Preparation and Validation of Gridded Emission Inventoryof Criteria Air Pollutants and Identification of EmissionHotspots for Megacity Delhi

Manju Mohan & Lalit Dagar & B. R. Gurjar

Received: 8 June 2006 /Accepted: 17 July 2006# Springer Science + Business Media B.V. 2006

Abstract Delhi is one of the many megacities strug-gling with punishing levels of pollution from industrial,residential, and transportation sources. Over the years,pollution abatement in Delhi has become an importantconstituent of state policies. In the past one decade a lotof policies and regulations have been implementedwhich have had a noticeable effect on pollution levels.In this context, air quality models provide a powerfultool to study the impact of development plans on theexpected air pollution levels and thus aid the regulatingand planning authorities in decision-making process. Inair quality modeling, emissions in the modeling domainat regular interval are one of the most important inputs.From the annual emission data of over a decade (1990–2000), emission inventory is prepared for the megacityDelhi. Four criteria pollutants namely, CO, SO2, PM,and NOx are considered and a gridded emissioninventory over Delhi has been prepared taking intoaccount land use pattern, population density, trafficdensity, industrial areas, etc. A top down approach isused for this purpose. Emission isopleths are drawn

and annual emission patterns are discussed mainly forthe years 1990, 1996 and 2000. Primary and secondaryareas of emission hotspots are identified and emissionvariations discussed during the study period. Validationof estimated values is desired from the available data.There is a direct relationship of pollution levels andemission strength in a given area. Hence, an attempthas been made to validate the emission inventory forall criteria pollutants by analyzing emissions in varioussampling zones with the ambient pollution levels. Forvalidation purpose, the geographical region encom-passing the study area (Delhi) has been divided intoseven emission zones as per the air quality monitoringstations using Voronoi polygon concept. Dispersionmodeling is also used for continuous elevated sourcesto have the contributing emissions at the ground levelto facilitate validation. A good correlation betweenemission estimates and concentration has been found.Correlation coefficient of 0.82, 0.77, 0.58 and 0.68 forCO, SO2, PM and NOx respectively shows a reason-ably satisfactory performance of the present estimates.

Keywords Emission inventory .Megacity Delhi .

Criteria pollutants . Voronoi polygon .

Top-down approach

1 Introduction

The air pollution in cities is due to multitude of differentsources. There are innumerable small sources such as

Environ Monit AssessDOI 10.1007/s10661-006-9400-9

M. Mohan (*) : L. DagarCentre for Atmospheric Sciences,Indian Institute of Technology,Delhi Hauz Khas,New Delhi 10016, Indiae-mail: mmohan6@hotmail.com

B. R. GurjarDepartment of Civil Engineering,Indian Institute of Technology,Roorkee, India

domestic fuel burning, traffic, small-scale industrialactivities, waste treatment etc. in addition to the largeindustries and power stations. In practice, it is seldompossible to examine all these sources in detail. Therefore itis common to treat major single sources individually, andto put all the small-scale emissions together in areasources. In order to handle the latter, the area may bedivided into quadratic grid an element of side length1–2 km and an emission value is estimated in each of thegrid element. This fine gridded emission data of airpollutants are of high importance for environmentalassessment studies and are essential input parameters forchemical transport models (Kuhlwein, Wickert,Trukenmuller, Theloke, & Friedrich, 2002). Delhi isone of the many megacities with alarming levels of airpollution from industries, power plants, residential, andtransportation sources. Therefore, gridded emissioninventory for the city is necessarily required in orderto address air quality management from atmosphericdispersion modeling based on deterministic concepts.Over the years, pollution abatement in Delhi hasbecome an important constituent of state policies.They are reflected in policy decision, legislativeframework and specific programmes most of whichbegun in 1990’s. This study deals with the preparationof gridded emission inventory from different sectorsfor criteria pollutants namely carbon monoxide (CO),sulfur dioxide (SO2), nitrogen dioxide (NO2) andsuspended particulate matter (SPM). The objective ofthe study is to generate emission isopleths and identifyemission hotspots for an annual time series of about adecade in support of air quality modeling as well as toperform validation of the same.

