lable at ScienceDirect

Environmental Pollution 231 (2017) 426e436

Contents lists avai

Environmental Pollution

journal homepage: www.elsevier .com/locate/envpol

Source apportionment of PM2.5 in North India using source-orientedair quality models*

Hao Guo a, Sri Harsha Kota b, Shovan Kumar Sahu b, Jianlin Hu c, Qi Ying d, Aifang Gao e, f,Hongliang Zhang a, c, *

a Department of Civil and Environmental Engineering, Louisiana State University, Baton Rouge, LA 70803, USAb Department of Civil Engineering, Indian Institute of Technology Guwahati, 781039, Indiac Jiangsu Key Laboratory of Atmospheric Environment Monitoring and Pollution Control, Jiangsu Engineering Technology Research Center of EnvironmentalCleaning Materials, Collaborative Innovation Center of Atmospheric Environment and Equipment Technology, School of Environmental Science andEngineering, Nanjing University of Information Science & Technology, 219 Ningliu Road, Nanjing 210044, Chinad Zachry Department of Civil Engineering, Texas A&M University, College Station, TX 77843, USAe School of Water Resources and Environment, Hebei GEO University, Shijiazhuang, Hebei Province 050031, Chinaf Hebei Key Laboratory of Sustained Utilization and Development of Water Resources, Shijiazhuang, Hebei Province 050031, China

a r t i c l e i n f o

Article history:Received 7 April 2017Received in revised form19 July 2017Accepted 4 August 2017

Keywords:Source apportionmentPM2.5

IndiaDelhiCMAQ

* This paper has been recommended for acceptance* Corresponding author. Department of Civil and

Louisiana State University, Baton Rouge, LA 70803, USE-mail address: hlzhang@lsu.edu (H. Zhang).

http://dx.doi.org/10.1016/j.envpol.2017.08.0160269-7491/© 2017 Elsevier Ltd. All rights reserved.

a b s t r a c t

In recent years, severe pollution events were observed frequently in India especially at its capital, NewDelhi. However, limited studies have been conducted to understand the sources to high pollutant con-centrations for designing effective control strategies. In this work, source-oriented versions of theCommunity Multi-scale Air Quality (CMAQ) model with Emissions Database for Global AtmosphericResearch (EDGAR) were applied to quantify the contributions of eight source types (energy, industry,residential, on-road, off-road, agriculture, open burning and dust) to fine particulate matter (PM2.5) andits components including primary PM (PPM) and secondary inorganic aerosol (SIA) i.e. sulfate, nitrateand ammonium ions, in Delhi and three surrounding cities, Chandigarh, Lucknow and Jaipur in 2015.PPM mass is dominated by industry and residential activities (>60%). Energy (~39%) and industry (~45%)sectors contribute significantly to PPM at south of Delhi, which reach a maximum of 200 mg/m3 duringwinter. Unlike PPM, SIA concentrations from different sources are more heterogeneous. High SIA con-centrations (~25 mg/m3) at south Delhi and central Uttar Pradesh were mainly attributed to energy, in-dustry and residential sectors. Agriculture is more important for SIA than PPM and contributions of on-road and open burning to SIA are also higher than to PPM. Residential sector contributes highest to totalPM2.5 (~80 mg/m3), followed by industry (~70 mg/m3) in North India. Energy and agriculture contribute~25 mg/m3 and ~16 mg/m3 to total PM2.5, while SOA contributes <5 mg/m3. In Delhi, industry and resi-dential activities contribute to 80% of total PM2.5.

© 2017 Elsevier Ltd. All rights reserved.

1. Introduction

Particulatematter (PM) can lead to reduced visibility (Ying et al.,2004), adversely affect human health (Pope and Dockery, 2006)and ecosystems (Qiao et al., 2015b) and change the earth's climateby perturbing the radiation balance (Charlson et al., 1992). PM withaerodynamic diameter of 2.5 mm or less (PM2.5) is harmful as it can

by Dr. Hageman Kimberly Jill.Environmental Engineering,A.

penetrate into lungs and be transported by bloodstream (Atkinsonet al., 2014; Bell et al., 2014; Kioumourtzoglou et al., 2016). PM2.5consists of primary PM (PPM), which is directly emitted into at-mosphere, and secondary PM, which is formed through chemicaland physical processes in atmosphere. The chemical composition ofPM2.5 is complex and typically includes elemental carbon (EC),primary organic carbon (POC), metals, sulfate (SO4

2�), nitrate (NO3�),

ammonium (NH4þ), and secondary organic aerosols (SOA). Sources,

chemical composition, formation, transformation and fate of PM2.5are quite different in different regions due to the variations inemissions and meteorological conditions (Querol et al., 2004;Shrivastava et al., 2015; Zhao et al., 2013).

H. Guo et al. / Environmental Pollution 231 (2017) 426e436 427

In India, the second most populous country in the world, rapidgrowth of industrialization and urbanization resulted in enormousincrease of anthropogenic emissions. Outdoor PM ranked the sev-enth in causes of death in India during 1990e2010 (IHME, 2013). In2010, out of 3.3 million global deaths due to outdoor PM2.5, around0.65 million deaths were in India of which 50% were due to resi-dential sector (Lelieveld et al., 2015). The situation in the Indiancapital has been alarming with extremely high PM2.5 concentra-tions. For example, annual PM2.5 concentrations in New Delhi was153 mg/m3 in 2014, more than 10 times higher than in WashingtonDC (WHO, 2014). Controlling PM2.5 concentrations can reduce thedeaths significantly. Sahu and Kota (2017) estimated that 41 out of100 thousand lives in Delhi could be saved by meeting the WorldHealth Organization (WHO) suggested PM2.5 annual average stan-dard of 10 mg/m3.

