Wintertime characteristics of aerosols at middle Indo-Gangetic Plain: Impacts of regional meteorology and long range transport

Atmospheric Environment 104 (2015) 162e175

Contents lists available at ScienceDirect

Atmospheric Environment journal homepage: www.elsevier.com/locate/atmosenv

Wintertime characteristics of aerosols at middle Indo-Gangetic Plain: Impacts of regional meteorology and long range transport M. Kumar a, S. Tiwari b, V. Murari a, A.K. Singh b, T. Banerjee a, * a b

Institute of Environment and Sustainable Development, Banaras Hindu University, Varanasi, India Atmospheric Research Laboratory, Department of Physics, Banaras Hindu University, Varanasi, India

h i g h l i g h t s
Exceptionally high aerosol mass loading for both PM10 and PM2.5 at middle IGP.
Space borne & ground based AOD reveal variability and moderate association with PM.
CALIPSO cross-section profiles depicts altitudinal distributions of aerosols.
At lower altitude continental wind accumulate fine PM from north-western dry part.
At higher altitude coarser PM accumulate due to strong intercontinental westerly.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 17 November 2014 Received in revised form 4 January 2015 Accepted 5 January 2015 Available online 6 January 2015

To develop a coherent picture of possible origin of Asian aerosol, transport and meteorological interaction; wintertime aerosol (January, 1 to March, 31, 2014 (n ¼ 90)) were measured in middle IGP in terms of aerosol mass loading, optical properties, altitudinal distributions and both high and low altitude transportation. Both space-borne passive (Aqua and Terra MODIS) and active sensor (CALIPSO-CALIOP) based measurements were concurrently used over the selected transect (25100 e25190 N and 82 540 e83 40 E). Exceptionally high aerosol mass loading was recorded for PM10 (233 ± 58.37 mg m3) and PM2.5 (138 ± 47.12 mg m3). Daily variations of PM2.5/PM10 persist in a range of 0.25e0.97 (mean ¼ 0.60 ± 0.14; n ¼ 90) and were in accordance to computed Angstrom exponent (0.078e1.407; mean: 1.002 ± 0.254) explaining concurrent contribution of both PM2.5 and PM10 for the region. Space borne (Aqua MODISAOD: 0.259e2.194) and ground based (MTP-AOD: 0.066e1.239) AODs revealed significant temporal variability and moderate association in terms of PM10 (MODIS-AOD: 0.46; MTP-AOD: 0.56) and PM2.5 (MODIS-AOD: 0.54; MTP-AOD: 0.39). Varying association of AOD and aerosol mass loading was also explained in terms of meteorological variables. CALIPSO altitude-orbit-cross-section profiles revealed presence of non-spherical coarse particulates (altitude: 1.2e5.4 km) and dominance of spherical fine particulates (altitude: 0.1e4.2 km). Contribution of trans-boundary aerosols transportation to mass loadings at middle IGP were recognized through lagrangian particle dispersion model, synoptic vector wind profiles at different geopotential heights and satellite images. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Aerosol Aerosol optical depth CALIPSO Indo-Gangetic Plain MODIS Trans-boundary

1. Introduction Aerosols are multi-component mixtures originate from a range of regional and global sources and have potential to alter Earth's climatic balance by affecting physical, chemical or optical properties. Aerosol-climate chemistry includes heterogeneity at spatial

* Corresponding author. Institute of Environment and Sustainable Development, Banaras Hindu University, Varanasi 221005, India. E-mail addresses: [email protected], [email protected] (T. Banerjee). http://dx.doi.org/10.1016/j.atmosenv.2015.01.014 1352-2310/© 2015 Elsevier Ltd. All rights reserved.

levels and hence their effects vary with topography, climate and meteorological conditions (Murari et al., 2014). Among several identified regional hotspots around the globe, Indo-Gangetic Plains (IGP) is considered to be most vulnerable to aerosol induced climate impacts. Thus creates the essentiality of conducting a comprehensive research to identify local and transboundary sources of aerosol, its association with regional meteorology, transport mechanism and altitudinal distribution. The nature of aerosols at middle IGP is mostly characterized by presence of mineral dust, organic aerosols and elemental carbon produced through burning

M. Kumar et al. / Atmospheric Environment 104 (2015) 162e175

of biomass and fossil fuels, which further get complicated by transboundary movement of aerosols originating from the Middle-east countries and Thar Desert. Additionally, burning of agricultural residues during winter has been a common practice in eastern Uttar Pradesh and Bihar which predominately adds large quantity of aerosols of different nature and sizes (Ram et al., 2010; Murari et al., 2014). A substantial number of studies on varying spatial and temporal scale at different locations of IGP, i.e. Lahore (Ghauri et al., 2012), Patiala (Mittal et al., 2009), Hisar (Raman et al., 2011), Delhi (Sharma et al., 2014), Varanasi (Murari et al., 2014; Tiwari and Singh, 2013) and Kolkata (Karar and Gupta, 2007) highlights the presence of quantum of both primary and secondary aerosols. However, for the present analysis, efforts were made to understand the entire spectrum of aerosol from origin to its transport, meteorological interaction and vertical distribution so that a coherent picture may be established. For the current analysis, ground based aerosol observations were made at Varanasi, India. Situated at middle IGP, Varanasi is not an exception of having worst statistics in terms of aerosols which exceeds the national standards on a frequent basis. The submitted manuscript initially presents an approach to associate the behavior of ambient aerosols (PM10 and PM2.5) with regional meteorological parameters. Such association is extremely critical in terms of forecasting aerosol chemistry in context of projected climate change. Additionally, efforts were made to identify interrelations of ground based aerosol mass loading with its optical properties collected both through ground and satellite based platforms. Altitudinal distribution of aerosol for the selected transect was made through active lidar instrument. Conclusively, based on aerosol vertical profile, synoptic meteorological data was plotted for entire region to understand trans-boundary origin and transport of aerosol for middle IGP. The implications of such findings may well be useful to understand uncertain association of aerosol mass concentrations with regional meteorology and its optical properties, aerosol origin and transport for a region long been projected as most vulnerable through aerosol induced climate change. 2. Experimental methods 2.1. Site description Varanasi, located at the bank of the river Ganges in the middle Indo-Gangetic Plains is considered as one of the ancient and holiest cities of the world. The entire study was carried out at the

