Introduction

The Indo Gangetic Plain (IGP) is the major source of pollutants over India and Indian subcontinents due to the combined effect of the usage of fossil fuels and biomass fuels as energy, rapid urbanization, and industrialization (Tare et al. 2006; Ram and Sarin 2011; Saud et al. 2013; Sharma et al. 2014a, b, c). It is well documented that atmospheric particulate matter (PM) significantly affects atmospheric chemistry, ambient air quality, visibility, climate change, cloud formation, and the Earth’s radiation budget (Ramgolam et al. 2009; Pope et al. 2009). Several studies have revealed that aerosols can lead to serious human health effects like cardiovascular and respiratory disorders (Pope and Dockery 2006). Generally, coarse mode PM are natural while fine-mode PM are produced due to chemical conversion such as the gas to particle conversion, combustion of fuels in vehicles, or high industrial processes (Sharma et al. 2015; Perrino et al. 2011). Therefore, chemical characterization and identification of PM sources are necessary to develop air quality improvement strategies in order to be able to control and reduce ambient PM concentrations through targeted action (Waked et al. 2014).

Particulate matter has been extensively studied in recent years over the continents including IGP region due to its potential impacts on human health and air quality (Schwartz et al. 1996; Li et al. 2009; Ram et al. 2012; Sharma et al. 1995; Sharma et al. 2014a, b; Sen et al. 2014). The previous studies (Ram and Sarin 2010; Ram et al. 2012; Sharma et al. 2014a, b) demonstrated that the soil dust, agricultural activities, biomass burning, vehicular emission, and industries, are responsible for the formation of secondary inorganic particles and other anthropogenic sources over IGP, India. Most of the studies on chemical characteristics of aerosol over IGP region are reported based on single location experiment/observation; long-term and simultaneous observations of chemical properties of PM over several locations of IGP are limited. The variation in inorganic secondary particle is of greater significance from the human health point of view and need to be examined (Liousse et al. 1996; Jacobson 2001; Sharma et al. 2007; Li et al. 2009; Bond et al. 2013).

Earlier studies have concentrated at a particular location only, rather representing regional features of whole IGP region. Present study documents the spatial and temporal variability of chemical composition of ambient aerosols through ground-based measurements over the whole IGP region. Keeping in view of the competitive importance and sensitivity of role of seasonal wind flow patterns, and production from local sources, chemical properties of aerosol over IGP, India, have also been discussed. The paper represents the simultaneous sampling of PM10 mass and analysis of chemical compositions (organic carbon (OC), elemental carbon (EC), and water-soluble inorganic ionic component (WSIC)) over three sites of IGP, India, (Delhi, Varanasi, and Kolkata) stretching from upper IGP region to lower IGP region for the whole year of 2011.

Methodology

Sites description

The observational sites are chosen to represent the characteristics of upper (Delhi), middle (Varanasi), and lower (Kolkata) IGP region of India. Saud et al. (2011, 2013) represent that IGP as a whole is a major emitter of pollutants, although emission density varies from spatially over IGP region. Therefore, three sites may represent the combination of local sources as well as the transported features of particulate matter. The description of the study sites are as follows: PM10 samples were collected at CSIR-National Physical Laboratory (28° 38′ N, 77° 10′ E; 218 m amsl), New Delhi, India (Fig. 1). The sampling site represents a typical urban atmosphere, surrounded by huge roadside traffic and agricultural fields in the southwest direction. During 2010–2011, the total number of registered vehicles in Delhi were approximately 6.35 million (Delhi Statistical Handbook 2012). This area is under the influence of free airmass flow from northeast to northwest in winter and from southeast to southwest in the summer (Goyal and Sidhartha 2002; Sharma et al. 2014a). During winter, Delhi experiences severe fog and haze weather conditions and poor visibility. During summer, the occasional occurrence of dust storm contributes the presence of mineral dust significantly to the aerosol loading (Ram et al. 2010). The ambient temperature of sampling site of Delhi varies from minimum (1 °C) in winter (November to February) to maximum (48 °C) in summer (March to June). During monsoon (July to October), the average rainfall at the sampling site is recorded as 780 mm.

Fig. 1
figure 1

Map of observational sites (Delhi, Varanasi, and Kolkata), IGP, India

Full size image

The sampling site, Varanasi (25° 18′ N, 83° 03′ E; 129 m amsl) located in the middle Ganges valley of North India (Fig. 1) represent an urban environment. Being located in the IGP of North India, the land is very fertile due to low-level floods in the Ganges continually replenishing the soil. Varanasi experiences a humid subtropical climate with large variations between summer and winter temperatures. The temperature of the sampling site ranges between 22 and 46 °C in the summer. During winter, Varanasi faces very large diurnal variations with warm days and downright cold nights. Cold waves from the Himalayan region cause temperatures to dip across the city in the winter from December to February, and temperature below 5 °C is common. The average annual rainfall is recorded as 1110 mm during 2011.