Contributions of different sources are important information forpolicy makers to formulate effective emission control strategies.Several source apportionment studies have been conducted in In-dia using different methods. For example, Srivastava et al. (2008)used principal component analysis (PCA) and found that crustalre-suspension is the major proportion of heavy metals in NewDelhi. Saxena et al. (2017) used PCA and concluded that secondaryaerosols, soil dust and biomass burning are the major sources ofwater soluble inorganic ions in PM2.5 of New Delhi, and theirfractional contributions are strongly dependent on seasons. Mandalet al. (2014) indicated that major parts of carbonaceous aerosols inPM2.5 in Delhi are from vehicles, coal smoke and biomass burningbased on measurement of EC to OC ratios. Sharma et al. (2016)applied positive matrix factorization (PMF) model to resolve ma-jor sources of PM2.5 as secondary aerosols, soil dust, vehicle emis-sions, biomass burning and fossil fuel combustion in New Delhi._ENREF_5These statistical methods are useful to understand thesources of PM2.5 at receptor locations, but the results are stronglydependent on availability of PM2.5 and its components data, andsometimes challenging to resolve sources to secondarycomponents.

Chemical transport models (CTMs) are widely used to analyzethe source origins of different air pollutants. Comprehensive airquality model with extensions (CAMx), coupled with plume risefunctions and hourly meteorology, has been used by Guttikundaand Jawahar (2014) to study PM2.5 related to coal-fired thermalpower plants nationwide in India and it was suggested thataggressive pollution control regulations were needed. Gupta andMohan (2013) predicted PM concentrations in New Delhi usingWeather Research and Forecasting Model (WRF-Chem), andobserved that emissions from North India was needed to improvethe performance of the model. Marrapu et al. (2014) used WRF-Chem model to study air quality during the CommonwealthGames in October 2010 and concluded that residential emissionswere the dominant source of PM2.5 in New Delhi and emissionsoutside New Delhi contributed to 20e50% of PM2.5 components.Source-oriented CTMs based on tagged tracer technique have beendeveloped and used for direct source apportionment of gas (Zhanget al., 2013; Zhang and Ying, 2011a) and particulate pollutants(Kleeman et al., 2007; Ying and Kleeman, 2006; Zhang et al., 2014).For example, using the source oriented UCD/CIT model, Zhang andYing (2010) found that road dust, diesel engines, internal com-bustion engines and coal burning are themajor sources for PPM, EC,primary organic carbon (POC) and SO4 in Southeast Texas. Shi et al.(2017) used source-oriented CMAQ system to quantify the contri-butions of different sources to PM2.5 in different provinces in China.Zhang et al. (2012) used the source-oriented Community MultiscaleAir Quality model (CMAQ) and observed that power generation isthe important source for SO4 and NO3 in China. Similar analysis byHu et al. (2015) discovered that residential/industrial emissions

from local and Hebei accounted for more than 90% of PPM inwinterat Beijing. Although many regional air quality studies were carriedout in Delhi and North India, a study using source-oriented CTMscan be a strong supplement to them.

In this paper, a source-oriented CMAQ model was applied toquantify the contributions of different source sectors to PM2.5 andits major components (PPM and SIA) in North India during 2015,covering New Delhi, Chandigarh, Lucknow and Jaipur. The seasonalvariations of contributions of different source sectors to total PM2.5and its components were also analyzed.

2. Methodology

2.1. Model description

The models used in this study were based on CMAQ 5.0.1 withthe SAPRC11 photochemical mechanism and aerosol moduleversion 6 (AERO6). The CMAQ model was modified to includeheterogeneous formation of SO4, NO3, and SOA formation fromsurface uptake (Hu et al., 2016; Ying et al., 2015). Source contri-butions of PPM and its chemical components were estimated usingtagged non-reactive PPM tracers. The tracers are set to 0.001% ofprimary emissions from each source sector and go through all at-mospheric processes same as other species. This small ratio doesnot significantly change particle size and mass. Then the PPMconcentrations from a given source is calculated by scaling thesimulated tracer concentrations from that source by 105, and sourceprofiles are used to estimate PPM components concentrations us-ing E1:

Ci,j ¼ PPMi � Ai,j (E1)

where Ci,j is component j concentration from source i, PPMj is theconcentration of total PPM from source i, and Ai,j is the ratio of jcomponent in PPM mass from source i. Details can be found in Huet al. (2015) and the references therein.

The source contributions to SIA were determined by trackingSO2, NOx, and NH3 through atmospheric processing using taggedreactive tracers. Both the photochemical mechanism and aerosolmodule were expanded so that SO4, NO3, and NH4 and their pre-cursors from different sources are tracked separately throughoutthe model calculations. Reactions R1, R2, and R3 show how thenitrate formation is tracked from NO2 reaction with hydroxylradical (OH) in original CMAQ (R1) to source-oriented version (R2and R3).

NO2 þ OH / HNO3(g)4 NO3� (R1)

In original CMAQ, HNO3(g) and NO3� are nitric acid gas and ni-

trate in PM.

NO2_X1 þ OH / HNO3_X1(g) 4 NO3�_X1 (R2)

NO2_X2 þ OH / HNO3_X2(g) 4 NO3�_X2 (R3)

In the source-oriented CMAQ, NO2 is expanded to two speciesNO2_X1 and NO2_X2, representing the emissions from two sources.R1 is then expanded to R2 and R3. NO3

�_X1 and NO3�_X2 represent

the contributions of NO2 from sources 1 and 2 to nitrate. Similartreatment is applied for all SIA precursors and related gas andaerosol processes. The readers are referred to previous studies fordetails (Qiao et al., 2015a; Zhang et al., 2014; Zhang et al., 2012).