163

premises of IESD-BHU campus, Varanasi (25160 2900 N, 82 590 4600 E) (Fig. 1). The sampling site is represented by an institutional set up surrounded by an urban environment typically characterized by traffic congested roads with mixed residential and commercial sprawl in the northern side and densely populated residential areas in the eastern, western and southern part. The region is climatologically affected by wide range of synoptic weather patterns but is devoid of any localized effects of oceans and mountains. Relatively flat topography is believed to simplify the atmospheric boundary layer structure of the region and improve the applicability of assimilated meteorological parameters used in the analysis. 2.2. Ground-based in-situ measurements All ground-based in-situ aerosol measurements for the current analysis were made at Varanasi (25160 N, 82 590 E, 77 m msl). Aerosol samples having aerodynamic diameter 10 mm (PM10) and 2.5 mm (PM2.5) were collected continuously seven days a week for entire winter months i.e. January, 1 to March, 31, 2014 (n ¼ 90). Aerosol samplers were placed at an elevation of 7.5 m at the roof of IESD-BHU and monitoring was continued for 22 h (1200e1000 h) on each consecutive day. Coarser particulates were collected through particulate sampler with size selective inlet (IPM-FDS, Instrumex). Ambient air was passed through the sampler using glass fiber filter (GF/A, Whatman, 47-mm diameter) with an airflow of 16.67 LPM (flow meter resolution of ±2% under actual operating condition). Additionally, fine particulates (PM2.5) were sampled through polytetrafluoroethylene filters (PTFE, Whatman, 47-mm diameter) with an airflow of 1 m3 h1 (accuracy ±2%) by fine particle sampler (APM 550, Envirotech). Filter papers were preconditioned in a desiccator for 24 h before sampling and preweighed using a microbalance (AY220, Shimadzu). Preconditioned filter papers were placed in filter holders (for PM2.5) and cloth-lined envelope (for PM10) before being taken to the field for sampling to avoid any possible contamination. Exposed filters were placed into cassettes and wrapped in aluminum foils to prevent exposure to sunlight and photooxidation. Aerosol mass concentration was gravimetrically calculated and exposed filters were stored under cool and dry condition (20  C) for further particulate speciation. All instruments deployed for the sampling have been frequently calibrated both before and after sampling.

Fig. 1. Geographical location of aerosol ground monitoring station.

164

M. Kumar et al. / Atmospheric Environment 104 (2015) 162e175

2.3. Micro-meteorology The climate of Varanasi is characterized as humid sub-tropical with severe cold during winter and extreme heat during summer. The flow of air mass is dominated by westerlies in summer and winter and easterlies during monsoon. The city typically experiences dense foggy conditions with reduced visibility during winter. Temperature of Varanasi varies from minimum of 8.4e15.0  C in winter (NovembereFebruary) to maximum of 38.5e41.2  C in summer (MarcheJune). The annual average rainfall at Varanasi is of the order of 1100 mm with minimum share during winter. The 24-h daily average micro-meteorological parameters were procured from wunderground.com, world-wide-web based meteorological database which provides a wide range of meteorological variables for a geographical area and previously been used by other researchers (Luvsan et al., 2012; Li et al., 2014). Additionally, observations were compared and validated with climatological means reported by regional weather monitoring station. Atmospheric boundary layer depths (ABL) were separately procured from Global Data Assimilation System (GDAS) which provides meteorological observations collected from various platforms to a gridded, 3-D, model space. The GDAS-ABL depths were collected from National Oceanic and Atmospheric Administration, Air Resources Laboratory (NOAA-ARL), Real-time Environmental Applications and Display System (READY) website (http://www.arl.noaa.gov/ready; Draxler and Rolph, 2003). The GDAS (1, 3-hourly) data of ABL for individual days of the study period were retrieved from the archived meteorology section of the website. The 3-hourly data were averaged and analyzed on daily basis to obtain the behavior of meteorological variables during the entire study period. 2.4. Synoptic distribution of aerosol 2.4.1. Satellite retrieved AOD AOD data corresponding to surface aerosol monitoring period (January to March, 2014) were retrieved from Aqua MODIS atmosphere level 2 products (3  3 km) at 550 nm. MODIS provides a spatial resolution of 250 me1 km (at nadir) with typical 1e2 days global coverage (swath of 2330 km ± 55 cross-track) with high radiometric resolution (12 bits) is essentially suitable for assessing spatial and temporal trends of finer particulates over large geographical areas (Liu et al., 2007). For the current analysis, area between 25100 3700 e25190 4700 N and 82 540700 e83 40 3000 E uniformly surrounding the study site was selected for retrieving AOD. Relations between aerosol optical properties and surface mass concentrations were previously established by different workers having diverse spatial resolutions i.e. 500 m (Bilal et al., 2013), 5-km (Kumar et al., 2007) and 10-km (Tsai et al., 2011). However, aerosol loadings and its prototypes frequently varied over lower spatial scales and therefore, AOD retrievals with higher spatial resolutions may found inappropriate to depict minute local variability and can be resulted into erroneous conclusions (Liu et al., 2007). This was the primary basis behind selective retrieval of Aqua MODIS 3-km product (MYD04_3K; MODIS collection 6) for the present study. MODIS 3-km atmospheric products have been newly introduced (January, 2014) and was retrieved through Atmosphere Archive and Distribution System (http://ladsweb.nascom.nasa.gov). Over land, the total MODIS AOD at 550 nm (Optial_Depth_Land_And_Ocean) (quality flag-3) denotes AOD encompassing both fine and coarser particulates. MODIS AOD data available in pre-selected grids were further averaged and represented for the entire site. Terra MODIS true images with corresponding AOD (collection 51) for each aerosol loading episodes were also compared for Indian subcontinent to identify composite anomalies of synoptic aerosol transportation within the region.

2.4.2. Ground based AOD In order to validate satellite retrieved aerosol optical properties and to determine physical properties of aerosol, ground based measurements of AOD was considered using hand held portable multiband sun photometer MICROTOPS-II (MTP-AOD) (Solar Light, USA). Details of MICROTOPSeII sun photometer, its calibration and performance have been provided by Morys et al. (2001). Ground based MTP-AOD data were collected at 500 nm on every half/or one hour on each clear sky day during the entire study period. Ground based measurements of AOD at study site were earlier reported by Tiwari et al. (2013); however, present analysis uniquely compare it with ground based PM mass concentrations and meteorological variables. 2.5. Aerosol vertical profile Aerosol vertical profiles over the selected transect were derived using Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) sensor onboard CALIPSO satellite for the selected transect. CALIOP sensor specifically detects and characterizes atmospheric aerosol profile using elastic laser backscatter at 1.064 mm and parallel and cross-polarized sections of 0.532 mm return signal (Kittaka et al., 2011). CALIPSO altitude-orbit cross-section profiles with 5-km ground horizontal resolution (level 2 version 3.30) were obtained from CALIPSO search and sub-setting web application (https:// www-calipso.larc.nasa.gov) on each 16-day CALIPSO repeat cycle for entire monitoring period. Total (b0 ) and perpendicular (b0 ⊥) attenuated backscattering at 532 nm and Total (b0 ) attenuated backscattering 1064 nm with vertical resolution of 1/3 km and 30 m respectively were initially compared to identify specific type of particulates. Further, Depolarization Ratio (DR), Vertical Feature Mask (VFM) and aerosol subtypes were also taken into account for assessing complete aerosol profile. 2.6. Particle backward trajectory and cluster analysis For the current analysis, trans-boundary aerosol distribution pattern was mapped through NOAA HYSPLIT model (Hybrid Single Particle Lagrangian Integrated Trajectory) (Draxler and Rolph, 2003). Also known as backward trajectory model, particle lagrangian model specifically simulate particle back trajectories in a 3-dimensional system by means of particle bulk-motion in preselected vector wind field and subsequently spread by turbulence. Used in combination with archived data and run interactively on the world-wide-web through the READY system (http://ready. arl.noaa.gov/HYSPLIT), it establishes the missing link between the synoptic air masses with spatial variations. Meteorological input was procured from National Center for Environmental Prediction/ National Center for Atmospheric Research (NCEP/NCAR REANALYSIS) global reanalysis meteorological datasets. Initially HYSPLIT was run on archived data set for the desired region (0e50 N, 0e100 E) to predict 5-days air mass back-trajectories for each aerosol loading episodes (HALDs, MALDs and LALDs) at three geopotential heights (1000 m, 2500 m and 3000 m). Subsequently a cluster analysis (CA) of 5 days air mass back trajectories at three geopotential heights were made for better source emission prediction. CA was especially performed to find particular air mass that traveled flowing identical trajectories and provide relative contribution of sources in different synoptic scales. Additionally, NCEP/NCAR REANALYSIS data was also used to assess the variation of 3-D wind fields from near surface (1000 m) to mid-troposphere (3000 m) with a horizontal resolution of 2.5  2.5 and 17 pressure levels (10e1000 hPa). Vector wind composite mean (m/s) for 700, 850, 925 hPa were plotted for desired study region (17e42 N, 50 e95 E) for three different