The sampling site, Kolkata, a megacity (22° 33′ N and 88° 20′ E; 9 m amsl), represents a typical urban environment located in the lower IGP influenced by upwind aerosol. The site is bounded to the west and northwest by Houghly river spread along 80 km. Sundarban mangroves forest, the world’s largest mangrove eco-region at the land-ocean boundary of the Ganges delta, is situated about 100 km from Kolkata. The average temperature, RH, and wind speed of the sampling site during study recorded as 23 °C, 74 %, and 2.1 m s−1, respectively.

Aerosol sampling

PM10 samples were collected (4–5 samples/month) at Delhi (n = 50), Varanasi (n = 45), and Kolkata (n = 54) of IGP, India, on precombusted (at 550 °C for 5 h to eradicate organic impurities and kept in desiccators for 24 h prior to sample collection) Whatman Quartz Microfibre filters (QM-A) at a 10-m height (AGL) during January to December 2011. Particle samplers (model: PEM-RDS 8NL, Make: M/s. Polltech Instruments, Mumbai, India) were used to collect the PM10 samples at all the locations. The blower flow rates of the samplers were calibrated (with the accuracy of ±1 % of full scale) with top loading orifice calibrator traceable to national standard. The flow meters of the samplers were also calibrated (with the accuracy of ±2 % of full scale) with air flow calibrator traceable to national standard. Ambient air was passed through QM-A filters (size 20 × 25 cm2) at a flow rate of 1.13 m3 min−1 (accuracy ±2 %) for 8 h on a day and night basis [during the daytime (1000–1800 h) and nighttime (1900–0300 h)]. In general, rush hour started from 0900 to 1100 hours and 1700 to 1900 hours in Delhi (Sharma et al. 2010); hence, the sampling had been started at 1000 hours (at daytime) and 1900 hours (at nighttime). The sample has been collected for 8 h also to follow the common protocol (i.e., 8 or 12 or 24 h) of Central Pollution Control Board (CPCB), New Delhi, India. The average concentration of day and night sampling of PM10 was reported in the paper. The sample filters were weighed before and after the sampling during the experiment in order to determine the mass of the PM10 collected. The concentration of PM10 (μg m−3) was calculated on the basis of the difference between final and initial weights of the QM-A filters (measured by a microbalance; resolution ±10 μg) and divided by the total volume of air passed during the sampling. After collecting samples, filters were stored under dry condition at −20 °C in deep-freezer prior to analysis.

Analysis

OC and EC analysis

Analysis of OC and EC of ambient PM10 samples was performed using an OC/EC carbon analyzer (model: DRI 2001 A, Atmoslytic Inc., Calabasas, CA, USA) following the USEPA Method (Improve Protocol) with negative pyrolysis areas zeroed (Chow et al. 2004). The detail principle and analytical procedure for analysis of OC and EC species of PM10 mass are described in Chow et al. (2004). Approximately 0.536 cm2 area of QM-A filter was cut using the proper punch, and the values are reported as μg cm−2 as given by the instrumental analysis software. Each filter was analyzed triplicate with blank filters (blank filters analyzed) to get the representative estimation of OC and EC concentration in PM10. To estimate the accurate OC and EC concentrations, the OC and EC values of blank filters were estimated and subtracted from the sample filters. The analytical error (repeatability) was estimated to be 3–5 % based on triplicate analysis of each filters.

Water-soluble inorganic ionic components analysis

PM10 samples were extracted in deionized water having conductivity >18.2 MΩ for 90 min in ultrasonic extractor for the determination of WSIC (Na+, NH4 +, K+, Ca2+, Mg2+ F−, Cl−, NO3 −, and SO4 2−). The WSIC of PM10 has been analyzed by ion chromatograph (model: DIONEX-ICS-3000, Sunnyvale, CA, USA). The concentrations of F−, Cl−, NO3 −, and SO4 2− were determined by 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 triple-distilled water as regenerator. Li+, Na+, NH4 +, K+, Ca2+, and Mg2+ were determined by using a separation column with a guard column, suppressor CSRS-300 and 5 mM MSA (methane sulfonic acid) as eluent. Calibration standards have been prepared by the National Institute of Standards and Technology (NIST, USA) traceable certified standards for calibration of IC (Sharma et al. 2014d). The blank filters were also analyzed for cations (Li+, Na+, NH4 +, K+, Ca2+, and Mg2+) and anions (F−, Cl−, NO3 −, and SO4 2−). The analytical error (repeatability) was estimated to be 3–5 % based on triplicate analysis of each filters (Sharma et al. 2012a).