SOA prediction has large uncertainties from the emissions of itsprecursors, unknown formation pathways, and limited observation(Hu et al., 2017; Ying et al., 2015; Zhang and Ying, 2011b). In this

Fig. 1. The 36-km domain (left) and 12-km resolution domain (right) with the locationof Chandigarh, Delhi, Jaipur and Lucknow. Color bar shows PM2.5 concentrations in mg/m3. (For interpretation of the references to colour in this figure legend, the reader isreferred to the web version of this article.)

H. Guo et al. / Environmental Pollution 231 (2017) 426e436428

study, the predicted SOA contributes <7% to total PM2.5, thus, it istreated as a separate “source type” while further studies areneeded. Its source apportionment using the tagged reactive tracersis still undergoing and will be provided in future manuscripts.

2.2. Model application

Fig. 1 shows the 36-km and 12-km domains selected for thesimulation in 2015 at India. The outer 36-km domain covered thewhole India and several adjacent countries and the inner 12-kmdomain covered areas around Delhi. Generation of the meteo-rology and emission inputs has been described in a companionmanuscript (Kota et al., 2017), and only a brief summary isdescribed here. The Weather Research & Forecasting model (WRF)v3.7.1 was utilized to generate meteorology inputs, and the

Fig. 2. Source apportionment of PPM in 12-km domain from sources types (a) energy, (b) indand (i) total. Units are in mg/m3.

monthly emissions from Emissions Database for Global Atmo-spheric Research (EDGAR) version 4.3 (http://edgar.jrc.ec.europa.eu/overview.php?v¼431) were used for anthropogenic emissions.The sectorial EDGAR inventories were then grouped into six broadsource categories: energy, industries, residential activities, on-roadtransportation, off-road transportation, and agriculture (Table S1).The monthly emissions were processed to hourly level based onweekly and diurnal emission profiles specific to sources from pre-vious studies (Olivier et al., 2003; Streets et al., 2003; Wang et al.,2010). Table S2 lists the total daily emission rates in 12-kmdomain of PM2.5, EC, OC, and gaseous species. Energy and in-dustry are the two largest anthropogenic sources of EC, while res-idential is the major source of OC. The top four source sectors ofPM2.5 emissions are energy (~5%), industry (~20%), residential(~55%) and agriculture (~15%), which explain 95% of total emissions.The Model for Emissions of Gases and Aerosols from Nature(MEGAN) v2.1 (Guenther et al., 2012) was used for biogenic emis-sions and open biomass burning emissions were generated fromthe Fire Inventory from NCAR (FINN), which is based on satelliteobservations (Wiedinmyer et al., 2011). Dust and sea salt emissionswere generated in line during simulations.

The default vertical distributions of concentrations that repre-sent clean continental conditions provided by the CMAQ modelwere used for 36-km domain initial and boundary conditions, andpredictions in the 36-km domain provides the boundary conditionsfor the 12-km domain (Hogrefe et al., 2017). The impact of initialconditions was minimal as the results of the first five days wereexcluded. Source apportionment simulations were conducted inthe 12-km domain and are presented in this paper.

The model performance for the entire year of 2015 has beenevaluated in a companion paper (Kota et al., 2017) and is summa-rized here. Generally, the WRF model performance on key

ustry, (c) residential, (d) on-road, (e) off-road, (f) agriculture, (g) open burning, (h) dust

Fig. 3. Seasonal variations of contributions of energy, industry, residential and agriculture to PPM concentrations. Units are mg/m3.

H. Guo et al. / Environmental Pollution 231 (2017) 426e436 429

meteorological parameters such as wind speed (WS), wind direc-tion (WD), temperature (T) and relative humidity (RH) is generallyacceptable accordingly to the criteria suggested by Emery et al.(2001) and comparable with similar studies in Asia (Hu et al.,2016). Model performance of major air pollutants including PM2.5also meets the criteria suggested by the US Environmental Pro-tection Agency as shown in Table S3. Specifically, there are twostations at Delhi and Lucknow located in the 12-km domain and themodel performance of PM2.5 is shown in Fig. S1. The model pre-diction meets the suggested performance criteria for most monthsexcept under-prediction at Lucknow in winter.

3. Results

3.1. Source apportionment of PPM

Fig. 2 shows regional variation of annual average source con-tributions to PPM from eight different sources, energy, industry,residential, on-road, off-road, agriculture, open burning and dust in2015. Energy, industry and residential sources contribute to 80% oftotal PPM in this domain. Dominance of energy and industrialsources is more obvious near New Delhi, Lucknow, Chandigarh,Jaipur and Lahore in Pakistan. South of Delhi is the most severelypolluted among all the regions. In south Delhi, energy sector, in-dustry and residential sources contributed to annual averaged

concentrations of 25, 53 and 40 mg/m3, respectively. Unlike energyand industry sectors, residential PPM is more spatially distributedand is high in Punjab, Haryana and parts of north Uttar Pradesh andEast Pakistan, which have high population density. The agriculturePPM is distributed evenly in Punjab, Haryana, Delhi and parts ofnorth Uttar Pradesh. Figs. S2 and S3, in the supplementary mate-rials, show the contributions of different sources to EC and POC.While industry (~7 mg/m3) is the largest source to EC, residentialsector (~20 mg/m3) is themain source for POC followed by industrialsector (~8 mg/m3).