2 Materials and Methods

Emissions are estimated specifically for individual airpollutants (e.g., sulfur dioxide, carbon monoxide) orclasses of pollutants (volatile organic compounds,particulate matter), or for other substances affecting theatmosphere’s properties (specifically, the earth’s radia-

tion budget: e.g., carbon dioxide). An inventoryprovides information on the type of emission source,and on the spatial and temporal variability of therespective emission (Orthofer & Winiwarter, 1998). Inthe current study the emission of four criteria pollutantsi.e., CO, SO2, PM, NOx from different sectors areestimated throughout the city from a top-downapproach for emission inventory simulation. The top-down approach applies the concept to a larger area,where statistical data may be more readily available(CORINAIR, 2001). Emissions are first calculated forthe total area, and only subsequently distributed withinthe respective area using surrogate information anddownscaling methods. Monforti and Pederzoli (2005)described a formal implementation of this scheme. Anassessment of area wise as well as sector wise emissionof criteria pollutants is made. The major sources of airpollution is attributed to mainly five sectors i.e., powerplants, industry, transportation, domestic and waste.Emission of each pollutant is calculated from eachsector and then summed to obtain the total emission forthat year. Out of the five sectors, vehicular emissionscontribute almost three fourth of the total emission inDelhi (ESD, 1999–2000). Thermal power plants havebeen accounted in industrial sector. Domestic andwaste emissions have been combined in one sector asboth have a little contribution. The percentage contri-bution of industrial (including thermal power plants),vehicular and domestic sources of pollution to theambient air at the beginning of each decade in the pastis included in Table I.

The emission of pollutants is mainly from urbanareas hence our computational domain consists ofonly urban Delhi (Figure 1). Selected area shown bythe straight blue line in Figure 1 encompasses urbanDelhi (26×30 km2) that is 780 km2. For inventorypurpose the selected area is divided into 173 squaregrids of 2 km×2 km resolution. This area covers allmonitoring stations in Delhi.

The study is based on the emission data of 11 yearsi.e., 1990 to 2000 (Gurjar, van Aardenne, Lelieveld,& Mohan, 2004). For this time period, we have

Source 1970–1971 (%) 1980–1981 (%) 1990–1991 (%) 2000–2001 (%) t1.1

Industrial 56 40 29 20 t1.2Vehicular 23 42 64 72 t1.3Domestic 21 18 7 8 t1.4

Table I Percentagecontribution of pollutionfrom different sources

Environ Monit Assess

analyzed the policies that were implemented forpollution control and their effect on the pollutionfrom each sector. Due to the fact that there are somesectors that are being affected or whose contributioncan practically hamper or enhance the emission inother sector, an inter-sector wise emission distributionis also considered. In this study, each polluting sectoris subdivided into two subsections: (a) that give anintroduction about the sector and its emissions and(b) that gives the methodology used to quantify thesetotal emissions into each grid. The methodology forestimating gridded emissions for each of these sectorsis adopted from Mohan and Dube (1998).

2.1 Transportation or vehicular pollution

(a) In this section we have aimed at the assessmentof air quality due to transportation. Delhi’sregistered vehicular population has nearly doubledto 4.1 million from 1.9 million in 1991, register-

ing a growth rate of 9% per annum (Figure 2). Ofthe total cars in India, 17% run in Delhi. The risein air pollution load form transportation can alsobe seen from the rise in the consumption of bothmajor fuels i.e., diesel and petrol. Petrol con-sumption has increased four times from133,000 tonnes in 1980–1981 to 535,000 tonnesin 1990–2000 (EEA, 2000). Diesel (HSDO)consumption has increased three times from377,000 tonnes to 1,237,000 tonnes in this period(European Environment Agency (EEA), 2001).Approximately 500 vehicles exist for everykilometer of the road stretch in Delhi as comparedto 350 for Mumbai and Calcutta and 100 forChennai.

Generally, vehicles that run on petrol emit largequantities of unburnt hydrocarbons besides CO andsoot (the technical term for which is ‘suspendedparticulate matter’). However, vehicles using dieselemit more of soot and NOx.

Figure 1 Land use pattern of Delhi.