Meteorological parameters and trajectory analysis

Meteorological parameters such as temperature (accuracy ±1 °C), relative humidity (RH) (accuracy ± 2 %), wind direction (accuracy ±3°), and wind speed (accuracy ±2 % of full scale) were collected at every 5-min interval using automatic weather station (AWS) at all the sampling sites. In order to identify the possible transport pathways of PM10 from their potential sources of origin to Delhi, Varanasi, and Kolkata of IGP, India, 5-day backward trajectory was calculated using the Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT) model and has been traced (Draxler and Rolph 2003). Air mass back-trajectories for each experimental day for 500 m above the ground level (AGL) during January to December 2011 have been calculated using Global Data Assimilation System (GDAS) meteorological data. HYSPLIT was run every day starting at 0500 hours, UTC, at a starting height of 500 m AGL on an hourly basis for each site. This height was chosen to diminish the effects of surface friction and to represent winds in the low boundary layer. The PM10 has the ability to travel long distance; therefore, 5 days were selected to calculate backward trajectories using HYSPLIT.

Principal component analysis

Principal component analysis (PCA) was performed to identify the possible sources of PM10 mass over Delhi, Varanasi, and Kolkata of IGP, India. In the present study, the orthogonal transformation method with Varimax rotation was employed. The lowest Eigenvalue for extracted factors was restricted to more than 1.0. Total 11 constituents of PM10 were used as variable in the dataset. In PCA, the chemical data are transformed into a dimensionless standardized form:

$$ Zij=\frac{Cij-\overline{Cj}}{\sigma j} $$
(1)

where i = 1, …, n samples; j = 1, …m elements; Cij is the concentration of element j in sample i; and \( \overline{Cj} \) and σj are the arithmetic mean concentration and the standard deviation for element j, respectively. The PCA model is expressed as

$$ {\mathrm{Z}}_{ij}={\displaystyle {\sum}_{k=1}^p{g}_{ik}{h}_{kj}} $$
(2)

where k = 1, …, p sources, and g ik and h kj are the factor loadings and the factor scores, respectively. This equation is solved by Eigenvector decomposition (Song et al. 2006).

Results and discussion

Mass concentration

In the previous section, we have mentioned that the observational sites are chosen to represent the characteristics of upper (Delhi), middle (Varanasi), and lower (Kolkata) IGP regions; therefore, in the following sections, we will discuss the results of individual stations. The annual average mass concentration of PM10 over Delhi (range 63.3–354.7 μg m−3), Varanasi (range 53.3–392.1 μg m−3), and Kolkata (range 92.9–382.6 μg m−3) was recorded as 202.3 ± 74.3, 206.2 ± 77.4, and 171.5 ± 38.5 μg m−3, respectively (Table 1), indicating the events of higher mass concentration over Varanasi as well as large variation among the three observational sites of IGP, India. Higher mass concentration of PM10 in the middle IGP indicates combined influences of the accumulation of fine particulates advected from the upper IGP region and local sources (transport for usage of adulterated oil, brick kiln, biomass burning, etc.). Combination of PCA analysis and trajectory analysis may provide better insight of the higher mass concentration of PM10 mass. The temporal variation of PM10 concentration over IGP region is also depicted in Fig. 2.

Table 1 The average concentrations of particulates (PM10), EC, OC, and WSIC of PM10 (μg m−3) over IGP, India
Full size table
Fig. 2
figure 2

Temporal variability in mass concentrations of PM10 and OC and EC of PM10 at urban sites of Delhi, Varanasi, and Kolkata

Full size image

Figure 3 shows the monthly average (average ± SD) of mass concentrations of PM10, OC, and EC over Delhi, Varanasi, and Kolkata of IGP, India, during 2011. Delhi and Varanasi have recorded almost similar order of magnitude of monthly average concentration of PM10 (Fig. 3) as 333.6 and 366.6 μg m−3, respectively, in the month of December (winter), whereas the highest concentration of Kolkata was recorded as 244.3 μg m−3 in the month of January. All the three sites, viz., Delhi, Varanasi, and Kolkata, have recorded the lowest monthly average value as 101.9 μg m−3 (July), 69.5 μg m−3 (August), and 119.5 μg m−3 (June), respectively, during monsoon. Average day and night ratio of PM10 over Delhi, Varanasi, and Kolkata were recorded as 1.01, 1.00, and 0.94 during the study (Fig. S1; in supplementary information). Several studies on PM10 concentration over individual stations of IGP are reported earlier, e.g., Tiwari et al. (2012) have reported the annual average concentration of PM10 mass as 219 ± 84 μg m−3 during 2007, whereas Sharma et al. (2014b) reported annual PM10 mass concentration as 213.1 ± 15.0 μg m−3 over Delhi during 2010. In another study, Sharma et al. (2014a) reported annual PM10 concentration over Delhi to be 191.4 ± 45.5 μg m−3 for the period 2010–2011. Pandey et al. (2013) also reported the PM10 mass concentrations over Lucknow during 2007 to 2009 of the order of 168.1 ± 29.1 μg m−3 (range 102.3–240.5 μg m−3). Murari et al. (2014) reported the annual average concentration of PM10 as 176.1 ± 85.0 μg m−3 during 2013–2014 at Varanasi. Das et al. (2006) have reported average PM10 mass concentration to be 303.85 ± 49.48 μg m−3 over Kolkata, whereas Gupta et al. (2007) reported as 288.93 ± 123.52 μg m−3. The highest concentration of PM10 mass during the winter months over IGP may be due to the combined effect of source strength and lower boundary layer height (Datta et al. 2010; Ram and Sarin 2010; Sharma et al. 2015).