Fig. 3 shows the seasonal variations in contributions to totalPPM from four source sectors of energy, industry, residential andagriculture, as the four source sectors contribute to 90% total PPM.PPM concentrations are higher at winter (December to February)followed by post-monsoon (September to November), and arelowest in monsoon (June to August) and pre-monsoon (March toMay). Low emissions and enhanced wet deposition are the mainreasons for the lowest PPM concentrations in monsoon, as shownin Fig. S4. The high residential PPM concentrations inwinter may bedue to the domestic heating in high population density areas atPunjab, Haryana, Delhi and parts of north Uttar Pradesh. Unlike thesignificant seasonal variation of the contributions from residentialsector, contributions from energy and industry sector are relativelysteady among different seasons. Agricultural PPM emissions are thehighest during pre-monsoon due to the residual burning (Awasthi

H. Guo et al. / Environmental Pollution 231 (2017) 426e436430

et al., 2010; Pandey et al., 2014). However, the agricultural PPMconcentrations are highest in winter indicating the importance ofmeteorology in this region. The seasonal variations in sourcecontribution of EC and POC (shown in Figs. S5 and S6) have verysimilar trend to total PPM.

Source apportionment of PPM in Delhi and three surroundingcities, Chandigarh, Lucknow and Jaipur are shown in Fig. 4 andFig. S7. Higher PPM concentrations are found in Delhi compared toother three cities. Residential sources are the dominant contributorto PPM in all cities. In Lucknow, Jaipur, Chandigarh and Delhi, res-idential sources contribute to 67, 57, 57 and 44%, respectively, toaverage PPM in 2015. Industries contribute to 47, 18, 31 and 20%,respectively, in Delhi, Chandigarh, Jaipur and Lucknow. While en-ergy is the thirdmost important source in Delhi, agriculture is moreimportant in the other three cities. The greatest contribution ofagriculture to PPM is in Chandigarh (17%). On-road PPM is at leasttwice off-road PPM on all seasons in Delhi, and the two sources are

Fig. 4. Daily contributions of PPM at New Delhi, Chandigarh, Jaipur and Lucknow cities fromburning and (h) dust. Units are mg/m3.

similar in the other cities. Although similar seasonal trends arefound in all these four cities, the contributions of sources to PPM aredifferent. The energy, industry and residential sectors at all citiesare maximum during post-monsoon and winter and least duringmonsoon. Highest daily PPM concentrations in Delhi, Lucknow,Chandigarh and Jaipur are 220, 92, 62 and 46 mg/m3, respectively.Contributions of industrial and residential sectors on days withhighest PPM concentrations in Delhi, Lucknow, Chandigarh andJaipur are 88, 90, 70 and 84%, respectively. Agricultural residueburning in India happens during OctobereNovember andAprileMay (Awasthi et al., 2010; Pandey et al., 2014). Unlike energy,residential and industrial PPM, which has higher concentrations inDecember than OctobereNovember, higher agriculture PPM isobserved in North India in winter. Large agriculture contribution isobserved across the year in Chandigarh in Fig. 4 and Fig. S7. Thecontribution of dust emissions is the highest in monsoon and theleast in winter at all the cities. In Delhi, Lucknow, Jaipur and

(a) energy, (b) industry, (c) residential, (d) on-road, (e) off-road, (f) agriculture, (g) open

Fig. 5. Source apportionment of annual SIA in 12-km domain from source types (a) energy, (b) industry, (c) residential, (d) on-road, (e) off-road, (f) agriculture, (g) open burning, (h)dust and (i) total. Units are mg/m3.

H. Guo et al. / Environmental Pollution 231 (2017) 426e436 431

Chandigarh, dust emissions contributed much higher to PPM inmonsoon than in winter. The maximum contribution of dustemissions to PPM (84%) is observed at Jaipur, which is in closeproximity to Thar Desert. This region experiences dust-storms,thunderstorms and dust raising winds during the hot wet season,which brings high dust contributions. It could also due to un-certainties in WRF predicted soil moisture used for dust emissionsestimation in East Asia as reported in a previous study (Darmenovaet al., 2009).

3.2. Source apportionment of SIA

Previous studies indicate that SIA account for a significantfraction of total PM2.5 (Wang et al., 2014; Zhang et al., 2012). Fig. 5shows the annual averaged regional contribution of each source toSIA in the 12-km domain. Similar to PPM, energy, industry, resi-dential and agriculture are the major source sectors for SIA. How-ever, unlike PPM, SIA concentrations from different sources aremore distributed due to their secondary nature. High SIA concen-trations (~30 mg/m3) are observed at south of Delhi and centralUttar Pradesh in sectors of energy, industry and residential, whilePunjab, Haryana, and parts of north Uttar Pradesh have a relativelylower SIA concentration (~10 mg/m3) in these three categories.Agriculture SIA is distributed evenly along Punjab, Haryana, Delhiand north Uttar Pradesh. Overall, from Figs. 2 and 5, relative con-tributions of agriculture emissions to SIA is higher than their con-tributions to PPM as the major of agriculture emission are NH3 andNO2, which are precursors to NH4

þ and NO3� in SIA (Paulot and Jacob,

2014).As Fig. S8 shows, energy is the dominant source of nitrate. Ni-

trate concentrations from energy production (~2 mg/m3) areconcentrated at south of Delhi as several coal-based power plants

located there (Guttikunda and Jawahar, 2014; Prasad et al., 2006).Moreover, the contribution of on-road sources to nitrate is alsosignificant. Also, energy is the major source of sulfate as shown inFig. S9, and agriculture is the major source of ammonium ion asshown in Fig. S10. As the dust sector has no gaseous precursoremissions, it does not contribute to sulfate and ammonium ion andthus is not shown in Figs. S9 and S10. Overall, it can be concludedthat control of energy and agriculture sources is expected to beeffective for reducing SIA and its components in this region.