Environ Monit Assess

Figure 3 shows the estimated emission of differentcriteria pollutants by transport sector and the contri-bution of each pollutant in that respective year. Theemissions in year 1996 show a downfall as a result ofthe policies implemented by the Government. Al-though there has been a steady growth of 8% inemission of all the pollutants from 1990 to 1995 butthere has been a decline in the pollution level in 1996.A significant decrease in the emission of CO (10%)and SO2 (40%) is observed. However, PM and NOx

emissions still show a marginal increasing trend.During 1990–1991 for the first time mass emission

norms for vehicles at manufacturing stage as well asfor in-use vehicles have been notified. The emissionnorms along with fuel quality specifications laid downin 1996 required the automobile manufacturers tomake modifications in the engine design. The changesobserved in emission rates (Figure 3) are a reflection ofthe policies implemented on vehicular pollution.(b) We prepared an emission estimate from trans-

portation sector for the selected area shown in

Figure 1. The emission from Transport is dividedinto 173 grids according to fraction of road areain each sector. In this source inventory, threetypes of roads have been considered, namely:highways, residential and feeders. The emissionin each grid was calculated according to thefollowing formula:

ETi ¼Ai

ATotal� ETi ð1Þ

Ai ¼X3

n¼1

Ain ð2Þ

where, ETi and Ai are the emission and road area forthe ith grid due to transport, n represents the differenttypes of roads, ET and ATotal are the total emission androad area in urban Delhi. Delhi has large number offlyovers and area covered by these on each intersec-tion is also added to the road area.

Figure 2 Statistics aboutnumber of vehicles.

Figure 3 Annual emissions (Gg) from transportation sector.

Environ Monit Assess

Control measures in the transport sector, such ascatalytic converters and the introduction of EURO Inorms, do not seem to have curbed the NOx emissions,although they have probably prevented even strongerincreases. Emission of CO and SO2 has shown theaffect of control policies and as a result a decliningtrend is registered after 1996.

2.2 Industries

(a) Industries contribute from 20% to 40% of thetotal pollution load in Delhi during the studyperiod (Table I). The master plan of Delhi rulesout setting up large and heavy industries. Butthere is a tremendous increase in the number ofsmall scale industries which rose from 85,000 in1991 to 126,000 in 1996. According to the studythere are 28 approved industrial areas in Delhiwith a total of 21,627 industrial units (MCI,2001). The index of industrial production showsan increasing trend until 1995–1996 and hasdeclined from 1996–1997 onwards.Figure 4 shows a decrease in industrial

production from 1996–1998 but this decrease isnot reflected in the yearwise emissions. This isbecause newer industrial locations are alsocoming up as developed industrial areas. Ofthese industrial units, about one fifth units areeither closed or, their allocated plot is lyingvacant.

(b) The emission from all small scale, cottage, large-scale industries and power plants is calculatedand a grid wise emission inventory is prepared.Delhi has 24 main industrial areas so the emissionobtained was mainly by the regions such as

Anand Parbat Industrial Estate, Wazirabad etc.But apart from these industrial areas there may besome small scale industries operating in residen-tial areas. Although the emission from thesesmall scale industries has been calculated andtaken into account but this emission could not betaken for industrial sectors as this has beengenerated from industries operating in mixedland use domestic sector and considered here as10% of the total industrial emission. So we havedivided the total industrial emission into twofragments: EI-90 and EI-10. EI-90 is basically90% of industrial pollution distributed amongstall industrial grids and EI-10 is 10% of the totalindustrial emission apportioned in all residentialgrids, which are non-industrial.

2.3 Thermal power plants

Amongst the large scale industries thermal powerplants are the most prominent contributors to airpollution. Three coal based thermal power plantsnamely; Rajghat, Indraprastha (IP) and Badarpurpower plant are responsible for as much as approxi-mately 10% of the air pollution load in 2001. Thesethermal power plants are located in the heart of the city.Particulate emission from the IP power plant in 1998reached an all time high of 422 mg/Nm3 as against thenational standard of 150 mg/Nm3. Figure 5 shows thepower plant emissions for various pollutants.

Power Plant emissions on assigning the totalemissions in the grid where they are located resultinto extremely high emissions when validation isattempted with air quality data in each sampling zoneof the city. Also, power plants being a large

Figure 4 Year-wise index of industrial production.

