Introduction

The atmospheric environment over a region is an open system with respect to trace gas concentrations, being affected by both local and regional contributions. These contributions are a result of the coupling between emission and transport processes, which vary with region of the atmosphere and season of the year. Further, trace gas levels over a location are also affected by chemistry, photochemistry, dynamics, and meteorology. The characterization of local variability becomes very important over major urban regions viz. megacities which are major sources (and sometime sinks) for a plethora of species affecting chemistry and budget at regional to global scales. The characterization is better elucidated if measurements involve multiple chemical species, which can serve as proxies for different source and sink terms. Some of these gases, e.g., ozone (O3), nitrogen oxides (NOx), carbon monoxide (CO), methane (CH4), non-methane hydrocarbons (NMHCs), sulfur gases including sulfur dioxide (SO2), etc. play important roles towards various atmospheric processes that affect the overall state of our health and ecosystem.

Measurements of SO2 are highly desirable over urban regions as it colligates important aspects of air pollution, atmospheric chemistry, and climate. It is a primary criteria pollutant with potentially adverse health effects and concomitant deleterious impacts on the flora and fauna. It is a precursor of sulfate aerosols which exert a direct radiative forcing (RF) of −0.40 ± 0.20 W m−2 (Forster et al. 2007). One half to two thirds of this RF is attributable to anthropogenic sulfur (mainly SO2). Despite global reductions in SO2 emissions, a 70 % growth is estimated over India during 1996–2010 (Lu et al. 2011). Ninety-one percent of these emissions are attributed to power-plants and industries (emitting 5,236 and 2,784 Gg SO2, respectively, in 2010). Various emission sources of SO2 are presented in Table 1. Garg et al. (2001) have estimated that power generation alone accounts for 46 % of total SO2 emissions in India. Annual electricity consumption in India is 600 million MWh (http://www.indexmundi.com/map/) with 7,000 tonnes SO2 emitted for every megawatt hour of electricity produced. Most of this electricity (about 60 %) comes from coal based thermal power plants. Annual coal (0.2–0.7 % S) consumption in India is around 340 Mt (http://www.eai.in/ref/fe/coa/coa.html). SO2 emissions from the oil sector vary between 0.5 and 8 % (Garg et al. 2001), the highest being for fuel oil. India consumed about 1,165 million barrels of crude oil (includes heating oil, gasoline, diesel, etc.) during 2012. On average, petroleum refineries in India emit about 48 tonnes of SO2 for every 1,000 barrels of production.

Table 1 Sources of SO2, CO, CH4, and NMHCs
Full size table

The importance of CO and NMHCs in controlling tropospheric O3 is very well documented. These gases also wield substantial influence on air quality and global climate. Being important consumers of hydroxyl radicals, they can enhance the lifetime of atmospheric greenhouse gases. Documentation of their levels over major emission regions like mega-cities is very important due to potentially large fluxes. The major sources of CO and NMHCs are presented in Table 1. It is to be noted that systematic representative measurements of SO2, CO, and NMHCs are still limited to very few locations in India (Tables 2 and 3), and hence knowledge about their abundance and variability could have cornerstone implications to the understanding and modeling of tropical, atmospheric chemistry. In the present paper, we report a year-long observation of SO2, CO, CH4, and NMHCs (C2–C5) over the megacity “Kolkata”, an urban region in Eastern India (Fig. 1). Being located in tropical South Asia, Kolkata can significantly influence the regional chemistry. Based on multi pollutant index (MPI) studies, the air quality of Kolkata is worse than several other megacities of the world (Gurjar et al. 2008). High levels of pollutants during winter over Kolkata are also clearly observed from surface measurements (Mallik and Lal 2013) as well as satellite measurements of CO. Electronic supplementary material (ESM) Fig. 1 shows that during April, high levels of CO are observed over Bangladesh (east of Kolkata) whereas during winter, high levels of CO occurs over Kolkata. Further, satellite based CO measurements clearly show higher levels over the eastern part of the Indo-Gangetic plains (IGP). The outflow of IGP pollution into the Bay of Bengal (BoB) occurs over this region as the winds turn towards south instead of going further east (Mallik et al. 2013). Thus, it is apt to describe Kolkata as the gateway for air pollutants from the IGP to the BoB during winter and post-monsoon and vice versa during monsoon and pre-monsoon. From the perspective of emission inventories, these measurements can characterize a major emission source region in Eastern India.