Fig. 3
figure 3

Monthly average variations in mass concentration of PM10, OC, and EC (μg m−3) over Delhi, Varanasi, and Kolkata of IGP, India, during 2011

Full size image

The mass concentration of PM10 during winter, summer, monsoon, and its seasonal differences over Delhi, Varanasi, and Kolkata of IGP, India, are also summarized in Table 1. The average mass concentration of PM10 was recorded as 263.7 μg m−3 (winter), 209.2 μg m−3 (summer), and 133.8 μg m−3 (monsoon) over Delhi. The higher average PM10 mass was recorded at all the observational sites of IGP, India, during winter, when low-level northeast wind flow carries large amount of pollutants along with as well as lower boundary layer caps pollutant increasing the average burden, whereas lower values were observed during monsoon, except Kolkata (Table 1). Large amount of sea salt might have increased the concentration of PM10 pouring into the atmosphere over Kolkata. This may be reconfirmed from the inorganic component of the chemical composition. In the present case, the significant seasonal variations in concentration of PM10 were observed at all the sites.

Chemical composition

Chemical compositions (OC, EC, and WSIC) of PM10 samples collected over Delhi, Varanasi, and Kolkata are summarized in Table 1 with seasonal changes. Figures 3 and 4 show the monthly average variations of mass concentrations and fractions of OC, EC, and WSIC of PM10 over IGP, India, during 2011. The detail descriptions of chemical characteristics of PM10 over the region are described in the next section.

Fig. 4
figure 4

Monthly average variations in mass concentration of WSIC (μg m−3) of PM10 over Delhi, Varanasi, and Kolkata of IGP, India, during 2011

Full size image

Time series of PM10, OC, and EC concentrations over Delhi, Varanasi, and Kolkata during January–December 2011 are depicted in Fig. 2. Over Delhi, the highest concentration of PM10 mass was recorded as 354.7 μg m−3 during January, whereas minimum concentration of PM10 mass was recorded as 63.3 μg m−3 during August. Over Varanasi, the highest concentration of PM10 mass was recorded as 392.1 μg m−3 during December, whereas minimum concentration of PM10 mass was recorded as 53.3 μg m−3 during June (Fig. 2). Unlike other two stations, the highest concentration of PM10 mass (as 382.6 μg m−3) was recorded over Kolkata during January the period of transition between monsoon and winter season. WSIC of PM10 also followed the similar temporal variation over the IGP region.

It has been noticed that concentration of OC, EC, and total carbon (TC = OC + EC) has followed the pattern of variation of PM10 mass concentration with maximum concentration broadly in winter (December–February) over the three stations. The monthly variations in chemical composition of PM10 at observational sites of IGP indicate that the (Figs. 3 and 4) higher average concentration of OC in PM10 (41.9 μg m−3) was observed in the month of December in Delhi, followed by Varanasi (38.5 μg m−3), whereas Kolkata recorded the highest value (25.9 μg m−3) in the month of February. The EC has followed similar pattern with the maximum concentration in the month of December at Delhi (20.9 μg m−3), followed by Kolkata (20.2 μg m−3), whereas Varanasi has recorded the highest value (19.1 μg m−3) in the month of February. It is to be reminded that total carbon (TC) content accounts for ~16 % of PM10 mass at Delhi during 2011 following the pattern of OC and EC with higher concentration during December (winter). During 2010, the total carbon (TC) had been accounted for ~18 % of PM10 mass at Delhi (Sharma et al. 2014c). Concentrations of OC, EC, and TC have varied with the increase of aerosol loading.