Fig. 6 presents seasonal variation of SIA concentrations based onthe top four source sectors: energy, industry, residential and agri-culture. Similar to PPM, SIA concentrations are higher in winterthan in other seasons because of higher emission resulting fromresidential heating in winter and unfavorable meteorological con-ditions for dispersion of pollutants as reported by previous studies(Kulshrestha et al., 1995; Norman et al., 2001). Industrial and resi-dential sectors have the peak concentration (~5 mg/m3) at south ofDelhi in winter while high concentration occurs along Punjab toNorth Uttar Pradesh in winter for energy and agriculture sector(~6 mg/m3). Fig. S11 shows seasonal variation of nitrate concentra-tions from these sectors. Compared to total SIA, nitrate has moresignificant seasonal variation as the extreme low concentrationsare observed in monsoon season from all these sectors. However,the peak values still occur in winter along Punjab to north UttarPradesh. Energy and industry sectors have higher sulfate concen-trations in pre-monsoon and monsoon at central Uttar Pradesh asobserved in Fig. S12. Residential and agriculture sectors do not haveobservable seasonal variations. The seasonal variation of ammo-nium ion is quite similar to that of nitrate and sulfate, even thoughthere are significant point sources in energy and industry sectorsand the major source of ammonium (agriculture) is more distrib-uted as shown in Fig. S13.

Fig. 6. Seasonal variation of contributions of energy, industry, residential and agriculture to SIA concentrations (Units are mg/m3).

H. Guo et al. / Environmental Pollution 231 (2017) 426e436432

Fig. 7 and Fig. S14 shows source apportionment of SIA in Delhi,Chandigarh, Lucknow and Jaipur. Higher SIA is found in Delhi, fol-lowed by Lucknow, Chandigarh and Jaipur. While energy sector isthe main contributor to SIA in Delhi and Jaipur, agriculture is themain source of SIA in Chandigarh and Lucknow. Energy and resi-dential sources together contribute to 45, 65, 61 and 55% of SIA inDelhi, Chandigarh, Jaipur and Lucknow, respectively. Unlike othercities, in Delhi, industry and residential sources are more importantthan agriculture. The relative contribution of on-road and openburning sources is more significant for SIA than PPM. Chandigarhhad highest contributions from open burning, i.e. 1.97%. Contribu-tion of SIA from agriculture is higher during biomass burning sea-sons, i.e. AprileMay and OctobereNovember, in all cities. Amongindustrial, residential and energy sources, while energy sourcesdominate during pre-monsoon and monsoon, contribution of res-idential sources to total SIA is maximum in winter and post-monsoon. On-road SIA reaches the highest concentrations inwinter, with maximum contributions being 20, 32, 29 and 53%,respectively.

3.3. Source apportionment of total PM2.5

PPM and SIA are the two major components of total PM2.5 inIndia (Kota et al., 2017). Fig. 8 shows the annual contributions of

different sources and SOA to total PM2.5. Residential sector con-tributes highest to total PM2.5 about ~80 mg/m3, followed by in-dustry sector (~70 mg/m3). Energy sectors and agriculture sectorcontribute to ~25 mg/m3 and ~16 mg/m3. Energy and industry con-centrations have a significant high concentration point at south ofDelhi and its surroundings. On the contrary, SOA, residential andagriculture sector distributed evenly at Indo-Gangetic plain. Addi-tionally, residential sources also peaked in north-Pakistan.

Fig. 9 shows the contributions of different sources to PM2.5 totalmass at selected four cities. Generally, residential sector contributesthemost to total PM2.5 followed by industry, energy and agriculturesectors. Energy, residential and industry sources contribute to atleast 75% of total PM2.5 in all the cities. Contribution of SOA ismaximum in Jaipur, 7%, and least in Delhi, 3%. Fig. S15 shows thecomparison of the relative source contributions to PM2.5 at Delhiusing PMF model for 2013e2014 (Sharma et al., 2016) and thesource-oriented CMAQ model in the present study for 2015. To beconsistent, sources in the present study e.g., residential, energy andagriculture were further reclassified as biomass and fossil fuel inTable S4 using the energy consumption data provided by the Indiangovernment (www.indiaenvironmentportal.org.in/files/file/pngstat.pdf). The source contributions estimated by these twomethods are generally consistent at Delhi expect for traffic andbiomass burning categories. The difference may be due to coarse

Fig. 7. Daily contributions of different sectors to SIA at specific cities from source types (a) energy, (b) industry, (c) residential, (d) on-road, (e) off-road, (f) agriculture (g) openburning and (h) dust. Units are mg/m3.

H. Guo et al. / Environmental Pollution 231 (2017) 426e436 433

grid (12-km) of emission used in our study and uncertaintiescontained in PMF model used in Sharma et al. (2016). According toanother study (Chowdhury et al., 2007), fossil fuel combustion isresponsible for about 25e33% of PM2.5 mass in Delhi, and 28% inChandigarh, which is consistent with our results: 20.7% in Delhi and23.2% in Chandigarh. The agreement with other studies at differentcities shows the reliability of the source apportionment results ofthis study.