Table 2 Average concentrations of SO2 (parts per billion by volume) over different regions in India and the world. Summer (s; premonsoon) and winter (w). Most of the datasets are based on continuous monitoring (UV Fluorescence) method. The data for Mumbai is based on samples collected once a month and analyzed using West and Gaeke method. The data for Europe is based on chemical ionization mass spectrometry
Full size table
Table 3 CO, CH4, and NMHCs (parts per billion by volume) over different regions in India and the world. The Ahmedabad data is based on direct online GC measurements for periods stretching 72 h (1-h interval) for each month
Full size table
Fig. 1
figure 1

The study location in (1) Kolkata (city map on the right). Major power plants (a) Farakka (2,100 MW), (b) Menjia (2,340 MW), (c) Bakreshwar (1,050 MW). Industrial areas: (2) Haldia (petrochemicals), (3) Durgapur (steel industry), (4) Asansol (steel, locomotives), (5) Raniganj (coal fields). Industrial areas in Kolkata: (6) Kasba, Tangra, Beliaghata (leather, polymer); (7) Behala, Taratala, Majerhat (engineering, pharmaceuticals, food); (8) Garden Reach (ship builders). Budge-Budge thermal power station (750 MW) lies 40 km south-west of study location. Kolaghat thermal power station (1,260 MW) is 40 km to west of Budge-Budge (not shown in map)

Full size image

Methodology

The study location

The study was conducted in Kolkata (formerly Calcutta; 22.55°N, 88.50°E; 6 m amsl), a megacity in Eastern India, spread over an area of 1900 km2 with population of about 15.7 millions. Kolkata is the capital city of the West Bengal state of India (Fig. 1). The port of Kolkata has two major dock agglomerations at Kolkata and Haldia. Haldia is a major port city, about 120 km south-west of Kolkata with a major refinery. Durgapur and Asansol are two other major industrial townships about 170 and 210 km to the north-west of Kolkata. Huge power plants lie within a few hundred kilometers in the north-west of Kolkata. The study site is located in the lush green campus of Jadavpur University at Salt Lake, a planned township in East Kolkata, with mostly office and residential buildings. The city itself has a few small-scale industries mostly located in the south-west of the study location. Dhapa, located south of the study location, is a major landfill site for dumping solid wastes of Kolkata. A major international airport is located within 10 km north of the study location. Public transport is more favored in Kolkata than private vehicles. Excluding two-wheelers (0.18 million) and cars (0.19 million), the city has about 0.45 million registered motor vehicles constituted by trucks (20 %), buses (6 %), taxi (45 %), and auto-rickshaws (28 %) (http://morth.nic.in/index2.asp?slid=291&sublinkid=137&lang=1). Most of these vehicles run on diesel, auto-rickshaws run on compressed natural gas (CNG) while two-wheelers and cars run on petrol (gasoline).

Experimental details

Continuous in situ measurements of surface SO2 were made over Kolkata during March 2012–February 2013, using an online gas analyzer from Thermo Scientific (43i-TLE). It is based on UV fluorescence (USEPA Equivalent Method EQSA-0486-060; http://www.epa.gov/ttn/amtic/criteria.html) and achieves a detection limit of 50 pptv for a 300-s averaging time. Ambient air is drawn through 4-m-long PTFE tubing (5-mm ID) into the analyzer through a 5-μm PTFE filter (to remove dust). The air then passes through a hydrocarbon kicker which removes hydrocarbons that fluoresce in the same region as SO2 (Mohn and Emmenegger 2001). The SO2 molecules are then excited by UV radiation (214 nm). The resulting fluorescence is measured by a PMT. Periodic zero calibration of the analyzer is carried out using zero air cylinders from British Oxygen Corporation, India, and a series of purafil/molecular sieve scrubbers. Span calibrations are done by diluting a gas mixture from Intergas (International Gases & Chemicals), UK (concentration: 471 ppbv in N2, traceable to National Physical Laboratory, UK) and a multi-gas calibrator (MCG146i; Thermo Scientific, USA).