Organic carbon and elemental carbon

Over Delhi, annual concentration of OC has been varied from 10.4 to 50.1 μg m−3 with an average value of 22.7 ± 7.3 μg m−3 (~11 % of PM10 mass), whereas mass concentration of EC has varied from 1.7 to 29.4 μg m−3 with an average value of 8.7 ± 4.0 μg m−3 (~5 % of PM10 mass). The average concentration of OC has been recorded as 16.9 ± 8.6 μg m−3 (~8 % of PM10 mass) and 13.7 ± 5.3 μg m−3 (~9 % of PM10 mass) at Varanasi and Kolkata, respectively, whereas average concentration of EC was observed as 7.3 ± 5.1 μg m−3 (~5 % of PM10 mass) and 7.8 ± 4.0 μg m−3 (~4 % of PM10 mass) over Varanasi and Kolkata, respectively (Table 1).

It is to be reminded that TC content accounts for ~16 % of PM10 mass at Delhi during 2011 following the pattern of OC and EC with higher concentration during December (winter). During 2010, the total carbon (TC) had been accounted for ~18 % of PM10 mass at Delhi (Sharma et al. 2014c). Ram and Sarin (2010) have reported that total carbonaceous aerosol (TCA = OM + EC) accounts for ~30–35 % of the total suspended particulate (TSP) mass at urban and rural sites of northern India (whereas in the present study, it is accounted for ~29 % of PM10 mass). Over an urban site (Beijing) in China, Guinot et al. (2007) have reported that 46 % of fine aerosol (<2 μm) mass is composed of carbonaceous aerosols. Recently, Mandal et al. (2014) had reported very higher annual average concentrations of OC (93.0 ± 44.7 μg m−3), EC (27.3 ± 13.4 μg m−3), and total carbonaceous aerosols (176.1 ± 84.7 μg m−3; ~66 % of PM10 mass) in PM10 (average 280.7 ± 126.1 μg m−3) mass at an industrial area of Delhi, India.

In the present case, the average day and night ratio of OC were recorded as 0.79, 0.85, and 0.90 over Delhi, Varanasi, and Kolkata, respectively, whereas the average day and night ratio of EC were recorded as 0.60, 0.73, and 0.68 at Delhi, Varanasi, and Kolkata, respectively (Fig. S1; in supplementary information). Ram and Sarin (2011) also reported the similar type of observations at Kanpur, a site in the middle IGP, India, and attributed this increment in mass concentration of PM10 during nighttime at the observational site to increase in the source strength of cabonaceous species and trapping of aerosols due to lowering of boundary layer height, etc. (Sharma et al. 2014b). Abundance of OC and EC concentration shows a linear increment with aerosol mass of PM10 at all the locations (Fig. 5).

Fig. 5
figure 5

Scatter plots between OC and EC, OC and PM10, and EC and PM10 at Delhi, Varanasi, and Kolkata of IGP, India, during 2011

Full size image

The monthly variations in chemical composition of PM10 at observational sites of IGP indicate that the (Figs. 3 and 4) higher average concentration of OC in PM10 (41.9 μg m−3) was observed in the month of December in Delhi, followed by Varanasi (38.5 μg m−3), whereas, Kolkata recorded the highest value (25.9 μg m−3) in the month of February. The EC has followed similar pattern with the maximum concentration in the month of December at Delhi (20.9 μg m−3), followed by Kolkata (20.2 μg m−3), whereas Varanasi has recorded the highest value (19.1 μg m−3) in the month of February.

Water-soluble inorganic ionic components

The annual average and seasonal variation in concentrations of WSIC are summarized in Table 1 with seasonal difference over Delhi, Varanasi, and Kolkata of IGP, India. Over Delhi, among WSIC, SO4 2− (11.6 μg m−3) reports the highest concentration followed by other ionic compositions, e.g., NO3 − (10.5 μg m−3) > NH4 + (5.1 μg m−3) > Cl− (4.5 μg m−3). Although Varanasi has encountered SO4 2− (10.7 μg m−3) as major component among WSIC, Kolkata has recorded SO4 2− (10.0 μg m−3) as a major component, followed by Cl−, NO3 −, NH4 +, and Na+ etc., (Table 1). The presence of Cl− and SO4 2− as a major ionic component over Kolkata may be due to the coastal region (Kolkata situated near the Bay of Bengal).

The monthly average variation in WSIC of PM10 mass over IGP, India, during 2011 is shown in Fig. 4. In the present study, the WSIC accounted for ~21, ~19, and ~24 % of PM10 mass at Delhi, Varanasi, and Kolkata, respectively, of IGI, India. Sharma et al. (2014b) reported the significant seasonal variation in concentration of particulate SO4 2−, NO3 −, and NH4 + over Delhi. The assimilation of HNO3 into NH4 + aerosols generally depends on the amount of H2SO4 used in the neutralization with NH3 (Sharma et al. 2012, 2014b). Thus, the samples containing lower SO4 2− and higher NO3 − concentration could be representative of local biogenic NOx emission leading to formation of NO3 − or marine masses having high Cl− concentrations whereas the sample containing higher SO4 2− and lower NO3 − concentration could be representative of aerosols transported over a long distance or precursors from a continental air mass.