M. Kumar et al. / Atmospheric Environment 104 (2015) 162e175

165

aerosol loading episodes. These pressure heights correspond to different altitude ranges in a standard atmosphere (700 hPa ~ 3000 m, 850 hPa ~ 2500 m, 925 hPa ~ 1000 m) were deliberately chosen to understand relative accumulation of aerosol layers in different altitudes over selected transect. 3. Results and discussion 3.1. Temporal characteristics aerosol mass loadings Mass concentrations of coarse and fine aerosols were continuously measured from January 1st to March, 31st 2014 (n ¼ 90) at IESD-BHU campus and its daily variability has been plotted in Fig. 2. The 3-months average mass concentrations of PM10 and PM2.5 were recorded as 233 ± 58.37 and 138 ± 47.12 mg m3, (mean ± sd) respectively. Mass concentrations of both type of aerosols exceeded the 24-h average national standards (PM10 ¼ 100 mg m3; PM2.5 ¼ 60 mg m3; cpcb.nic.in), USEPA standards (PM10 ¼ 150 mg m3; PM2.5 ¼ 35 mg m3) (http://www.epa.gov/air/ particlepollution/) and the EU standards (PM10 ¼ 50 mg m3; PM2.5 ¼ N/A for 24-h, http://ec.europa.eu/environment/air). Additionally, every single 24-h particulate mass concentrations were found to persist (PM10 and PM2.5: 100%) well above the prescribed standards, which pose a considerable threat to surrounding inhabitants. For the entire monitoring period, particulate mass depicts higher variability from 106 to 373 mg m3 (PM10) and 70286 mg m3 (PM2.5) which typically resemble the pattern of particulate mass concentrations for the same station reported in earlier studies (Murari et al., 2014; Sen et al., 2014). It was expected that both PM2.5 and PM10 demonstrate a close statistical relation as finer particulates constitute a fraction of coarser suspended particulates. Fig. 3 shows a linear bi-variate plot of PM2.5 vs PM10 with computed coefficient of determination (R2 ¼ 0.42). Significant correlation (r ¼ 0.65) was also observed between PM10 and PM2.5 which depicts the similarity of their origins. Additionally, average and standard deviation of particulate mass ratio is often considered as an indicator of finer particulates relative contribution to coarser one. For the current analysis, daily variations of PM2.5/PM10 persist in a range of 0.25e0.97 with an average of 0.60 ± 0.14 (n ¼ 90) (Fig. 4). Such was a clear indication of fine particulate accountability on coarser particulate variability for the region. The accounted ratio (0.60 ± 0.14) was in line to the previously monitored value for the same station as reported by Murari et al. (2014) (0.59 ± 0.18, n ¼ 104).

Fig. 3. Ground level PM2.5 as a function of PM10.

3.2. Micro-meteorology and its impacts on winter aerosol The uncertain sensitivity of airborne particulates in context of regional meteorology partly depends on the interaction of its specific components with meteorological variables (Banerjee et al., 2011a,b). In order to assess the relationship between aerosols and meteorological variables regression analysis was made and plotted in Fig. 5aej. Wintertime coarser particulates were found to be significantly negatively correlated with the ambient temperature (r ¼ 0.57) and boundary layer height (ABL) (r ¼ 0.51). Both meteorological variables were found responsible for 26e33% of PM10 variability. This was according to the expectations as with gradual increase of temperature the possibility of enhanced wind circulation and simultaneous increase in ventilation coefficient rises which ultimately helps to increase in atmospheric dispersive capacity. A weak positive correlation between PM10 and relative humidity (RH, r ¼ 0.42, R2 ¼ 18%) was possibly due to the hygroscopicity of specific aerosol components. In our earlier study, a proportion of ambient aerosols were found to be constituted of water-soluble ionic constituents (PM10: 26.9%; PM2.5: 27.5%) (Murari et al., 2014) which due to its hygroscopic/sorption behavior gain additional weight in humid atmosphere. Negative correlation (r ¼ 0.32) between PM10 and WS indicates the influence of local sources including regional practices of biomass-waste incineration, vehicular emissions, and resuspension of crustal elements. However, poor coefficient of variation (11%) implies the possible existence of other significant sources for coarser particulates. Airborne

Fig. 2. Time series of wintertime airborne particulates over Varanasi.

166

M. Kumar et al. / Atmospheric Environment 104 (2015) 162e175

Fig. 4. Time series of Angstrom Exponent in respect to particulate ratio at ground monitoring station.

coarser particulates significantly scatter radiations resulting in visibility reduction on the event of higher particulate loadings. Therefore, computed correlation coefficient (r ¼ 0.60, R2 ¼ 36%) was according to expectations. The finer fractions of airborne particulates also denotes statistical association with prevalent meteorological variables. Finer particulates were found to be negatively correlated both with temperature (r ¼ 0.51) and ABL (r ¼ 0.47) with 22e30% of particulate variability was found to be associated with surface temperature and ABL height. Statistical association of air pollutants and meteorological variables is highly location specific and rely on existing meteorological characteristics. Nature of existing correlation between particulate and meteorological components for other stations is reported in Table 1. Visibility attenuation due to finer aerosols is an important index of urban air quality. During winter, fog and haze is a common phenomenon in IGP which potentially contributes to visibility reduction. A negative correlation between PM2.5 and visibility (r ¼ 0.52, R2 ¼ 27%) for Varanasi is in well agreement with certain expectations. To understand aerosol variability as a function of surface meteorology, 22-h particulate mass concentrations were plotted in reference to existing meteorology (Fig. 6). However, instead of presenting individual particulate mass loadings, time-averaged particulate mass concentration anomalies in respect to 90-days mean were compared. Such presentation of aerosol variability clearly establishes three distinguishable aerosol mass loading episodes, namely high aerosol loading days (HALDs, Julian days 1e23), medium aerosol loading days (MALDs, Julian days 24e60) and low aerosol loading days (LALDs, Julian days 61e90). These pollution episodes were distinctively articulated based on frequencies of how often particulate mean concentrations (PM10:233 ± 58.37; PM2.5:138 ± 47.12 mg m3) were exceeded. It is noteworthy to mention that such categorization were only intended to enhance the inter-comparability of aerosol loading days and were in line of thoughts of other contemporary researchers (Kim et al., 2014). The HALDs were characterized by high particulate concentrations of both coarser (298 ± 42 mg m3) as well as finer ones (163 ± 46 mg m3) with typical 100% (PM10) and 74% (PM2.5) exceedance over prevailing average. The events of HALDs were well supported by the persistence of shallow ABL depths (275.8 ± 65.5 m), high RH (85.55 ± 7.1%), lower surface temperature (14.58 ± 1.7  C) and moderate wind speed (1.2 ± 0.8 m s1) (Table 2). Prevailing minimum atmospheric dispersive capability (330.9 m2 s1) might have well-acted as a lid which subsequently increased the airborne particulate concentrations. Few counts of comparatively lower aerosol loadings days within HALDs were possibly due to occasional rainfall (2 counts, 12.2 mm) which