In addition to the online SO2 measurements, 71 air samples were collected during February 2012–February 2013, for offline GC analysis. The air samples were collected from 4 m above ground level using a Teflon tube at 3-h intervals between 0730 to 1930 IST into pre-evacuated (to 10−5 mbar) 900-ml borosilicate glass sampling bottles using an oil-free air compressor (MB-158-E, Metal Bellow, USA). The Indian Standard Time (IST) is 5 h and 30 min ahead of Greenwich Mean Time (GMT). Each bottle was flushed several times with the ambient air before finally filling the sample to a pressure of around 2.5 bar leading to sample sizes of about 2.25 L (STP) each. These samples were analyzed in the laboratory at PRL, Ahmedabad for CH4, CO, and light NMHCs (C2–C5), using gas chromatographic methods given in Sahu and Lal (2006a), within a month of sampling at Kolkata. The stability tests regarding storage of these hydrocarbons showed variations of the order of 4–5 % w.r.t. the mean for a 2-month period (ESM Fig. 2). The response of the FID detector was found to be linear over the range 100 pptv to 60 ppbv for ethane (C2H6), ethene (C2H4), propane (C3H8), propene (C3H6), acetylene (C2H2), n-butane (n-C4H10), i-butane (i-C4H10), and i-pentane (i-C5H10). The reproducibility in NMHCs analyses ranged between 3 and 10 % depending upon the compounds, in general lower molecular weight hydrocarbons showed better reproducibility.

Auxiliary data sources

Hourly means of temperature, relative humidity, wind speed, and direction during March 2012–February 2013 from the nearby Kolkata Airport were used for the analysis. In addition to this, daily data for surface wind components (zonal, meridional; in meter per second), omega (in Pascal per second), and lifted index (degree kelvin) over 22.5°N, 87.5°E were obtained from 2.5° × 2.5°-resolution NCEP reanalysis (http://www.esrl.noaa.gov/psd/data/gridded/data.ncep.reanalysis.surface.html). Trajectory data and boundary layer heights were obtained from NOAA archives using HYSPLIT (http://ready.arl.noaa.gov/HYSPLIT.php). Further, SO2 mass mixing ratios (in kilogram per kilogram) and columnar dust AOD at 550 nm were obtained from European MAAC model reanalysis (22.5°N, 88.5°E; http://data-portal.ecmwf.int/data/d/macc_reanalysis/) for comparison. CO mixing ratios (1° × 1° monthly gridded, level 3) are obtained from “Measurements Of Pollution In The Troposphere” (MOPITT) for the year 2012 (ftp://l4ftl01.larc.nasa.gov/MOPITT/MOP03M.004.).

Results and discussion

Wind pattern over Kolkata during the study period

The prevailing wind patterns over Kolkata during the study period are shown in Fig. 2. Based on their spatial coverage, the air masses may be divided into three major types. During pre-monsoon (March, April, May), Kolkata is influenced by low-level southerly air masses from the Bay of Bengal and coastal India. These air masses spend 75 % time below 500 m and 15 % time between 500 and 1,000 m. During monsoon (June, July, August, and early September), the wind regime shifts to south-westerly and brings air parcels from southern peninsular India. These air parcels spend about 64 % time below 500 m, 18 % time between 500 and 1,000 m and 15 % time between 1,000 and 2,000 m. A complete reversal of winds takes place during late September (ESM Fig. 3). During the post-monsoon (late September, October, November), wind is mostly north-westerly over the study location. During winter (December, January, February), the prevailing wind-pattern is same as during post-monsoon, bringing in air masses from the IGP. Reversal of winds again takes place in March. During post-monsoon, air masses reside about 37 % time below 500 m, 21 % time between 500 and 1,000 m and 35 % time between 1,000 and 3,000 m. The residence times during winter are 37 % below 500 m, 31 % between 500 and 1,000 m and 28 % between 1,000 and 3,000 m. During pre-monsoon, the trajectories reside 89 % time over oceans whereas during monsoon, this value reduces to 43 %.