Environ Monit Assess

continuous elevated point source with significantbuoyancy and momentum of the plume made itinfeasible to account for its total emission in a singlegrid where it is located as the gridded emissioninventory ought to have these emissions at the groundlevel in each grid. Thus to overcome this problem,emissions at the ground level were distributedassuming the pattern to be quite similar to themaximum ground level concentration distributionpattern. In order to obtain the emission distributionaround the plant at the ground level a Gaussian Modelnamed Industrial Source Complex Short Term(ISCST) model was utilized. Annual averaged groundlevel concentration distribution pattern has beenobtained for all the power plants in analysis area.This distribution pattern is used to evolve emissionpatterns by assigning weights associated to each zonerepresenting an air quality monitoring station. EZonerepresents the total power plant emission.

If aj are the weights in each of the seven samplingzones, then the emission in each grid from powerplants may be summarized as:

EPP ¼X3

i¼1

EPi ð3Þ

EPPj ¼ EPP � aj ; where j ¼ 1 to 7 ð4Þ

EPPj;i ¼ EPPj

Njð5Þ

where,P7

j¼1aj ¼ 1. EPP is the total emission from all

power plants and EPPj is the emission in the jthemission zone where aj is the dispersion ratioobtained from the dispersion model for that zone.

EPi is the emission of ith power plant, EPPj,i is thegrid averaged emission in the jth sector and Nj is thetotal number of grids in that emission zone.

The reduction in the emission from power plants isa clear indication of the policy change in the form ofreduction of sulphur content. All the three powerplants have installed high efficiency electrostaticprecipitators in all their units to control PM emission.Besides this the power plants are using benificiatedcoal (ash content less than 34%) since 1999 as againstthe coal used earlier (ash content 40%). In addition,low NOx burners have been installed at Pragati (gasbased) power plant.

2.4 Domestic and waste usage

(a) Domestic sector shares 8% of the pollution load ofDelhi. The domestic sector too has a share, albeitsmall as the percentage of households usingcleaner fuels such as LPG (liquefied petroleumgas) and kerosene is well above the all-Indiaaverage. Figure 6 shows the consumption of dif-ferent fuels for household.In 1999 the estimated quantity of garbage

generated in Delhi was 8,203 MT based on0.61 kg/capita per day against which only4,885 MT is properly disposed off (http://www.gisdevelopment.net/application/natural_hazards/overview/nho0014pf.htm).

Figure 5 Year-wise emissions (Gg) from power plants in Delhi.

Environ Monit Assess

The consumption of LPG in the domestic sectorwas almost double in 1998/1999 to that consumed in1990/1991 (Delhi Statistical Handbook (DSH), 2000).The kerosene consumption has dropped by 20% in thesame period, reflecting a typical fuel shift towardsLPG along with increasing urbanization. Such a shifthas a definite impact on the total emission load fromdomestic sources.

Several steps were taken to reduce the emission fromdomestic as well as waste sector. Government policiesfor waste management were taken up seriously regard-ing waste generation, recycling, waste incineration etcto control emission from this sector. Governmentmotivated the people to use LPG, which is a cleanerfuel as compared to dung, coal, charcoal, wood,kerosene, etc by subsidizing LPG (CSE, 1996). Figure 7

shows the emissions from domestic sector and wastecombined.(b) The total emission from domestic and waste has

been divided in all 173 grids according to ratioof population in each grid. Due to the fact thatthere are some small scales and cottage indus-tries that may lie in residential grids and mixedland use, an appropriate apportionment of theindustrial emission has been considered with thedomestic emission. Of the total industrial emis-sion, 10% has been added to the domesticemission as a result of which the total emissionthat we account for the domestic sector orresidential non-industrial grids become:

TEDW ¼ ED þ EW þ EI10 ð6Þ

Figure 6 Households using different fuels for cooking (%) 1991.

Figure 7 Emission from domestic sector and waste production.

Environ Monit Assess

The emission of each grid from domestic andwaste can be given as:

EDi ¼ Pi

P� TEDW ð7Þ

where Pi and P are the population in ith grid and totalpopulation. EDi denotes the emission in ith grid dueto emission from domestic sector and waste, ED andEW denotes the total emission from domestic andwaste sectors respectively. EI10 denotes 10% of the

Figure 8 Air qualitymonitoring stations inDelhi.

Figure 9 Distribution ofanalysis area using Voronoipolygon technique.

Environ Monit Assess

industrial emission, which has been discussed indetails in Section 2.

