzell, P.D., Hopke, P.K., 2009. Comparison of the results obtained by four receptor modelling methods in aerosol source apportionment studies. Atmospheric Environment 43, 3989–3997. Taylor, S.R., McLennan, S.H., 1985. The Continental Crust: Its Composition and Evolution. Blackwell, Oxford, p. 312.

Please cite this article in press as: Sharma, S.K., et al. Source apportionment of PM10 by using positive matrix factorization at an urban site of Delhi, India. Urban Climate (2013), http://dx.doi.org/10.1016/j.uclim.2013.11.002

S.K. Sharma et al. / Urban Climate xxx (2013) xxx–xxx

15

Thurston, G.D., Spengler, J.D., 1985. A quantitative assessment of source contributions to inhalable particulate matter pollution in metropolitan Boston. Atmospheric Environment 19, 9–25. Tiwari, S., Srivastava, A.K., Bisht, D.S., Bano, T., Singh, S., Behura, S., Srivastava, M.K., Chate, D.M., Padmanabhamurthy, B., 2009. Black carbon and chemical characteristics of PM10 and PM2.5 at an urban site of North India. Journal of Atmospheric Chemistry 62, 193–209. USEPA, 2008. EPA Positive matrix Factorization (PMF) 3.0 Fundamentals and User Guide, USEPA Office of Research and Development. Vallius, M., Janssen, N.A.H., Heirich, J., Hoek, G., Ruuskanen, J., Cyrys, J., Grieken, R.V., Hartog, J.J., Kreyling, W.G., Pekkanen, J., 2005. Sources and elemental composition of ambient PM2.5 in three European cities. Science of Total Environment 337, 147– 162. Venkataraman, C., Rao, G.U.M., 2002. Emission factors of carbon monoxide and size-resolved aerosols from biofuel combustion. Science of Total Environment 35, 2100–2107. Wu, C.F., Larson, T.V., Wu, S.Y., Williamson, J., Westberg, H.H., Liu, L.J.S., 2007. Source apportionment of PM2.5 and selected hazardous air pollutants in Seattle. Science of Total Environment 386, 42–52. Zeng, F., Shi, G.L., Li, X., Feng, Y.C., Bi, X.H., Wu, J.H., Xue, Y.H., 2010. Application of a combined model to study the source apportionment of PM10 in Taiyuan, China. Aerosol and Air Quality Research 10, 177–184. Zhang, Q., Jimenez, J., Canagaratna, M.R., Ulbrich, I.M., Ng, N.L., Worsnop, D.R., Sun, Y., 2011. Understanding atmospheric organic aerosols via factor analysis of aerosol mass spectrometry: a review. Analytical and Bioanalytical chemistry 401, 3045–3067.

Please cite this article in press as: Sharma, S.K., et al. Source apportionment of PM10 by using positive matrix factorization at an urban site of Delhi, India. Urban Climate (2013), http://dx.doi.org/10.1016/j.uclim.2013.11.002