Fig. 2
figure 2

Seasonal patterns of prevailing winds over Kolkata. One hundred twenty-hour backward trajectory data from HYSPLIT are grouped into three major regimes representing pre-monsoon (March–May), monsoon (June–August) and post-monsoon + winter (late September–February). The altitude of these trajectory groups is shown by the color bar. The colored circles represent power plants (which are large point sources for SO2). Size and color of the circles are scaled according to the capacity of the power plants; e.g., the Mudra Thermal power station in western Gujarat, fifth largest single location coal-based thermal power, has an installed capacity of 4,620 MW

Full size image

Seasonal variation in surface SO2

Hourly average SO2 concentrations are calculated from 5-min averages recorded by the instrument. A comparison of SO2 values obtained over Kolkata with other locations is presented in Table 2. The median concentrations of SO2 for Kolkata are 0.8 for premonsoon and 5.2 during winter. Assuming local emissions do not change much throughout the year, the several fold increase in winter concentrations seem to be a result of atmospheric dynamics (meteorology and regional transport). The median concentration of SO2 over Kolkata during premonsoon is slightly higher than median concentration observed over Ahmedabad (0.6; Mallik et al. 2012). The median SO2 concentration observed over Mumbai over a 4-year sampling period was, however, 7.5 ppbv. The mean winter concentrations of SO2 over Kolkata are comparable to concentrations observed over Seoul and Hong Kong (Table 2).

The variation in observed hourly average SO2 concentrations as a function of wind direction is shown in Fig. 3. It is clearly observed that the hourly SO2 concentrations during post-monsoon (mean = 3.5 ppbv) and winter (mean = 6.4 ppbv) are far higher than those during pre-monsoon (mean = 0.83 ppbv) and monsoon (mean = 0.72 ppbv). From the figure, there seems to be a dominant effect of prevailing wind regimes, resulting in different catchment areas for SO2 during the different periods. During post-monsoon and winter, sources are mainly in north and north-west (also Fig. 4a). Decent correlations between daily means of SO2 and zonal wind (R 2 = 0.23 and 0.31 during winter and post-monsoon; n = 90 for each season) and a skewed distribution of SO2 with wind direction indicate a regional influence. Further, a sudden change in SO2 levels is observed towards the end of September, along with a concurrent change in wind regime (ESM Fig. 3). This change is also reciprocated by the MAAC simulations (ESM Fig. 4). There seems to be an association between SO2 and lifted index (LI; a measure of atmospheric stability) during post-monsoon (slope = 0.43, R 2 = 0.67, n = 90). During this period, SO2 values start to increase as the LI values change from negative to positive, i.e., atmosphere becomes more stable.

Fig. 3
figure 3

Variation of hourly SO2 (parts per billion by volume) with wind direction during different seasons over Kolkata during the study period

Full size image
Fig. 4
figure 4

Probable source regions contributing to elevated SO2 concentrations over the study location during a January and b July. The maps are derived from Potential source contribution function analysis using TrajStat software (Wang et al., 2009). Runtime, 120 h, cell size = 2°×2°. Pollution criteria, 75 percentile of monthly variation of SO2. The red grids represent a probability of 0.7–1, followed by pink (0.42–0.7) and orange (0.17–0.42)