The result reveals that the molar ratio of NH4 + positively correlated with SO4 2− and NO3 − at Delhi Varanasi and Kolkata during winter (Figs. S5, S6, and S7; in supplementary information). However, molar ratio of NH4 + shows nonsignificant positive correlation with molar ratio of SO4 2− and NO3 − during monsoon at all locations. The assimilation of HNO3 into NH4 + aerosols generally depends on the amount of H2SO4 used in neutralization with NH3. Thus, the samples containing lower SO4 2− and higher NO3 − concentration could be representative of local biogenic NOx emission leading to formation of NO3 − or marine masses having high Cl− concentrations whereas the sample containing higher SO4 2− and lower NO3 − concentration could be representative of aerosols transported over a long distance or precursors from a continental air mass, whereas NH4 + generally combines with NO3 − and SO4 2− in atmosphere and forms NH4NO3 and (NH4)2SO4, respectively (Utsunomiya and Wakamatsu 1996; Behera and Sharma 2010). In (NH4)2SO4, the molar ratio of NH4 + to SO4 2− is 2:1. If the observed molar ratio is greater than 2, it signifies that excess NH4 + is present and possible may be combined with NO3 − or other ions. In the present study, the equivalent molar ratio of NH4 +/SO4 2− computed as 2.73, 2.08, and 3.96 at Delhi, Varanasi, and Kolkata, respectively, which indicate the complete neutralization of atmospheric acid gas H2SO4 and predominant to aerosol formation during the study period. Similarly, the equivalent molar ratio of NH4 +/NO3 − i.e., 1.95 (Delhi), 1.54 (Varanasi), and 2.62 (Kolkata) indicate the formation of NH4NO3.

During the study period, SO4 2− and NO3 − ions also show the significant correlation with NH4 + indicating formation of (NH4)2SO4 and NH4NO3 salt (Sharma et al. 2012a, b). The remainder of HNO3 (after neutralization with NH3) and/or H2SO4 may react with mineral dust to form Ca(NO3)2 and/or CaSO4 in the coarse mode (Kholder and Hassan 2008; Guo et al. 2010; Rastogi and Sarin 2006). Significant correlation of NO3 − with Ca2+ in PM10 samples indicates the possible association of NO3 − with Ca2+ in the coarse mode. Thus, particulate-NH4 + is the major neutralizing agent of the acidic species (SO4 2− and NO3 −) in PM10, whereas Ca2+ plays an important role in the neutralizing of the acidic species in the coarse mode aerosol over IGP (Ram and Sarin 2010).

Source apportionment

Figure 5 shows the scatter plot between OC and EC at Delhi (R 2 = 0.797; at P < 0.05), Varanasi (R 2 = 0.862; at P < 0.05), and Kolkata (R 2 = 0.833; at P < 0.05) during the study period (January–December 2011), whereas Figs. S2, S3, and S4 (in supplementary information) depicted on seasonality. A significant correlation between OC and EC is usually indicative of their common sources like vehicular traffic (Salma et al. 2004). In contrast, a poor correlation between OC and EC indicates the formation of secondary aerosol under favorable conditions for the gas to particle conversion of VOCs through a photochemical reaction in the atmosphere. Overall, a positive linear trend (R 2 = 0.797 for Delhi; R 2 = 0.862 for Varanasi; R 2 = 0.833 for Kolkata) is observed between OC and EC for at urban sites especially when OC concentrations are >50 μg m−3, and it indicates the influence of vehicular emission (Salma et al. 2004). The scatter plots between OC and EC shows that during winter, summer, and monsoon over Delhi, Varanasi, and Kolkata of IGP, India, are significantly correlated and indicated the common sources of OC and EC over the region (Figs. :2, S3, and S4; in supplementary information).

In the present study, the annual ratio of OC/EC, K+/OC, K+/EC, Cl−/EC, and SO4 2−/EC of PM10 mass is recorded as 2.61, 0.08, 0.19, 0.47, and 1.15, respectively, over Delhi, whereas it was 1.75, 0.21, 0.35, 1.10, and 1.10, respectively, over Kolkata. Average values of K+/EC and Cl−/EC ratios are 0.19 and 0.47, respectively, over Delhi, which may give indication of presence of the biomass burning as well as fossil fuel combustion as a source of the PM10 mass at the observational site of Delhi (Yamasoe et al. 2000; Ferek et al. 1998; Saud et al. 2012).