CO emissions from domestic fuel use show a strongdecrease from 1997 to 2000 caused by a change fromtraditional fuels to cooking gas. As a result of moreefficient burning, NOx emissions from cooking gas arehigher than under conditions of burning traditional fuels.Thus the shift of traditional fuels towards cooking gasresults in an additional increase of NOx emissions.

3 Air Quality Monitoring Networkand Emission Zoning

Air Quality Monitoring is an important part of the airquality management. Central Pollution Control Boardis the governing body responsible for monitoring ofdifferent pollutants in Delhi. The CPCB’s NationalAmbient Air Quality Monitoring Programme(NAAQMP) has been established with objectives todetermine the present air quality status and trends andto control and regulate pollution from industries andother sources to meet the air quality standards. InDelhi, CPCB has seven monitoring stations namely,Ashok Vihar, Shahadra, Nizzamudin, Siri Fort, Shaza-dabagh, Janakpuri and ITO (TERI, 1997). AnnualConcentration data for SO2, PM and NOx has beentaken for the entire period at each observational site.However, data for CO was not available for all sitesand all time except the busy traffic intersection at ITO.

The total analysis area is divided into sevenemission zones corresponding to seven CPCB obser-vational sites and divided into seven zones also namedas emission zones. Figure 8 shows all the sevenCPCB’s observational sites located inside the analysisarea. These seven zones are constructed usingVoronoi Polygon technique that is a standard mathe-matical application (Okabe, Boots, & Sugihara, 1992;Du & Gunzburger, 2002). Voronoi diagrams representthe region of influence around each of a given set ofsites. In this technique, we decompose the analysisarea into zones around each point such that all thegrid points of that zone are closer to that observationalsite than they are to any other observational point inthe study area. Every zone constructed contains oneobservational site and a set of emission grids, whichare closest to that observational site. This was done inorder to associate the emissions of each grid to thenearest observational site. The Voronoi distribution of

the analysis area is given below in Figure 9. Theboundaries of the resulting seven zones are furtherapproximated using Stair step grid structure.

The emissions for each zone were hence calculated.Since there is a remarkable difference in the number ofgrids in each zone, so to normalize this effect; averageemission per grid for a zone was used as compared tothe total emission.

4 Results and Discussion

4.1 Trends of emission isoplethsand emission hotspots

Emission isopleths were plotted for the emission data ofthe year 1990, 1996 and 2000, for each of the pollutantsindividually. Figures 10, 11, 12 and 13 shows theemission isopleths from gridded inventory for SO2,CO, NOx and PM respectively from total emissions oncombining contributions from all sectors. Emissionhotspots or the worst affected areas are shown for theyears 1990, 1996 and 2000 in these figures. All theemissions shown are in Giga grams (Gg).

The hotspots of emission are identified as: (a)primary and (b) secondary which are represented inFigures 10, 11, 12, & 13 by star (✩) and ring (⊚)respectively. A pollutant wise list of primary andsecondary hotspots is given in Table II.

Primary hotspots are the places with maximumemission load and secondary hotspots are the placesthat have a prominent contribution to emissions nextto primary ones. Common primary hotspots for COand NOx are identified as ITO (Income Tax Office),AIIMS (All India Institute of Medical Sciences),Mangolpuri and Shahadra whereas CP (CannaughtPlace) for CO and Nariana for NOx are differentwhich are accounted due to the heavy traffic load atCP and clusters of industries and domestic emissionsat Nariana. Thus, similar emission hotspots for COand NOx can be explained due to their same emissionsource. Similarly, for PM and SO2, common primaryhotspots are identified as ITO, CP and India Gate dueto the fact that their source of generation is mainlyfrom Power Plants.

Secondary emission hotspots are the places withsignificant emission load and wider area but which areless in magnitude than the primary ones. These placesare also common for CO and NOx and are identified as

Environ Monit Assess

Karolbagh, Model Town, Badarpur, Okhla, VasantVihar and Gurgaon Road. Patparganj, Uttam Nagarare the secondary hotspots for NOx due to the influenceof power plant and industries in these areas. For PMand SO2 the common areas are Karolbagh, Shahadra,AIIMS, Mangolpuri whereas Badarpur for PM isreplaced by Model Town for SO2 which may be due

to domestic sources, generators etc. and industries inneighbouring areas.