Full size image

The regional sources potentially contributing to higher SO2 concentrations during January and July are shown in Fig. 4a, b). These maps are based on potential source contribution function (PSCF) analysis (Wang et al. 2009). Figure 4a shows that SO2 concentrations greater than 75 percentile of January values are related to air masses passing over steel industries in Bokaro and Durgapur, coal mining regions of Jharia and Raniganj, apart from numerous other industrial townships and polluted areas in West Bengal, Jharkhand, Bihar, and Uttar Pradesh. These air masses also intersect with numerous power plants including Menjia (2,340 MW), Farakka (2,100 MW), and Bakreshwar (1,050 MW). Previous modeling studies have also shown these regions as hotspots for SO2 emissions (Adhikary et al. 2007). Coal combustion in power plants and steel industries are major contributors to regional SO2. Figure 4b shows that SO2 concentrations greater than 75 percentile of July values are related to air masses coming from the industrial townships of Haldia and Paradeep.

The average SO2 concentration during March–August is about 0.8 ppbv, which is about 16 % of the average concentration during September–February (4.9 ppbv). This is because the air masses during winter pass over more polluted regions of the IGP. Although numerous power plants lie in the catchment area for SO2 during March–August (Fig. 2), dilution effect is larger during this period. Meng et al. (2010) attribute lower SO2 concentrations during summer at several Chinese sites to vertical mixing and removal mechanisms (precipitation, photochemical). Dilution effects can be measured from ventilation coefficient (VC), obtained by multiplying wind speed with mixed layer depth (MLD). In source regions such as megacities, VC represents the rate by which pollutants will be blown out of the mixed layer over a particular location. Chin and Jacob (1996) have estimated that ventilation accounts for 90 % of SO2 export from USA to the free troposphere in summer while this reduces to 30 % in winter. MLD values over the study location are generally below 500 m during winter and post-monsoon but extend to 2,000 m and above during pre-monsoon and monsoon. Further, wind speeds are higher over Kolkata during monsoon and pre-monsoon (mean = 3.4 m/s) accentuating the VC values during these periods. Thus higher VC during pre-monsoon will cause dilution of local emissions while lower VC during winter and post-monsoon results in stronger trapping of transported air parcels from other source regions as well as local emissions. Figure 5 confirms that higher SO2 concentrations generally occur at lower VC. However, the correlations of SO2 with VC or MLD are not strong enough to quantify their effects.

Fig. 5
figure 5

Variation of SO2 with ventilation coefficient

Full size image

Apart from the effect of source contributions, removal mechanisms play a vital role in governing ambient SO2 levels. The major sinks are oxidation, dry-deposition, and wet-deposition, with global average contributions of 53, 36, and 8 %, respectively (Faloona 2009). The columnar dust AOD at 550 nm over Kolkata during summer is 0.36, which is about 45 % higher than winter average (0.25) and could result in higher heterogeneous losses of SO2. Adhikary et al. (2007) have discussed the importance of heterogeneous pathways in conversion of SO2 to sulfate over the Indian region, especially during the dry season. Guttikunda et al. (2001) have also estimated high sulfur deposition over the Indian region during summer. High SO2 dry deposition velocities were calculated over China during summer (0.45 cm/s; Wang et al. 2003). The efficacy in wet removal processes in determining SO2 concentrations over heavy rainfall areas like Kolkata could be significant during monsoon (Matsuda et al. 2006). Rai et al. (2010) have calculated a deposition rate of 185 μg m−2 s−1 for SO2 removal at a rainfall rate of 168 mm/h. However, more detailed analysis is needed to quantify these effects.