4. Conclusion

A source-oriented CMAQmodeling system driven by the off-line

meteorological inputs from the WRF model was used to quantifythemajor source contributions to primary, secondary inorganic andtotal PM2.5 in Delhi and surrounding cities in North India in 2015.This study finds that industrial and residential activities are thedominating sources (60e70%) for EC and POC, while energy andagriculture are also important sources to SIA and its componentsi.e. nitrate and sulfate. The spatial distribution of energy and in-dustry sectors shows significant point sources at south of Delhi inboth PPM and SIA source apportionment analysis while residentialand agriculture sectors are distributed evenly in the Indo-Gangeticplain. A strong seasonal variation in the sectors’ contribution toPPM and SIA is also predicted. Inmost areas, the peak concentration

Fig. 8. Total PM2.5 source apportionment in 12-km domain from source types (a) energy, (b) industry, (c) residential, (d) on-road, (e) off-road, (f) agriculture, (g) open burning, (h)dust and (i) SOA concentrations. Units are mg/m3.

H. Guo et al. / Environmental Pollution 231 (2017) 426e436434

is observed at winter followed by post-monsoon and pre-monsoonseasons and lowest at monsoon in all sources. All the selected citiesare all facing severe PM2.5 pollution during winter. PPM concen-tration can reach ~200 mg/m3 at Delhi. Contributions of agricultureto PPM and SIA were higher during OctobereNovember coincidingwith the residual burning. Industry, residential and energy are thethree major sources to PM2.5 in all the cities. The variation in theseasonal, spatial and source sector contributions emphasizes theimportance of a better understanding the sources of PPM, SIA and

Fig. 9. Contributions of different source se

its components when designing efficient regional emission controlstrategies towards reducing severe air pollution issues occurs atDelhi. Future studies should be carried out to better interpolate theresults of this study, including obtaining source apportionment ofSOA in this region, conducting correction for the bias error (Solazzoet al., 2017), and considering feedbacks of aerosols on meteorologyusing online coupled source-oriented models in future (Im et al.,2015).

ctors to total PM2.5 at selected cities.

H. Guo et al. / Environmental Pollution 231 (2017) 426e436 435

Acknowledgement

Portions of this researchwere conductedwith high performancecomputing resources provided by Louisiana State University(http://www.hpc.lsu.edu) and Indian Institute of Technology,Guwahati (http://www.iitg.ernet.in/param-ishan/index.html). Theproject is funded by European Climate Foundation (G-1606-00917).Open fund by Jiangsu Key Laboratory of Atmospheric EnvironmentMonitoring and Pollution Control (KHK1512), A Project Funded bythe Priority Academic Program Development of Jiangsu HigherEducation Institutions (PAPD).

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.envpol.2017.08.016.

References

Atkinson, R., Kang, S., Anderson, H., Mills, I., Walton, H., 2014. Epidemiological timeseries studies of PM2.5 and daily mortality and hospital admissions: a system-atic review and meta-analysis. Thorax 69 (7), 660e665. http://dx.doi.org/10.1136/thoraxjnl-2013-204492.

Awasthi, A., Singh, N., Mittal, S., Gupta, P.K., Agarwal, R., 2010. Effects of agriculturecrop residue burning on children and young on PFTs in North West India. Sci.Total Environ. 408, 4440e4445.

Bell, M.L., Ebisu, K., Leaderer, B.P., Gent, J.F., Lee, H.J., Koutrakis, P., Wang, Y.,Dominici, F., Peng, R.D., 2014. Associations of PM2.5 constituents and sourceswith hospital admissions: analysis of four counties in Connecticut and Massa-chusetts (USA) for Persons> or¼ 65 Years of age. Environ. Health Perspect. 122,138 (Online).

Charlson, R.J., Schwartz, S.E., Hales, J.M., Cess, R.D., Coakley, J.A., Hansen, J.E.,Hofmann, D.J., 1992. Climate forcing by anthropogenic aerosols. Science 255,423e430.

Chowdhury, Z., Zheng, M., Schauer, J.J., Sheesley, R.J., Salmon, L.G., Cass, G.R.,Russell, A.G., 2007. Speciation of ambient fine organic carbon particles andsource apportionment of PM2.5 in Indian cities. J. Geophys. Res. Atmos. 112.

Darmenova, K., Sokolik, I.N., Shao, Y., Marticorena, B., Bergametti, G., 2009. Devel-opment of a physically based dust emission module within the WeatherResearch and Forecasting (WRF) model: assessment of dust emission parame-terizations and input parameters for source regions in Central and East Asia.J. Geophys. Res. Atmos. 114.

Emery, C., Tai, E., Yarwood, G., 2001. Enhanced Meteorological Modeling and Per-formance Evaluation for Two TEXAS Episodes. Report to the Texas Natural Re-sources Conservation Commission. prepared by ENVIRON, International Corp,Novato, CA. available at: http://www.tceq.state.tx.us/assets/public/implementation/air/am/contracts/reports/mm/EnhancedMetModelingAndPerformanceEvaluation.pdf.

Guenther, A., Jiang, X., Heald, C., Sakulyanontvittaya, T., Duhl, T., Emmons, L.,Wang, X., 2012. The Model of Emissions of Gases and Aerosols from NatureVersion 2.1 (MEGAN2. 1): an Extended and Updated Framework for ModelingBiogenic Emissions.

Gupta, M., Mohan, M., 2013. Assessment of contribution to PM10 concentrationsfrom long range transport of pollutants using WRF/Chem over a subtropicalurban airshed. Atmos. Pollut. Res. 4, 405e410.

Guttikunda, S.K., Jawahar, P., 2014. Atmospheric emissions and pollution from thecoal-fired thermal power plants in India. Atmos. Environ. 92, 449e460.