The average OC/EC ratios of the present study over Delhi (2.61), Varanasi (2.31), and Kolkata (1.75) are relatively lower than those reported for the urban locations in China (Cao et al. 2003; Ho et al. 2007) and matches with earlier observations at few urban locations of the IGP (Table S1; in supplementary information). Such difference in the OC/EC ratio might depend on the differences in emission sources of carbonaceous aerosols (Ram and Sarin 2011; Sharma et al. 2014c). The OC and EC concentrations in present study sites might be attributed to the combined effects of traffic emission, biomass burning, wood burning, and crop residue burning. Analysis of stable carbon and nitrogen isotopic composition of PM10 mass indicated that the vehicular emissions, biomass burning, and secondary aerosols are the major source of PM10 mass over IGP, India (Sharma et al. 2015). Figure 5 shows the scatter plot between OC and EC during the study period for Delhi (R 2 = 0.797), Varanasi (R 2 = 0.862), and Kolkata (R 2 = 0.833). Ram and Sarin (2011) reported a significant linear relation between OC and EC (r 2 = 0.66) and OC and K+ (r 2 = 0.59) of PM10 at Kanpur, IGP of India and attributed to biomass burning emissions which is one of the sources of PM10 over IGP India. A significant correlation between OC and EC is usually indicative of their common sources like vehicular traffic, biomass burning, etc. (Salma et al. 2004, Sharma et al. 2014c). Potassium (K) is generally used as a tracer of crustal dust in the coarse range and soluble K+ for biomass burning in the fine range of PM (Sharma et al. 2014a). Soluble K+ has been used as a key elemental marker for biomass/wood combustion for TSP, PM10, and PM2.5 (Khare and Baruah 2010; Shridhar et al. 2010). K+ and NH4 + have also been used as markers for wood burning and agricultural activities (Khare and Baruah 2010). Fossil fuels could be the dominant sources of OC and EC in mega-cities like Delhi and Kolkata. Comparison of OC/EC, K+/OC, K+/EC, Cl−/EC, and SO4 2−/EC ratios in the particulates with present study and at different sites in India and China along with ratios for Savanna burning and agricultural waste burning are summarized in Table S1 (in supplementary information). In the above section, we have seen that variation of PM10 and their chemical composition (OC, EC, and WSIC) over three IGP regions, indicating the role of several sources. PCA and trajectory analysis may support and consolidate the idea depicted in the above section.

PCA was performed by the Varimax-rotated factor matrix method of SPSS for factor analysis on PM10 constituents collected over Delhi, Varanasi, and Kolkata of IGP, India. Eleven constituents of PM10 (OC, EC, Na+, K+, NH4 +, Ca2+, Mg2+, F−, Cl−, NO3 −, SO4 2) were used as variables in the dataset at Delhi. Total 11 components were obtained, out of which three were extracted as principle components (Eigenvalue >1) that explained the 76 % of the variance of the data at Delhi. Three components corresponds to the different sources that were identified as secondary aerosol, biomass burning, fossil fuel combustion + vehicular emission, based on the loading of the variables in the factor (Table 2). The first factor is responsible for 41.14 % of the variance and high loading NH4 +, NO3 −, and SO4 2− attributed to secondary aerosol whereas second factor attributed to biomass burning sources (OC, EC, and K+). The third component indicated the combination of fossil fuel combustion + vehicular emission (~9.5 % of the total variance) exhibits high loading of EC, OC, and other metals.

Table 2 PCA factor loadings for PM10 collected over Delhi, Varanasi, and Kolkata of IGP, India
Full size table

PCA analysis also demonstrated the three factors to the different sources that were identified as secondary aerosol, biomass burning, vehicular emission + sea salt, and secondary aerosol based on the loading of the variables in the factor (Table 2) at Varanasi. The first factor is resolved 40.75 % of the variance and high loading OC, EC, and K+ attributed to biomass burning whereas second factor attributed to vehicular emission + sea salt (EC, OC, K+, Cl−, Na+, Mg2+). The third factor attributed to secondary aerosol due high loading of NH4 +, NO3 −, and SO4 2− in PM10 mass over Varanasi (Table 2).

The total 11 components were obtained, out of which four factors were extracted (with 82.5 % of the variance of the data) of PM10 mass at Kolkata. The four factors were identified as sea salt, biomass burning + vehicular emission, secondary aerosol, and soil dust based on the loading of the variables in the factor (Table 2). The first factor have high loading of Na+ and Cl− attributed to sea salt aerosol (~37 % of the total variance). The second factor attributed to biomass burning + vehicular emission (OC, EC, and K+) whereas third component indicated secondary aerosol (Table 2).