subsequently scavenged particulates off from the atmosphere. In between Julian days 24e60, relatively lower aerosols loading were accounted for both in coarser (222 ± 49 mg m3) and finer (148 ± 50 mg m3) fractions with lower exceedance level (PM10: 38%; PM2.5: 49%). This interlude was characterized by gradual increase in ABL depths (353.4 ± 114.2 m), surface temperature (17.34 ± 2.4  C), comparatively low RH (76.24 ± 10.1%) and prevailing moderate wind speed (1.4 ± 0.72 m s1). Relative increment of atmospheric ventilation coefficient (494.7 m2 s1) during MALDs eases the particulate dispersion and subsequently results in lower particulate mass loading. Additionally, the MALDs also experienced higher rainy days count (6, Julian days 44, 45, 53, 58e60), with aggregate rain fall of 57.4 mm. The concluding section of particulate monitoring period (LALDs, Julian days 61e90) distinctively experienced moderate particulate loading episodes (PM10: 195 ± 33 mg m3; PM2.5: 108 ± 24 mg m3) possibly due to persistence of higher atmospheric dispersive capability (1143.0 m2 s1). A strident increase in mean ABL depths (762.0 ± 227.3 m) in comparison to two other concurrent episodes with gradual rise of average surface wind speed (1.5 ± 0.9 m s1), temperature (24.6 ± 3.4  C) and minimum RH (57.54 ± 10.7%) possibly developed conducive atmospheric condition for better particulate mixing. Prevailing air pressure for the entire study period revealed minimum fluctuations (1012 ± 3.2 hPa) and therefore, statistical relations of surface pressure with airborne particulate were not plotted. 3.3. Trends in aerosol optical depths Daily variations of both space-based MODIS-AOD and groundbased MTP-AOD at 550 nm along with the daily particulate mass (PM) concentrations (PM10 and PM2.5) were plotted in Fig. 7. MODIS-AOD varied in a range of 0.259e2.194 (mean: 0.690 ± 0.464) while MTP-AOD prevailed within 0.066e1.239 (mean: 0.620 ± 0.240). MODIS-AOD essentially established a clear declining trend with respect to monitoring period typically identical to reduction of surface particulate mass loading for the region. A significant positive correlation (r ¼ 0.79) between these two variables reveal well agreement. During initial days of analysis (Julian days: 1e17), very high MODIS-AOD (range: 0.715e2.194, mean: 1.334 ± 0.594) were found to be associated with very high particulate mass loadings (PM10: 308 ± 40; PM2.5: 169 ± 48 mg m3). Contradictorily at later stage (Julian days: 68e82), both MODISAOD (range: 0.279e0.535, mean: 0.366 ± 0.077) and particulate mass loading (PM10: 196 ± 20; PM2.5: 100 ± 19 mg m3) reveal a decreasing trend. The features of MODIS-AOD temporal distribution were further reevaluated in respect to ground observations

M. Kumar et al. / Atmospheric Environment 104 (2015) 162e175

Fig. 5. (aej) Wintertime aerosol loading as function of micro-meteorological variables and visibility.

167

168

M. Kumar et al. / Atmospheric Environment 104 (2015) 162e175

Table 1 Correlation coefficient between particulate and meteorological components for other stations. Particulate parameters

Station

Temperature RH

PM10 PM2.5 PM2.5

Varanasi Varanasi Delhi

0.57 0.55 0.52

TSPM

Pantnagar 0.76

TSPM

Afyon

0.622

TSPM

Balikesir

0.30

WS

ABL

References

0.42 0.32 0.39 0.47 0.56 0.45

0.51 Present Study 0.30 Present Study 0.45 Tiwari et al. (2014) 0.25 0.11 e Banerjee et al. (2011a,b) _ g a and Sabah þ0.47 þ0.047 e Iça (2009) þ0.28 0.43 e Ilten and Selici (2008)

(MTP-AOD) and scatter plot between MODIS-AOD and MTP-AOD is given in Fig. 8. The observed MODIS-AOD trend was in identical with in-situ ground based MTP-AOD. For the HALDs episodes, very high particulate mass loading (PM10: 298 ± 42; PM2.5: 163 ± 46 mg m3) were characteristically associated with very high MODIS-AOD (1.230 ± 0.671) and MTP-AOD (0.827 ± 0.197). During MALDs reduction in both MODIS-AOD (0.711 ± 0.408) and MTPAOD (0.657 ± 0.276) were in continuation of reduced particulate mass loadings (PM10: 222 ± 49 mg m3; PM2.5: 148 ± 50 mg m3). Similar trends of reduced particulate distribution and their optical depths were also evident for LALDs episode. 3.4. Angstrom Exponent (AE) The value of AE, a quantitative measure of aerosol size distribution was in the order of 0.078e1.407 (mean: 1.002 ± 0.254) signifying substantial proportion of aerosol loadings were due to finer particulates of anthropogenic origin (Fig. 4). As seen in Fig. 4, temporal variation of AE was significant with dominance of coarser non-spherical particulate during initial HALDs episodes (0.934 ± 0.138) followed by gradual upsurge in finer particulates during MALDs (1.008 ± 0.320). Later (LALDs), AE was computed as 1.020 ± 0.207 signifying dominance of both finer and coarser particulates. The AE values were in accordance with expectations as prevalence of coarser particulate during HALDs may have originated through crustal resuspension and trans-boundary pollutants which further aggravated during MALDs through regional anthropogenic emissions. Further, AE was found to be inversely correlated (r ¼ 0.13) with AOD. 3.5. Association of AOD and surface particulate loading There has always been a desire for atmospheric scientists to essentially establish a link between space-borne retrievals of aerosol properties and ground based observations so that a costeffective monitoring approach may be developed. In absence of particle extinction coefficient and particle number count, here emphasis was made to establish AOD-PM relation based on statistical analysis. For the current experiment, ground based aerosol mass loadings were found to be consistent with the columnar AOD. Linear regression analysis was initially performed between surface mass concentrations and columnar AOD both from MODIS-AOD and MTP-AOD and presented in Fig. 9aed. Both coarser and finer particulates were found to be moderately correlated both in terms of MODIS-AOD (PM10 ¼ 0.46; PM2.5 ¼ 0.54) and MTP-AOD (PM10 ¼ 0.56; PM2.5 ¼ 0.39). Several instances are there when empirical relations between aerosol optical depth and particulate mass loading were established (Slater et al., 2004; Liu et al., 2007; Guo et al., 2009; Barladeanu et al., 2012). As seen in Table 3,