Hogrefe, C., Roselle, S.J., Bash, J.O., 2017. Persistence of initial conditions in conti-nental scale air quality simulations. Atmos. Environ. 160, 36e45.

Hu, J., Chen, J., Ying, Q., Zhang, H., 2016. One-year simulation of ozone and partic-ulate matter in China using WRF/CMAQ modeling system. Atmos. Chem. Phys.Discuss. 2016, 1e31.

Hu, J., Wang, P., Ying, Q., Zhang, H., Chen, J., Ge, X., Li, X., Jiang, J., Wang, S., Zhang, J.,Zhao, Y., Zhang, Y., 2017. Modeling biogenic and anthropogenic secondaryorganic aerosol in China. Atmos. Chem. Phys. 17, 77e92.

Hu, J., Wu, L., Zheng, B., Zhang, Q., He, K., Chang, Q., Li, X., Yang, F., Ying, Q.,Zhang, H., 2015. Source contributions and regional transport of primary par-ticulate matter in China. Environ. Pollut. 207, 31e42.

IHME, 2013. India Global Burden of Disease Study 2010. Institue for Health Metricsand Evaluation, Seattle, United States.

Im, U., Bianconi, R., Solazzo, E., Kioutsioukis, I., Badia, A., Balzarini, A., Bar�o, R.,Bellasio, R., Brunner, D., Chemel, C., 2015. Evaluation of operational on-line-coupled regional air quality models over Europe and North America in thecontext of AQMEII phase 2. Part I Ozone. Atmos. Environ. 115, 404e420.

Kioumourtzoglou, M.-A., Schwartz, J.D., Weisskopf, M.G., Melly, S.J., Wang, Y.,Dominici, F., Zanobetti, A., 2016. Long-term PM2.5 exposure and neurologicalhospital admissions in the northeastern United States. Environ. Health Perspect.124, 23.

Kleeman, M.J., Ying, Q., Lu, J., Mysliwiec, M.J., Griffin, R.J., Chen, J.J., Clegg, S., 2007.

Source apportionment of secondary organic aerosol during a severe photo-chemical smog episode. Atmos. Environ. 41, 576e591.

Kota, S.H., Guo, H., Hu, J., Sahu, S.K., Ying, Q., Zhang, H., 2017. Year-long simulation ofgaseous and particulate air pollutants in India. Environ. Pollut. Submitted).

Kulshrestha, U., Kumar, N., Saxena, A., Kumari, K., Srivastava, S., 1995. Identificationof the nature and source of atmospheric aerosols near the Taj Mahal (India).Environ. Monit. Assess. 34, 1e11.

Lelieveld, J., Evans, J., Fnais, M., Giannadaki, D., Pozzer, A., 2015. The contribution ofoutdoor air pollution sources to premature mortality on a global scale. Nature525, 367e371.

Mandal, P., Sarkar, R., Mandal, A., Saud, T., 2014. Seasonal variation and sources ofaerosol pollution in Delhi, India. Environ. Chem. Lett. 12, 529e534.

Marrapu, P., Cheng, Y., Beig, G., Sahu, S., Srinivas, R., Carmichael, G.R., 2014. Airquality in Delhi during the Commonwealth games. Atmos. Chem. Phys. 14,10619e10630.

Norman, M., Das, S., Pillai, A., Granat, L., Rodhe, H., 2001. Influence of air masstrajectories on the chemical composition of precipitation in India. Atmos. En-viron. 35, 4223e4235.

Olivier, J., Peters, J., Granier, C., Petron, G., Müller, J., Wallens, S., 2003. Present andFuture Surface Emissions of Atmospheric Compounds. POET report {#} 2, EUproject EVK2-1999-00011. Tech. rep., European Union.

Pandey, A., Sadavarte, P., Rao, A.B., Venkataraman, C., 2014. Trends in multi-pollutant emissions from a technology-linked inventory for India: II. Residen-tial, agricultural and informal industry sectors. Atmos. Environ. 99, 341e352.

Paulot, F., Jacob, D.J., 2014. Hidden cost of US agricultural exports: particulate matterfrom ammonia emissions. Environ. Sci. Technol. 48, 903e908.

Pope, C.A., Dockery, D.W., 2006. Health effects of fine particulate air pollution: linesthat connect. J. Air Waste Manag. Assoc. 56, 709e742.

Prasad, A.K., Singh, R.P., Kafatos, M., 2006. Influence of coal based thermal powerplants on aerosol optical properties in the Indo-Gangetic basin. Geophys. Res.Lett. 33.

Qiao, X., Tang, Y., Hu, J., Zhang, S., Li, J., Kota, S.H., Wu, L., Gao, H., Zhang, H., Ying, Q.,2015a. Modeling dry and wet deposition of sulfate, nitrate, and ammonium ionsin Jiuzhaigou National Nature Reserve, China using a source-oriented CMAQmodel: Part I. Base case model results. Sci. Total Environ. 532, 831e839.

Qiao, X., Xiao, W., Jaffe, D., Kota, S.H., Ying, Q., Tang, Y., 2015b. Atmospheric wetdeposition of sulfur and nitrogen in Jiuzhaigou national nature reserve, Sichuanprovince, China. Sci. Total Environ. 511, 28e36.

Querol, X., Alastuey, A., Ruiz, C.R., Arti~nano, B., Hansson, H.C., Harrison, R.M.,Buringh, E., ten Brink, H.M., Lutz, M., Bruckmann, P., Straehl, P., Schneider, J.,2004. Speciation and origin of PM10 and PM2.5 in selected European cities.Atmos. Environ. 38, 6547e6555.