The emission trends show an increase from the year1990–2000 for CO, PM and NOx whereas there is adecrease of SO2 emission from 1990–1996 but afterthat it shows a slight increment till 2000. The increasein emission for NOx was only magnitude wise from

0 5 10 15 20 (km)0

5

10

15

20

25

(km

)

0

1

2

3

4

5

6

7

8

CO-20009

VASANT VIHAR

GURGAON RD

(c)

0 5 10 15 200

5

10

15

20

25

(km

)

0

1

2

3

4

5

6

7

8

(km)

CO-1990

9

ITO

CP

AIIMS

MANGOLPURI

SHAHADRA

MATHURA RD

ROHTAK RD.

(a)

0 5 10 15 200

5

10

15

20

25

(km

)

0

1

2

3

4

5

6

7

8

(km)

9

CO-1996

KAROL BAGH

MODEL TOWN

OKHLA

BADARPUR

(b)

Figure 10 Emission isopleths for CO for 1990, 1996 and 2000.

Environ Monit Assess

1990–2000 and the spatial distribution of emissionremained same. However, for other pollutants the trendsfrom 1990–2000 showed an increase in both magnitudeand area. Though primary hotspots remained same from1990–2000, there was addition of new areas assecondary hotspots in the subsequent years since 1990as can easily be seen in Figures 10, 11, 12, & 13.

The main polluting areas or primary hotspotsshows combined emissions for all pollutants in areas

near ITO, Rajghat, Badarpur and CP. Secondaryhotspots are identified near Shahadra and South Delhiand Model Town with combined emissions from allpollutants. However areas of South and North Delhiare found to be relatively less polluted as compared toITO etc. Although there has been a noticeable changein the pollution level of SO2 and CO but still a lotneeds to be done for NOx and PM, which remainsabove the prescribed standards. Moreover, decrease in

0 5 10 15 200

5

10

15

20

25

(km

)

00.20.40.60.811.21.41.61.822.22.42.62.83.0

PM-1990

(km)

ITO

IG

CP

(a)

0 5 10 15 200

5

10

15

20

25

(km

)

00.20.40.60.811.21.41.61.822.22.42.62.8

PM-1996

(km)

3.0

MANGOLPURI

KAROL BAGH

SHAHADRA

AIIMS

BADARPUR

(b)

0 5 10 15 200

5

10

15

20

25

(km

)

00.20.40.60.811.21.41.61.822.22.42.62.8

PM-2000

(km)

3.0

(c)Figure 11 Emission isopleths for PM for 1990, 1996 and 2000.

Environ Monit Assess

some of the pollutants has been witnessed but theprototype/blueprint of the polluting areas remains thesame i.e., the areas of high pollution still remainsintact and follows the same pattern.

The emission trends show that emissions inmegacity Delhi are partly associated with a stronginter-annual variability based on introduction ofcontrol measures, fuel shifts, changes in agricultural

practice, increase in the vehicle fleet or problemswithin the power generation sector. Further, theemission trends and air quality analysis reveal thatthe impact of several control measures is not verylarge except for SO2. For example, NOx emissions hadbeen steadily increasing throughout the study period,and TSP levels in Delhi’s air still violate thepermissible limits.

0 5 10 15 200

5

10

15

20

25

(km

)

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

SO2-1990

(km)

2.2

ITO

CP

IG

BADARPUR

(a)

0 5 10 15 200

5

10

15

20

25

(km

)

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

SO2-1996

(km)

2.2

(b)

0 5 10 15 200

5

10

15

20

25

(km

)

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

(km)

SO2-20002.2

MODEL TOWN

MANGOLPURI

KAROL BAGH

SHAHADRA

AIIMS

(c)Figure 12 Emission isopleths for SO2 for 1990, 1996 and 2000.

Environ Monit Assess

4.2 Validation of emission strength

The most important and crucial part of an emissioninventory is its validation with the observed data. It ispractically impossible to prepare an accurate emissionestimate for each of the sources concerned since thenumber of sources incorporated is randomly changing

both temporally and spatially and therefore it isequally difficult to perform validation studies for thesame.