Diurnal variation in surface SO2

The monthly, diurnal patterns for SO2 are constructed from averaged hourly data for all days of the month. Virtually, no diurnal variation in SO2 is observed during March and August. During premonsoon and monsoon, under the southerly wind regime (180°, Fig. 3), elevated daytime values in SO2 are observed (Fig. 6). Daytime enhancement in SO2 envisages effects of photochemistry. Daytime enhancements in SO2, observed at Hawaii (De Bruyn et al. 2006), were attributed to oxidation of reduced sulfur compounds (RSCs; mainly DMS). It was shown in Fig. 1 that vast landfill regions exist to the exact south of the study location within 5-km distance. Figure 3 shows elevated SO2 concentrations when wind is coming from south during premonsoon and monsoon. Thus, oxidation of RSCs (which have major sources in landfill sites) may be surmised. There could also be other anthropogenic sources of SO2 but their emissions will occur both during day and night. Nevertheless, the hypothesis regarding photo-oxidation of RSCs leading to daytime enhancements in SO2 needs to be excogitated with quantitative estimations over several years.

Fig. 6
figure 6

Diurnal variations of SO2 during different seasons. The boxes represent the 25 and 75 percentiles

Full size image

Early morning enhancements are observed during September along with hint of late afternoon peaks. These early morning and afternoon to evening enhancements become very prominent in post-monsoon months of October and November (Fig. 6). Daytime dip in SO2 concentrations occur during winter and post-monsoon. These daytime dips are conspicuous during October and November and could be the result of boundary layer dilution. Similar features were also observed at Lin’an, China (Wang et al. 2004b). High SO2 concentrations during winter mornings (~8 IST) could be a result of colligated impacts of local emissions and boundary layer dynamics. During February, a gradual increase in SO2 concentrations from midnight to early morning seems to be a clear effect of decreasing MLD. Absence of evening peaks during winter asserts that vehicular traffic does not significantly impact SO2 concentrations over this region. A weekday to weekend ratio analysis shows that the median values are close to 1 during the study period. However, these values are much lower than 1 during winter and post-monsoon, suggesting more pre-dominant impacts of regional sources (power plants and industries which prevent the increase of weekday values) rather than local emissions (vehicular etc. which tend to increase weekday values).

Variations in CO, CH4, and NMHCs

Figure 7 shows the diurnal variation of CO and hydrocarbons over Kolkata based on seasonal means for 0730, 1030, 1330, 1630, and 1930 IST observations. Similar to MOPITT observations (ESM Fig. 1), CO values in surface measurements are always elevated during winter compared to pre-monsoon (Fig. 7). Further, the daytime patterns in hydrocarbons during winter reveal high values during morning and evening. The morning values seem to be more related to atmospheric stability conditions rather than traffic activity (which should have peaked between 0900 and 1100 IST). Mixed layer height is lower during winter mornings compared to summer mornings, resulting in lower dilution of accumulated pollutants during night, leading to higher concentrations of hydrocarbons at 0730 IST. Further, it is clearly observed from Fig. 7 that unlike other gases, daytime concentrations (excluding morning and evening) of C2H4 and C3H6 are comparable for premonsoon and winter. However, early morning values (0730 IST) for these gases are far higher in winter than premonsoon, pointing to non-vehicular influences. The peak traffic hours in the morning (0900–1100 IST) are marked by elevated i-C5H10 values, pointing to gasoline evaporation.

Fig. 7
figure 7

Daytime variations of CO, CH4, and NMHCs (C2–C5) over Kolkata during March 2012–February 2013

Full size image

The average concentrations of CO, CH4, and NMHCs over Kolkata during the study period are presented in Table 3. A comparison has been made with other available measurements in India and some other urban regions of the world. It is observed that CO and C2H6 levels over Kolkata are comparable to several Chinese sites while C3H8 and C4H10 are substantially higher over the study region. The September–February to March–August ratios are 3.2, 2.2, 1.9, 3.2, 2.3, and 2.4 for C2H6, C2H4, C3H8, C3H6, C2H2, and C4H10, respectively. However, in case of i-C5H10, CH4 and CO, these ratios are 0.84, 1.1, and 1.3, respectively. Among the measured hydrocarbons, C4H10 contributes about 29 % to the C2–C5 total, followed by C3H8 (25 %). Over Kolkata, the propylene equivalents, which are also an indicator of O3 forming potential (Chameides et al. 1992), are highest for C3H6, followed by C2H4 and i-C5H10.