association between the MODIS-AOD and ground-based aerosol mass loading are highly location specific. High correlation was specifically observed in Northern Italy (r ¼ 0.82, Chu et al., 2003); Alabama (r ¼ 0.70, Wang and Christopher, 2003) and in Bucharest (r ¼ 0.89, Barladeanu et al., 2012) while moderate correlation was found in New Hampshire (R2 ¼ 0.66, Slater et al., 2004); Hong Kong (r > 0.5, Li et al., 2005a,b); Eastern China (r ¼ 0.52, Guo et al., 2009) and poor association was found in Taiwan (r ¼ 0.47, Tsai et al., 2011). Additionally, negative correlation was found in western United States (Engel-Cox et al., 2004). Principally, a high AOD-PM correlation is expected for a region where aerosols exist in well-mixed lower boundary layers during the satellite overpass time (likewise in Bucharest and Alabama) while presence of aerosols in higher altitude typically through long-range transport may generate poor correlations (likewise in Beijing and Taiwan). For the current submission, a moderate AOD-PM association for middle IGP instigate further research in altitudinal distribution of aerosol (Section 3.7). 3.6. Sensitivity of AOD-PM relations with meteorological variables Varying association of AOD and PM loading at the study site necessitated to assess the role of existing meteorological variables on AOD-PM relations. In principle, with increased RH, hygroscopic particle grows in size which subsequently changes its optical cross sections and simultaneously its ability to scatter light (Khoshsima et al., 2014). However, such theoretical explanation is not universal possibly due to some extraneous effects. For the current analysis, MODIS-AOD and MTP-AOD expressed variable correlations from poor (0.17) to strong (0.73) with ground level aerosol mass loading under different relative humidity conditions. During HALDs with high RH (mean: 85.55 ± 7.1%), MTP-AOD was having strong correlation with PM10 (r ¼ 0.73) but not with PM2.5 (r ¼ 0.14) while MODIS-AOD also exhibited poor association both in terms of PM10 or PM2.5. With decrease in RH (mean: 76.24 ± 10.1%) during MALDs, moderate to strong association of AOD-PM mass were obtained for both in terms of MODIS-AOD (PM10: 0.37; PM2.5: 0.50) and MTPAOD (PM10: 0.54; PM2.5: 0.40). Results clearly indicate that as per expectations, AOD-PM association increased with relative decrease in RH. Similar trends were also achieved by Gupta et al. (2006) and Khoshsima et al. (2014). However, during LALDs further decrease in RH (57.54 ± 10.7%) only results in reduced AOD-PM association. Possible reasons for such variability in AOD-PM association may be many. The most conceivable reason in terms of middle IGP is may be extraneous effects like altitudinal accumulation of aerosol. Multiple events of long range air mass transport both during HALDs and LALDs could lead to the insensitivity of AOD-PM relations with RH. The long range transport of air masses has been described under separate heading. Poor AOD-PM relations with high RH conditions were also reported by Khoshsima et al. (2014) at North Western Iran. Khoshsima et al. (2014) found moderate correlation of AOD-PM10 (r ¼ 0.52) with low RH while with increase in RH the correlation further recedes. For the entire monitoring period, surface wind speed variability was extremely low (1.2e1.5 m s1) which results in poor to moderate association of AOD-PM in context of WS. For HALDs, AOD-PM correlation in respect to WS was found to be highly variable (r ¼ 0.14e0.73) while relatively higher correlations of AOD-PM (r ¼ 0.37e0.54) were observed during MALDs. The possible reason of such low to moderate correlation may be regional practices of biomass-waste incineration during winter months which possibly exert a sudden local flush of finer particulates without possibly modifying long-term aerosol trends for the region. Additionally, convection of winds due to temperature variation and boundary layer fluctuations could simultaneously affect the

M. Kumar et al. / Atmospheric Environment 104 (2015) 162e175

169

Fig. 6. (aed) Variation of 24-h average aerosol mass loading with meteorological variables along with designation of aerosol loading episodes.

Table 2 Mean of all pre-identified variables for different aerosol loading periods. Period

RH (%)

Temp ( C)

ABL (m)

WS (m s1)

MODIS AOD

MTP AOD

PM10 (mg m3)

PM2.5 (mg m3)

HALD MALD LALD

85.55 ± 7.1 76.24 ± 10.1 57.54 ± 10.7

14.58 ± 1.7 17.34 ± 2.4 24.60 ± 3.4

275.8 ± 65.5 353.4 ± 114.2 762.0 ± 227.2

1.2 ± 0.8 1.4 ± 0.72 1.5 ± 0.9

1.230 ± 0.671 0.711 ± 0.408 0459 ± 0.149

0.827 ± 0.197 0.657 ± 0.273 0.510 ± 0.141

298 ± 42 222 ± 49 195 ± 33

163 ± 46 148 ± 51 108 ± 24

170

M. Kumar et al. / Atmospheric Environment 104 (2015) 162e175

Fig. 7. Daily variations of AOD from MODIS and MTP in reference to aerosol mass loading.

generation and transport of aerosols and change AOD-PM relations. Surface temperature is another significant variant that possibly modify AOD-PM association by changing existing ABL and atmospheric circulation. For the current analysis, increasing temperature from HALDs (mean: 14.58 ± 1.7  C) to MALDs (mean: 17.34 ± 2.4  C) had increased the correlations of MODIS-AOD and MTP-AOD with both PM10 and PM2.5. However, during LALDs AOD-PM correlations decreased with increase in the temperature (mean: 24.60 ± 3.4  C). Increase in surface temperature significantly modifies ABL and atmospheric ventilation coefficient which could have eased the movement of aerosols. Accumulation of such high-altitude aerosol may incur significant extinction coefficient but far enough to be measured by ground monitoring and lead to a weaker correlation between the aerosol optical properties and the surface aerosols. 3.7. Aerosol vertical profile In search of possible explanations for diverse AOD-PM association, aerosol vertical profiles for pre-identified transect were analyzed. Altitudinal distribution of aerosols is extremely critical for Varanasi as long-range transboundary particulates are believed to be significant contributors of total particulate loading (Murari et al., 2014; Sen et al., 2014). CALIPSO total (b0 532) and perpendicular (b0 532⊥) attenuated backscattering profiles comparison initially identifies two separated regions (X and Y). For region X (altitude: 1.2e5.4 km) both total (b0 532) and perpendicular (b0 532⊥) backscattering denotes enhanced signal possibly due to presence of non-spherical particles. For the same region (X) enhanced total backscattering for both b0 532 and total b0 1064 channel signify the presence of coarser particulates. In contrast, for a separate region Y (altitude: 0.1e4.2 km), comparison of enhanced signal in b0 532 channel with little or no enhancement in the b0 532⊥ channel denotes the presence of spherical particles while with reduced signal in b0 1064 channel denotes presence of finer particulates. Thus conclude the presence of non-spherical coarse particulates (possibly dust and cloud) in region X (altitude: 1.2e5.4 km) while (altitude: 0.1e4.2 km) dominance of spherical fine particulates (possibly of anthropogenic origin like smoke, polluted dust) for region Y. Additionally, lower DR profile (0.1e02) for region Y represent spherical particles mostly above the ABL (270 m) where as higher DR values (0.3e0.6) over region X signify dominance of non-spherical particles (Fig. 10d). The VFM profile (Fig. 10f) reconfirms the presence of huge volume of aerosol at selected transect. Level-2 CALIPSO aerosol subtype profile (Fig. 10e), eventually establish the presence of dust (subtype 2) in region X (altitude: 1.4e4.2 km) while enormous quantity of polluted