The only way to assess how realistic the results areis a comparison to independently derived data. Sturmet al. (2000) have compiled possibilities to performsuch validations. Such a comparison exercise has

0 5 10 15 200

5

10

15

20

25

(km

)

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

2.2

2.4

2.6

(km)

2.8

NOx-1996

0 5 10 15 200

5

10

15

20

25

(km

)

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

2.2

2.4

2.6

(km)

NOx-20002.8

0 5 10 15 200

5

10

15

20

25

(km

)

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

2.2

2.4

2.6

NOx-19902.8

(km)

ITOSHAHADRA

NARAINA

MANGOLPURI

AIIMS

MODEL TOWN

KAROL BAGH

PATPARGANJ

UTTAM NAGAR

BADARPUR

(a) (b)

(c)Figure 13 Emission isopleths for NOx for 1990, 1996 and 2000.

Environ Monit Assess

been performed, using results from a top–downAustrian emission inventory compendium (Ritter,Ohr, & Gugele, 1999) that had been disaggregatedto the level of provinces. Xia and Shao (2005) haveattempted to validate estimated traffic emissions at threeintersections using air quality monitoring data of threesampling stations in Hong Kong Island. Hourly emis-sion estimates are compared with hourly concentrationsfor 24 hours and equal number of points is used toobtain linear fit and correlations in their study. In thepresent study, annual average emission per grid of a

zone is plotted with concentration at seven different sitesrepresentative of that zone. The number of grids in eachzone is varying, so an average emission of each grid wasused as compared to the total emission of a zone sinceconcentration at a site is a quantitative representation ofpollutants at that particular point/grid.

Central Pollution Control Board (CPCB) concen-tration data for each pollutant has been used tovalidate the emissions obtained for these pollutants.For validation purpose, an average emission for eachof the sampling zone is prepared using the Voronoi

Table II Primary and secondary emission hotspots in Delhi

Pollutant CO PM NOx SO2

Primaryhotspots

Mangolpuri, ITO, CP,Shahadra, AIIMS

ITO, CP,India Gate

Mangolpuri, ITO, Shahadra,AIIMS Naraina

ITO, India Gate,Badarpur, CP

Range (>4 Gg) (>2.2 Gg) (>2 Gg) (>1.5 Gg)Secondaryhotspots

Karolbagh, Model Town,Badarpur, Okhla, VasantVihar, Gurgaon Road

Karolbagh, Shahadra,AIIMS, Badarpur,Mangolpuri

Karolbagh, Model Town, OkhlaBadarpur, Vasant Vihar, Gurgaon RoadPatparganj, Uttam Nagar

Karolbagh, Shahadra,AIIMS, Mangolpuri,Model Town

Range (2.5–4 Gg) (1–2.2 Gg) (1–2 Gg) (0.8–1.5 Gg)

(a)

SO2y = 12.85x + 9.5677

R2 = 0.4704

0

10

20

30

40

50

0 0.5 1 1.5 2

Average Emission

NOxy = 26.819x + 11.534

R2 = 0.5987

10

20

30

40

50

60

70

80

0 0.5 1 1.5 2

Average EmissionEmission (Gg)

Co

nce

ntr

atio

n (

µg

m-3)

Co

nce

ntr

atio

n (

µg

m-3)

Co

nce

ntr

atio

n (

µg

m-3)

Co

nce

ntr

atio

n (

µg

m-3)

Emission (Gg)

(b)

(c)

COy = 543.98x + 1996.2

R2 = 0.6766

200025003000350040004500500055006000

0 1 2 3 4 5 6

Average Emission

PMy = 61.742x + 317.17

R2 = 0.3385

200

250

300

350

400

450

500

550

0 0.5 1 1.5 2 2.5

Average EmissionEmission (Gg) Emission (Gg)

(d)Figure 14 Emission versus concentration plot for (a) SO2, (b) NOx, (c) CO, (d) PM.

Environ Monit Assess

Figure 15 Comparison of concentration and emission trends for all years (1990–2000) and for all seven stations for (a) NOx, (b) SO2,(c) CO, (d) PM.

Environ Monit Assess

diagram discussed above in Section 3 that identifiesnumber of grids to be considered for each of thesampling zone. Using Voronoi distributed zones inthe city, the estimated emissions are compared withthe corresponding ambient air pollution concentra-tions at all seven monitoring stations for all 11 years.There are seven stations and 11 years so the totalnumber of points should be 77 for each pollutantspecies but due to lack of observational data for ITO(from 1990–1995), which is available only from1996–2000, we have a total of 72 points for NOx,SO2 and PM. The monitoring data for CO was notavailable for all the sites and all years so we havevalidated the emission with the observational data attwo sites only namely ITO and Siri Fort. Thuspredicted emission of each pollutant against theobserved concentration of CO, NOx, SO2, and PM isshown in Figure 14 (a–d). The linear regressionequations and the correlation coefficients R2 are alsoshown in these figures. The relationship obtained forCO, SO2, PM and NOx are 0.82, 0.77, 0.58 and 0.68respectively. Here the unit of concentration is micro-grams/m3 and emission is Giga grams (109 gm).