Sahu, S.K., Kota, S.H., 2017. Significance of PM2.5 air quality at the Indian capital.Aerosol Air Qual. Res. 17, 588e597.

Saxena, M., Sharma, A., Sen, A., Saxena, P., Saraswati, Mandal, T.K., Sharma, S.K.,Sharma, C., 2017. Water soluble inorganic species of PM10 and PM2.5 at an urbansite of Delhi, India: seasonal variability and sources. Atmos. Res. 184, 112e125.

Sharma, S.K., Mandal, T.K., Jain, S., Saraswati, Sharma, A., Saxena, M., 2016. Sourceapportionment of PM2.5 in Delhi, India using PMF model. Bull. Environ. Contam.Toxicol. 97, 286e293.

Shi, Z., Li, J., Huang, L., Hu, J., Wu, L., Ying, Q., Zhang, H., Lu, L., Liu, X., Liao, H., 2017.Source apportionment of fine particulate matter in China in 2013 using asource-oriented chemical transport model. Sci. Total Environ. Submitted).

Shrivastava, M., Easter, R.C., Liu, X., Zelenyuk, A., Singh, B., Zhang, K., Ma, P.L.,Chand, D., Ghan, S., Jimenez, J.L., 2015. Global transformation and fate of SOA:implications of low-volatility SOA and gas-phase fragmentation reactions.J. Geophys. Res. Atmos. 120, 4169e4195.

Solazzo, E., Bianconi, R., Hogrefe, C., Curci, G., Tuccella, P., Alyuz, U., Balzarini, A.,Bar�o, R., Bellasio, R., Bieser, J., 2017. Evaluation and error apportionment of anensemble of atmospheric chemistry transport modeling systems: multivariabletemporal and spatial breakdown. Atmos. Chem. Phys. 17, 3001e3054.

Srivastava, A., Gupta, S., Jain, V.K., 2008. Source apportionment of total suspendedparticulate matter in coarse and fine size ranges over Delhi. Aerosol Air Qual.Res. 8, 188e200.

Streets, D.G., Bond, T., Carmichael, G., Fernandes, S., Fu, Q., He, D., Klimont, Z.,Nelson, S., Tsai, N., Wang, M.Q., 2003. An inventory of gaseous and primaryaerosol emissions in Asia in the year 2000. J. Geophys. Res. Atmos. 108.

Wang, D., Hu, J., Xu, Y., Lv, D., Xie, X., Kleeman, M., Xing, J., Zhang, H., Ying, Q., 2014.Source contributions to primary and secondary inorganic particulate matterduring a severe wintertime PM2.5 pollution episode in Xi'an, China. Atmos.Environ. 97, 182e194.

Wang, X., Liang, X.-Z., Jiang, W., Tao, Z., Wang, J.X., Liu, H., Han, Z., Liu, S., Zhang, Y.,Grell, G.A., 2010. WRF-Chem simulation of East Asian air quality: sensitivity totemporal and vertical emissions distributions. Atmos. Environ. 44, 660e669.

WHO, 2014. World Health Statistics 2014.Wiedinmyer, C., Akagi, S.K., Yokelson, R.J., Emmons, L.K., Al-Saadi, J.A., Orlando, J.J.,

Soja, A.J., 2011. The Fire INventory from NCAR (FINN): a high resolution globalmodel to estimate the emissions from open burning. Geosci. Model Dev. 4,625e641.

Ying, Q., Kleeman, M.J., 2006. Source contributions to the regional distribution ofsecondary particulate matter in California. Atmos. Environ. 40, 736e752.

Ying, Q., Li, J., Kota, S.H., 2015. Significant contributions of isoprene to summertimesecondary organic aerosol in eastern United States. Environ. Sci. Technol. 49,7834e7842.

Ying, Q., Mysliwiec, M., Kleeman, M.J., 2004. Source apportionment of visibility

H. Guo et al. / Environmental Pollution 231 (2017) 426e436436

impairment using a three-dimensional source-oriented air quality model. En-viron. Sci. Technol. 38, 1089e1101.

Zhang, H., Hu, J., Kleeman, M., Ying, Q., 2014. Source apportionment of sulfate andnitrate particulate matter in the Eastern United States and effectiveness ofemission control programs. Sci. Total Environ. 490, 171e181.

Zhang, H., Li, J., Ying, Q., Guven, B.B., Olaguer, E.P., 2013. Source apportionment offormaldehyde during TexAQS 2006 using a source-oriented chemical transportmodel. J. Geophys. Res. Atmos. 118, 1525e1535.

Zhang, H., Li, J., Ying, Q., Yu, J.Z., Wu, D., Cheng, Y., He, K., Jiang, J., 2012. Sourceapportionment of PM2.5 nitrate and sulfate in China using a source-orientedchemical transport model. Atmos. Environ. 62, 228e242.

Zhang, H., Ying, Q., 2010. Source apportionment of airborne particulate matter inSoutheast Texas using a source-oriented 3D air quality model. Atmos. Environ.44, 3547e3557.

Zhang, H., Ying, Q., 2011a. Contributions of local and regional sources of NOx toozone concentrations in Southeast Texas. Atmos. Environ. 45, 2877e2887.

Zhang, H., Ying, Q., 2011b. Secondary organic aerosol formation and sourceapportionment in Southeast Texas. Atmos. Environ. 45, 3217e3227.

Zhao, P., Dong, F., He, D., Zhao, X., Zhang, X., Zhang, W., Yao, Q., Liu, H., 2013. Charac-teristics of concentrations and chemical compositions for PM2.5 in the region ofBeijing, Tianjin, and Hebei, China. Atmos. Chem. Phys. 13, 4631e4644.