Source signatures from inter-species relationships

Table 4 presents the inter-species relationships among CO, SO2, and hydrocarbons. To account for the impact of prevailing wind regimes, the data is divided into summer (March–August) and winter (September–February). Since it is likely that photochemistry can influence the relationships significantly (Leuchner and Rappengluck 2010), the table presents relationship based on all observations as well as observations excluding the noontime period (when photochemical activity is maximum). Excellent correlations are observed for butane isomers for both the seasons (Table 4). The slope matches the values (0.34–0.52) obtained over several sites in UK which was attributed to evaporative emissions from petrol station refueling and refinery operations (Derwent et al. 2000). C4H10 also shows good correlations with C3H8 (Table 4), alluding to LPG emissions (Table 1). Good correlation of n-C4H10 with C2H2 confirms impact of vehicular emissions (Xie and Berkowitz 2006). Overall, strong correlations among C3H8, C4H10, and C2H2 suggest impacts of local vehicular emissions as well as influence of LPG. Associations between C2H4 and C3H8 suggest additional co-emissions from combustion sources. Strong correlations between CH4, C2H6, and C2H4 during winter compared to pre-monsoon also indicate influence of combustion sources (biomass/biofuel).

Table 4 Correlations between SO2 (parts per billion by volume), CO (parts per billion by volume), CH4 (parts per million by volume), and NMHCs (parts per billion by volume) over Kolkata for the period March 2012 to February 2013. The data is divided into two time periods (a) winter + postmonsoon, (b) premonsoon + monsoon. The analysis is given for two datasets, one including all data, the other excluding noontime data. The numbers in tables are presented in the format of slope (R 2)
Full size table

CO is strongly correlated to CH4, C2H6, C2H4, C2H2, and C4H10 during winter indicating impacts of biomass/biofuel burning. Since not much biomass burning occurs during this period, it is more likely that biofuel burning along the east IGP is the major source of CO (Table 1). The C2H6/CO slope obtained in this study (0.007 ppbv/ppbv) is very close to the values obtained for Ahmedabad (0.008; Sahu and Lal 2006a) and Hissar (0.01; Lal et al. 2012) as well as Japan/Korea (Wang et al. 2004b). Further, C2H4 to CO slope of 0.02 obtained by Lal et al. (2012) is close to winter values obtained in this study indicating impacts of biofuel on CO values. However, there are bound to be significant impacts of fossil fuel combustions to CO levels in the IGP as suggested by low C2H6/C3H8 (slope = 0.2, R 2 = 0.6) during winter. Good correlations also exist between CO, C3H8 and n-C4H10 during winter, corroborating co-emissions from petrochemical industries/LPG sources (Table 1). Efficient mixing of various emissions from a plethora of collocated sources accompanied by boundary later dilution results in weak source signatures during pre-monsoon. Although the correlations of CO with hydrocarbons are very poor during pre-monsoon, the R 2 increases significantly when the data analysis excludes noontime values, indicating influence of photochemistry in CO–hydrocarbon relationships. Particularly for CO-CH4 relationship, when noontime data is excluded, the R 2 increases significantly for summer data but decreases for winter data. Nevertheless, source signatures can be difficult to interpret over urban regions, unless measurements of specific tracers like 3-methyl pentane (gasoline marker) or isoprene (biomass burning indicator) are available. The major difficulty in source interpretation over urban regions is that there could be a plethora collocated sources emitting the same gas.

The unique combination of SO2, CO, and NMHC measurements allows to establish that SO2 is more likely to be impacted by regional anthropogenic sources (power plants, industries, and refineries) rather than local emissions (transport, residential). This is because SO2 is more strongly associated with CO and C2H6 and exhibits poorer correlations with gases emitted locally (C3H8, n-C4H10 and C2H2) during winter. This is corroborated by PSCF analysis and weekday/weekend ratios. Nevertheless, being a major urban region, there would still be some local emissions of SO2 impacting its background concentrations over Kolkata. The observed SO2/CO slope during winter is lower than an estimate of 0.045 over China, which was attributed to coal burning (Wang et al. 2004b). This could be due to lower sulfur content of Indian coals.