continental (subtype 3), polluted dust (subtype 5) and smoke (subtype 6) for region Y (altitude: 0.2e4.3 km). Collapse of ABL and reduced convective activity during winter nights (Satheesh et al., 2009) or long range transports (Sen et al., 2014) of aerosols are the possible reasons of aerosol dominance in such high altitude. These aerosols are well capable of modifying extinction of incident light and consequently modify total AOD. Identical approaches followed for other stations reveal similar conclusions likewise Bridhikitti (2013) for Bangkok Metropolitan Region, Kim et al. (2014) for Gosan climate observatory, Korea; Satheesh et al. (2009) for east coast of India.

3.8. Assessment of trans-boundary origin of particulates It was evident from CALIPSO profiles (Fig. 10) that substantial amount of aerosols with diverse morphological and chemical characteristics were present in various altitudes which of course not attributed through ground monitoring station. To establish the possible origin of such elevated-layer aerosol profile over Varanasi transect, lagrangian trajectory model NOAA-HYSPLIT with essential input from NCEP/NCAR reanalysis meteorological data was run for each individual aerosol loading episodes. Initially, five-day air mass back-trajectories at three different heights (1000 m, 2500 m and 3000 m) were simulated for each individual aerosol episodes (HALDs, MALDs and LALDs). As seen in Fig. 11aec, majority of the approaching air masses at 1000 m altitudes were continental and originated from various regions of IGP,

Fig. 8. MODIS AOD as a function of MTP AOD at ground monitoring station.

M. Kumar et al. / Atmospheric Environment 104 (2015) 162e175

171

Fig. 9. (aed) Scatter plot of MODIS and MTP AOD in reference to ground level aerosol loading.

Central India, Middle-east countries, Pakistan, Afghanistan and surrounding regions. However, proportions of wind during HALDs and MALDs also correspond to marine origin. Characteristically, for the same aerosol episodes implications of regional source contribution seems to be higher in comparison of LALDs episodes where particle profiles were primarily trans-boundary in nature. However, certain evidences of long range transport have also been observed during HALDs. Such observations were in accordance to expectations as ground observations during HALDs and LALDs depicts PM2.5/PM10 ratio as minimum (0.55) which may possibly due to accumulation of coarser particulates through trans-boundary movement. The same may be reestablished by observed AE profiles which validate the gross accumulation of coarser particulates both during HALDs (0.934 ± 0.138) and LALDs (1.020 ± 0.207). At higher altitudes i.e. ~2500 m and ~3000 m, most of the air masses

reaching the study site were originated from relatively longer distances (Fig. 11b and c). Contrary to air masses at 1000 m, the aerosol mass loadings were more contributed by air masses at 2500 m and 3000 m in all aerosol loading episodes. At 1000 m height, few signs of trans-boundary movement were seen during HALDs and mostly in LALDs, but at higher altitudes winds of greater magnitudes were coming to the study site from Arabian Sea, Middle East countries, Pakistan, Afghanistan, China and also from the Bay of Bengal. Possible accumulation of coarser particulates over IGP was also experienced by some other researchers. Venkataraman et al. (2005) reported a huge emission of BC and smoke from these regions due to biomass burning events. Tiwari et al. (2013) reported the aerosol loadings over IGP originating from different source regions while Sharma et al. (2014) reported similar trends of air mass movement in Delhi, India. Sen et al. (2014) through a coordinated study traced

Table 3 Correlations between AOD and particulate matter reported at different locations. Location

Source of AOD measurement

AOD vs PM10

AOD vs PM2.5

References

Varanasi Varanasi Northern Italy New Hampshire Alabama Hong-Kong Beijing Eastern China Bucharest Taiwan

MODIS Microtops Sun Photometer AERONET Multi-Filter Rotating Shadow band Radiometer (MFRSR) MODIS MODIS AERONET MODIS MODIS MODIS

r ¼ 0.46 r ¼ 0.55 r ¼ 0.82 e r ¼ 0.70 r > 0.5 r ¼ 0.59 e r ¼ 0.89 r ¼ 0.47

r ¼ 0.54 r ¼ 0.39 e r2 ¼ 0.66 e r > 0.5 e r ¼ 0.52 e r ¼ 0.65

Present Study Present Study Chu et al. (2003) Slater et al. (2004) Wang and Christopher (2003) Li et al. (2005a,b) Jiang et al. (2007) Guo et al. (2009) Barladeanu et al. (2012) Tsai et al. (2011)

172

M. Kumar et al. / Atmospheric Environment 104 (2015) 162e175

Fig. 10. (aef) CALIPSO profiles showing vertical distribution of aerosols at middle IGP. (a) Total Attenuated Backscatter at 532 nm, (b) Perpendicular Attenuated Backscatter at 532 nm, (c) Total Attenuated Backscatter at 1064 nm, (d) Depolarization Ratio (e) Aerosol Subtype (f) Vertical Feature Mask. The daytime CALIPSO altitude-orbit cross-section profile (version 3.30) was obtained on each 16-day CALIPSO repeat cycle, however, only plotted for 15-01-2014 at UTC 7:54.

the continental outflow of airborne particulates in IGP. The simulated particle back-trajectories eventually establish the meteorological influences of particle accumulation in middle IGP. However, in order to understand the mechanism in a broader perspectives, NCEP/NCAR reanalysis wind vector profiles was analyzed. NCEP/NCAR reanalysis wind vector data were purposively chosen at identical heights for which back-trajectories were developed. As seen in Fig. 12g and h, at ~1000 m altitude (925 mb) both HALDs and MALDs are characterized with moderate wind speed (calm-4.5 m s1) with highly variable wind directions. Moderate continental wind at lower altitudes possibly accumulates

anthropogenic fine particulates from western parts of country to middle IGP while strong continental northwesterly with moderately high wind speed (67 m s1) during LALDs (Fig. 12i) recognizes possible contribution of northwestern dry deserts. The entire situation predominantly remains identical at 2500 m (850 mb) with prominent contribution of continental northwesterly and proportion of southeasterly wind. Such prevailing wind vectors both at 1000 and 2500 m possibly explain the existing high altitudinal accumulation of aerosol at middle IGP. Aerosol accumulation even at higher altitudes (Fig. 10) may explicitly due to strong intercontinental westerly winds. A strong westerly wind

Fig. 11. (aec) HYSPLIT backward trajectories at 1000 m (a), 2500 m (b) and 3000 m (c) altitude for different aerosol loading days.

M. Kumar et al. / Atmospheric Environment 104 (2015) 162e175

173

Fig. 12. (aei) NCEP/NCAR composite means of geopotential heights and wind vector over south-east Asia.