Sharma et al. (2014a) reported soil dust (22.4 %), vehicle emissions (16.0 %), secondary aerosols (20.3 %), fossil fuel burning (16.6 %), biomass burning (11.8 %), industrial emissions (7.2 %), and sea salts (5.6 %) as the major sources of PM10 mass concentration at the observational site of Delhi using positive matrix factorization (PMF). Tiwari et al. (2009) reported that soil dust contributed 27.0 % of PM10 mass in Delhi whereas Khillare et al. (2004) estimated as 22.0 % soil dust of coarse particle at Delhi. Chelani et al. (2008) reported that soil dust and vehicular emission contributed 17.7 and 23.0 % of PM10 mass, respectively, at Mumbai whereas soil dust contributed 37.0 % of PM10 mass at Kolkata (Gupta et al. 2007).

In the above sections, we have observed the possible combined effect of biomass burning and fossil fuel combustion in the mass concentration of PM10 mass over three locations of IGP, India. However, the combined effect of biomass and fossil fuel largely depends on the local as well outflow of the long-distance source region. HYSPLIT has allowed for understanding the role of air mass parcel from long range transport at the receptor sites (Fig. 6). The approaching air mass at the receptor sites Delhi, Varanasi, and Kolkata during winter, summer, and monsoon seasons are depicted in Fig. 6. The approaching air mass at the receptor site Delhi is mainly continental type, being transported from IGP and its surroundings during winter and summer whereas air mass transported mainly from IGP and Arabian Sea through Thar Desert during monsoon (Sharma et al. 2014c). The approaching air mass at the receptor site of Varanasi is mainly from Punjab, Haryana, and Uttar Pradesh of IGP and surrounding areas during winter and summer whereas during monsoon, air mass parcel approaching mainly from Arabian Sea. At Kolkata the air mass is approaching the observational site from the IGP and surrounding areas during winter season. During summer and monsoon, air mass is approaching at the observational site of Kolkata mainly from coastal areas of the Bay of Bengal and Arabian Sea, respectively.

Fig. 6
figure 6

Air parcel back trajectory (using HYSPLIT) during winter, summer, and monsoon period (GDAS meteorological data) over Delhi, Varanasi, and Kolkata of IGP, India

Full size image

Conclusions

Source apportionment of simultaneously observed PM10 at three locations, viz., Delhi (upper IGP), Varanasi (middle IGP), and Kolkata (lower IGP) and its chemical properties (OC, EC, and WSIC) for the year 2011 provide following points:

  • During study period, higher average mass concentration of PM10 at Varanasi (206.2 μg m−3) followed by Delhi (202.3 μg m−3) and Kolkata (171.5 μg m−3) indicates combination of the growth of fine particulates on the way from upper IGP region, local sources (transport for usage of adulterated oil, brick kiln, biomass burning, etc.).

  • The higher average PM10 mass during winter are either due to capping by low-level boundary layer increasing the average burden (Delhi and Varanasi) or due to the movement of ITCZ when low-level northeast wind flow carries large amount of pollutants along with (Kolkata). Particulate matter has been washed out during heavy rain over three locations during monsoon. Over Kolkata, large marine sources (sea salt) might have increased the concentration of PM10 during monsoon.

  • It has been noticed that concentration of OC, EC, and TC has followed the pattern of variation of PM10 mass concentration with maximum concentration broadly in winter (December–February) over the three stations. The maximum average concentration of OC has been recorded at Delhi (23.57 μg m−3) and minimum at Kolkata (12.74 μg m−3).

  • OC/EC, K+/OC, K+/EC, Cl−/EC, and SO4 2−/EC ratios of PM10 indicate the role of different sources over three IGP regions. The average OC/EC ratios of the present study over Delhi (2.61), Varanasi (2.31), and Kolkata (1.75) are relatively lower than those reported for the urban locations in China and matches with earlier observations at few urban locations of the IGP

  • The positive linear relationship of molar ratios of SO4 2− and NO3 − with NH4 + indicates the possibility of the formation of secondary inorganic aerosol [(NH4)2SO4 and NH4NO3] over IGP region. Air mass flow at the observational sites indicate the continental and marine types PM10 transported from the Bay of Bengal, Pakistan, Thar Desert, IGP region and surrounding areas.

  • PCA analysis of PM10 quantifies that secondary aerosol, biomass burning, vehicular emission, and fossil fuel combustion are the major sources of PM10 at Delhi whereas biomass burning, secondary aerosol, vehicular emission, and sea salt are the major sources at Varanasi. At Kolkata, the major sources of PM10 mass resolve as sea salt, secondary aerosol, biomass burning, vehicular emission, and soil dust.