Source apportionment from PMF analysis

Positive matrix factorization (PMF) is a multivariate factor analysis tool that decomposes a matrix of speciated sample data into matrices that facilitates grouping of sources based on observations at the receptor site (Paatero and Tapper 1994). In general, the PMF model provides robust results for the identification of sources in complex atmospheric environments (Leuchner and Rappengluck 2010). PMF v3.0 was utilized in this study using 10 species and 55 samples. In order to reduce the influence of photochemistry, the noontime samples were excluded from the PMF analysis (Leuchner and Rappengluck 2010). Details regarding this method can be found in Ling et al. (2011) as well as Leuchner and Rappengluck (2010). The analysis resulted in six major factors (Fig. 8). Factor 1, dominated by CO and CH4. Since CH4 over Kolkata is not likely to be influenced much by agricultural sources, this factor connotes biofuel/biomass combustion (mainly biofuel). Factors 2 and 4, dominated by C2H2 and C2H4, seem to be related to traffic emissions (Ling et al. 2011). Factor 2 is dominated by alkenes and is related to petrochemistry while factor 4, dominated by C2H2 and C2H4, is more closely related to vehicle exhausts (Leuchner and Rappengluck 2010). Factor 3 is dominated by SO2 and related to coal combustions (power plants and industries). However, abundance of C2H6 in factor 3 also indicates influence of natural gas and evaporation, which could occur from collocated emission sources followed by wind borne transport (it may be noted that significant correlations exist between SO2 and C2H6, mainly during winter).

Fig. 8
figure 8

Source profiles (percentage of species total) resolved with PMF analysis

Full size image

Factor 5 is dominated by i-C5H10 and identified as fuel (mainly gasoline) evaporation (Table 1). Factor 6 is dominated by C3H8 and C4H10 and this combination of species is typically found in liquefied petroleum gas (Table 1; Leuchner and Rappengluck 2010).

Conclusion

A year-long study of SO2, CO, and hydrocarbons over Kolkata during 2012–2013 enabled interpretation of emission characteristics over this region. The observed SO2 concentrations are comparable to several Asian sites but higher than European sites. High levels of SO2 during winter (>6 ppbv) are attributed to regional emissions and subsequent trapping of these air masses favored by a stable atmosphere with low ventilation coefficient. Coal burning in industrial areas and power plants in eastern IGP are identified as potential source regions for SO2 during winter. Daytime elevations in SO2 during summer seem to be related to photo-oxidation of RSCs from a nearby landfill region. The daytime enhancement in MOPITT derived CO over Kolkata during December (478 ppbv) over April (331 ppbv) is about 44 %, which is also reciprocated by in-situ, sample-based measurements during April (323 ppbv) and December (469 ppbv) for the year 2012. Early morning enhancements during winter for several trace gases indicate the role of MLD dynamics. Interspecies correlations show the dominant influence of LPG leakage and petrochemical industries to local air quality during winter apart from vehicular traffic. Correlation analysis shows that CO is dominated by biofuel combustions. Overall, it may be concluded that the major emission sources of different gases are not collocated. The “sui generis” combination of measured species in this study has helped to segregate impacts of transport from local emissions. From the point of view of emission inventories, these measurements over Kolkata (the world’s 16th largest megacity in terms of population) would be very useful for characterizing urban Indian emissions and potentially also represent South Asia. The concentrations over Kolkata may be interpreted as the end point of anthropogenic inputs to the IGP outflow into the BoB and subsequently the Indian Ocean during winter. Thus, these measurements (e.g., C2H2 to CO ratios) can serve as initial values to calculate photochemical processing of air masses over the BoB. Nevertheless, measurements of additional tracers, e.g., isoprene, methyl chloride, etc. are required to better characterize the emission scenarios over this region.