(7e10 m s1) prevail around IGP at 700 mb carry coarser particulates from western deserts and middle-east countries and dump it into IGP (Fig. 12aec). However, the relative contribution of particulate loading possibly varied during MALDs due to low vector wind speeds (78 m s1). Trans-boundary origin and transport of airborne particulates were further assessed through Terra MODIS true color images over Indian subcontinent with its corresponding AOD images for three individual aerosol loading episodes (Fig. 13aef). The images are self-explanatory as they depict the presence of high turbid atmosphere that prevail over IGP. When these images are superimposed with NCEP/NCAR reanalysis wind vector profile (700 and 925 mb) previous explanation regarding possible origin and transport of airborne particulates fits well. 4. Conclusion To generate a coherent picture of regional interactions of

airborne particulates with existing meteorology and particle optical properties, both airborne PM2.5 and PM10 were monitored for entire winter months for a station that typically represents the IndoGangetic Plain. Initially particulate mass loadings from ground based in-situ measurement were analyzed in respect of regional meteorology. Further, in order to understand horizontal and vertical distribution of aerosol profiles, both space-borne passive and active sensor based measurements were concurrently used. The entire experimental setup was logically design to get the entire spectrum of airborne particulate sources, movements and interactions. The principal findings of the research may be summarized as follows: (a) The average wintertime mass concentrations of PM10 and PM2.5 were recorded as 233 ± 58.37 mg m3 and 138 ± 47.12 mg m3 at Varanasi which typically exceeds prescribed Indian, WHO, USEPA and EU standards. Not a

174

M. Kumar et al. / Atmospheric Environment 104 (2015) 162e175

Fig. 13. Terra MODIS true color and corresponding AOD images with vector wind profiles over Indian sub-continent during different aerosol loading episodes. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

(b)

(c)

(d)

(e)

single particulate loading for the entire monitoring period were found below regulatory limits. The similarity of origins of PM10 and PM2.5 could be anticipated based on the significant correlation observed between the two parameters (r ¼ 0.65). Additionally, daily deviations of PM2.5/PM10 ratio persist in a range of 0.25e0.97 (mean ¼ 0.60 ± 0.14; n ¼ 90) which represent greater association of fine than coarser ones. The sensitivity of airborne particulates in context of regional meteorology was also statistically assessed. As expected, both PM10 and PM2.5 were found to be negatively correlated with temperature, wind speed and ABL. Negative correlations between aerosol mass concentrations and wind speed indicate the dominance of local sources, however, degree of correlations necessitates further research. Gradual increase of ABL and atmospheric ventilation coefficient were found to enhance the particulate dispersion for each subsequent aerosol loading episodes and consequently results in lower particulate loading. An ensemble of both ground-based in-situ and space-borne satellite retrieved information has revealed optical and physical properties of Asian aerosols and synoptic pattern of regional and trans-boundary transport. Satellite retrieved Aqua MODIS-AOD (0.259e2.194) as well as ground measured MTP-AOD (0.006e1.239) was in agreement with the surface measured aerosol mass loading. Variability of both MODISAOD and MTP-AOD were found to be statistically more dependent on PM10 in contrast to PM2.5.

(f) Contributions of trans-boundary transport of aerosols to mass loadings at middle IGP were also recognized. CALIPSO altitude-orbit cross-section profiles revealed the presence of non-spherical coarse particulates (possibly dust) within altitude 1.2e5.4 km while dominance of spherical fine particulates (possibly of anthropogenic origin like smoke, polluted dust) in altitude: 0.1e4.2 km. (g) Persistence of high altitude aerosol mass loading in middle IGP was found to modify AOD-PM association within the region. (h) Cluster analysis of air mass back trajectories revealed the contribution of both marine and continental aerosol mass transportation from various regions of the IGP, middle-east Asia, central and southern India. The transboundary movement of aerosols of different size fractions was found to be governed by wind vectors at different geopotential heights. Mostly north-westerly winds at lower altitudes (~1000 m) contributed to the transport of finer aerosols from relatively lower distances where as high altitude westerlies (~3000 m) were found responsible for coarse aerosol mass loading in middle IGP. Acknowledgments Present submission is financially supported by University Grants Commission, New Delhi (F. No. 41-1111/2012, SR), Department of Science and Technology, New Delhi (F. No. SR/FTP/ES-52/2014) and by ISRO, Bangalore under ISRO-SSPS program. The Terra and Aqua MODIS AOD is courtesy of NASA's Earth-Sun System Division,

M. Kumar et al. / Atmospheric Environment 104 (2015) 162e175

distributed by LAADS Web. The CALIPSO data were obtained from the Atmospheric Science Data Center at NASA Langley Research Center. Meteorological data from wunderground.com is also acknowledged. Authors also acknowledge the NOAA-ARL for HYSPLIT transport model and NCEP/NCAR Reanalysis team for providing synoptic meteorological maps. The guidance and cooperation provided by Director, IESD-BHU is also acknowledged. References Banerjee, T., Singh, S.B., Srivastava, R.K., 2011a. Development and performance evaluation of statistical models correlating air pollutants and meteorological variables at Pantnagar, India. Atmos. Res. 99, 505e517. Banerjee, T., Barman, S.C., Srivastava, R.K., 2011b. Application of air pollution dispersion modeling for source-contribution assessment and model performance evaluation at integrated industrial estate-Pantnagar. Environ. Pollut. 159, 865e875. Barladeanu, R., Stefan, S., Radulescu, R., 2012. Correlation between the particulate matter (PM10) mass concentrations and aerosol optical depth in Bucharest, Romania. Rom. Rep. Phys. 64 (4), 1085e1096. Bilal, M., Nichol, J.E., Bleiweiss, M.P., Dubois, D., 2013. A simplified high resolution MODIS Aerosol Retrieval Algorithm (SARA) for use over mixed surfaces. Remote Sens. Environ. 136, 135e145. Bridhikitti, A., 2013. Atmospheric aerosol layers over Bangkok metropolitan region from CALIPSO observations. Atmos. Res. 127, 1e7. Chu, D.A., Kaufman, Y.J., Zibordi, G., Chern, J.D., Mao, J.M., Li, C., Holben, H.B., 2003. Global monitoring of air pollution over land from EOS-Terra MODIS. J. Geophys. Res. 108 (D21), 4661. Draxler, R.R., Rolph, G.D., 2003. HYSPLIT (HYbridSingle-Particle Lagrangian Integrated Trajectory) Model Access Via NOAA ARL READY Website. NOAA Air Resources Laboratory, Silver Spring, MD. http://www.arl.noaa.gov/ready/hysplit4. html. Engel-Cox, J.A., Holloman, C.H., Coutant, B.W., Hoff, R.M., 2004. Qualitative and quantitative evaluation of MODIS satellite sensor data for regional and urban scale air quality. Atmos. Environ. 38 (16), 2495e2509. Ghauri, B., Mansha, M., Khalil, C., 2012. Characterization of cytotoxicity of airborne particulates from urban areas of Lahore. J. Environ. Sci. 24 (11), 2028e2034. Guo, J.P., Zhang, X.Y., Che, H.Z., Gong, S.L., An, X., Cao, C.X., et al., 2009. Correlation between PM concentrations and aerosol optical depth in eastern China. Atmos. Environ. 43, 5876e5886. Gupta, P., Christopher, S.A., Wang, J., Gehrig, R., Lee, Y., Kumar, N., 2006. Satellite remote sensing of particulate matter and air quality assessment over global cities. Atmos. Environ. 40, 5880e5892. Ilten, N., Selici, T., 2008. Investigating the impacts of some meteorological parameters on air pollution in Balikesir, Turkey. Environ. Monit. Assess. 140, 267e277. _ g a, Y., Sabah, E., 2009. Statistical analysis of air pollutants and meteorological Iça parameters in Afyon, Turkey. Environ. Model. Assess. 14, 259e266. Jiang, X., Liu, Y., Yu, B., Jiang, M., 2007. Comparison of MISR aerosol optical thickness with AERONET measurements in Beijing metropolitan area. Remote Sens. Environ. 107, 45e53. Karar, K., Gupta, A.K., 2007. Source apportionment of PM10 at residential and industrial sites of an urban region of Kolkata, India. Atmos. Res. 84 (1), 30e41. Khoshsima, M., Givi, F.A., Bidokhti, A.A., Sabetghadam, S., 2014. Impact of meteorological parameters on relation between aerosol optical indices and air pollution in a sub-urban area. J. Aerosol Sci. 68, 46e57. Kim, Y., Kim, S.W., Yoon, S.C., Kim, M.H., Park, K.H., 2014. Aerosol properties and associated regional meteorology during winter pollution event at Gosan Climate Observatory, Korea. Atmos. Environ. 85, 9e17. Kittaka, C., Winker, D.M., Vaughan, M.A., Omar, A., Remer, L.A., 2011. Intercomparison of column aerosol optical depths from CALIPSO and MODIS-Aqua. Atmos. Meas. Tech. 4, 131e141. Kumar, N., Chu, D.A., Foster, A., 2007. An empirical relationship between PM2.5 and

175

aerosol optical depth in Delhi Metropolitan. Atmos. Environ. 4492e4503. Li, C.C., Mao, J.T., Lau, A.K.H., Yuan, Z.B., Wang, M.H., Liu, X.Y., 2005a. Application of MODIS satellite products on the air pollution research in Beijing. Sci. China Ser. D48 (Suppl. II), 209e219. Li, C., Lau, A.K.H., Mao, J., Chu, D.A., 2005b. Retrieval, validation, and application of the 1-km aerosol optical depth from MODIS measurements over Hong Kong. IEEE Trans. Geosci. Remote Sens. 43 (11), 2650e2658. Li, L., Huang, C., Huang, H.Y., Wang, Y.J., Yan, R.S., Zhang, G.F., et al., 2014. An integrated process rate analysis of a regional fine particulate matter episode over Yangtze River Delta in 2010. Atmos. Environ. 91, 60e70. Liu, Y., Koutrakis, P., Kahn, R., Turquety, S., Yantosca, R.M., 2007. Estimating fine particulate matter component concentrations and size distributions using satellite-retrieved fractional aerosol optical depth: part 2 e a case study. J. Air Waste Manag. Assoc. 57 (11), 1360e1369. Luvsan, M.E., Shie, R.H., Purevdorj, T., Badarch, L., Baldorj, B., Chan, C.C., 2012. The influence of emission sources and meteorological conditions on SO2 pollution in Mongolia. Atmos. Environ. 61, 542e549. Mittal, S.K., Singh, N., Agarwal, R., Awasthi, A., Gupta, P.K., 2009. Ambient air quality during wheat and rice crop stubble burning episodes in Patiala. Atmos. Environ. 43, 238e244. Morys, M., Mims, F.M., Hagerup, S., Anderson, S.E., Baker, A., Kia, J., et al., 2001. Design, calibration, and performance of MICROTOPS II handheld ozone monitor and sun photometer. J. Geophys. Res. 106, 14573e14582. Murari, V., Kumar, M., Barman, S.C., Banerjee, T., 2014. Temporal variability of MODIS aerosol optical depth and chemical characterization of airborne particulates in Varanasi, India. Environ. Sci. Pollut. Res. http://dx.doi.org/10.1007/ s11356-014-3418-2 (Published Online). Ram, K., Sarin, M.M., Tripathi, S.N., 2010. A 1 year record of carbonaceous aerosols from an urban site in the Indo-Gangetic Plain: characterization, sources, and temporal variability. J. Geophys. Res. 115, D24313. Raman, R.S., Ramachandran, S., Kedia, S., 2011. A methodology to estimate sourcespecific aerosol radiative forcing. J. Aerosol Sci. 42 (5), 305e320. Satheesh, S.K., Vinoj, V., Babu, S.S., Moorthy, K.K., Nair, V.S., 2009. Vertical distribution of aerosols over the east coast of India inferred from airborne LIDAR measurements. Ann. Geophys. 27, 4157e4169. Sen, A., Ahammed, Y.N., Arya, B.C., Banerjee, T., Begam, G.R., Baruah, B.P., et al., December 2014. Atmospheric fine and coarse mode aerosols at different environments of India and the Bay of Bengal during Winter-2014: implications of a coordinated campaign. MAPAN J. Metrol. Soc. India 29 (4), 273e284 (Published Online). Sharma, S.K., Kumar, M., Rohtash, Gupta, N.C., Saraswati, Saxena, M., Mandal, T.K., 2014. Characteristics of ambient ammonia over Delhi, India. Meteorol. Atmos. Phys. 124, 67e82. Slater, J.F., Dibb, J.E., Campbell, J.W., Moore, T.S., 2004. Physical and chemical properties of surface and column aerosols at a rural New England site during MODIS overpass. Remote Sens. Environ. 92, 173e180. Tiwari, S., Singh, A.K., 2013. Variability of aerosol parameters derived from ground and satellite measurements over Varanasi located in the Indo-Gangetic Basin. Aerosol Air Qual. Res. 13, 627e638. Tiwari, S., Srivastava, A.K., Singh, A.K., 2013. Heterogeneity in pre-monsoon aerosol characteristics over the Indo-Gangetic Basin. Atmos. Environ. 77, 738e747. Tiwari, S., Bisht, D.S., Srivastava, A.K., Pipal, A.S., Taneja, A., Srivastava, M.K., et al., 2014. Variability in atmospheric particulates and meteorological effects on their mass concentrations over Delhi, India. Atmos. Res. 145e146, 45e56. Tsai, T.C., Jeng, Y.J., Chu, D.A., Chen, J.P., Chang, S.C., 2011. Analysis of the relationship between MODIS aerosol optical depth and particulate matter from 2006 to 2008. Atmos. Environ. 45, 4777e4788. Venkataraman, C., Habib, G., Eiguren-Fernandez, A., Miguel, A.H., Friedlander, S.K., 2005. Residential biofuels in South Asia: carbonaceous aerosol emissions and climate impacts. Science 307, 1454e1456. Wang, L., Christopher, L., 2003. Inter-comparison between satellite-derived aerosol optical thickness and PM2.5 mass: implications for air quality studies. Geophys. Res. Lett. 30 (21).