The trend of concentration of each pollutant andcorresponding emission for all sampling points andyears is shown in Figure 15 (a–d). The trends shownby concentration plot with that of emission in thesefigures shows a good relationship, which furtherjustifies the good correlations obtained above. Thepeaks at all the stations and for all the years have beenwell captured though the magnitude may be differentfor each case. The lower correlation of PM isattributed to the magnitude difference in the earlyyears but in later years this is well captured.

Based on linear regression and trend analysis it canbe inferred that a satisfactory validation of the esti-mated emission inventory is performed. Nevertheless,this can be improved with more data when available.

5 Conclusions

1. Gridded emission inventory has been pre-pared for all criteria pollutants for megacityDelhi annually for the years 1990–2000 insupport of air quality modeling.

2. Emission isopleths are prepared and primary andsecondary emission hotspots are identified forthe entire decade and for all criteria pollutants.

3. A new approach is adopted for emissionzoning using Voronoi polygon technique forvalidation purpose.

4. For apportioning of power plant emissionswhich are needed to obtain total all sectoremissions near ground level is performedwith due weightings in all the grids from adispersion modeling approach.

5. A good correlation for the estimated emissioninventory is obtained that shows a reasonablevalidation.

6. Different approaches can be used for valida-tion with the availability of more data.

7. Although there has been a noticeable reduc-tion in the pollution level of SO2 and CO butstill a lot needs to be done for NOx and PMthose remains above the prescribed standards.

8. Moreover, decrease in some of the pollutantshas been witnessed but the prototype/blue-print of the polluting areas remains the samei.e., the areas of high pollution still remainsintact and follows the same pattern.

References

CORINAIR Atmospheric Emission Inventory Guidebook(2001). 2001_3/en (EEA technical report n. 30) Third ed.Copenhagen: European Environment Agency. Retrievedfrom: http://reports.eea.eu.int/technical_report.

CSE (Centre for Science and Environment) (1996). Slowmurder. A CSE publication. Centre for Science andEnvironment, 41, Tughlakabad Institutional Area, NewDelhi, India. E-mail: cse@cseindia.org.

DSH (Delhi Statistical Handbook) (2000). Directorate of eco-nomics and statistics. New Delhi: Govt. of NCT of Delhi.

Du, Q., & Gunzburger, M. (2002). Grid generation andoptimization based on centroidal Voronoi tessellations.Applied Mathematics and Computation, 133, 591–607.

EEA (European Environment Agency) (2000). COPERT III.Computer program to calculate emissions from roadtransport. Methodology and emission factors (version 2.1).

EEA (European Environment Agency) (2001). EEA TechnicalReport No. 49. Joint EMEP/ESD (Economic Survey ofDelhi) (2001–2002). Planning Department Government ofNational Capital Territory of Delhi, 6th Level, B-Wing, DelhiSecretariat, I. P. Estate. New Delhi (India). URL: http://delhiplanning.nic.in/Economic%20Survey/Ecosur2001–02/Ecosur2001–02.htm.

ESD (Economic Survey of Delhi) (1999–2000). Q3PlanningDepartment Government of National Capital Territory ofDelhi, 6th Level, B-Wing, Delhi Secretariat, I. P. Estatehttp://www2.toki.or.id/book/AlgDesignManual/BOOK/BOOK4/NODE187.HTM.

Environ Monit Assess

Gurjar, B. R., van Aardenne, J. A., Lelieveld, J., & Mohan, M.(2004). Emission estimates and trends (1990–2000) formegacity Delhi and implications. Atmospheric Environment,38, 5663–5681.

Kuhlwein, J., Wickert, B., Trukenmuller, A., Theloke, J., &Friedrich, R. (2002). Emission-modelling in high spatialand temporal resolution and calculation of pollutantconcentrations for comparisons with measured concen-trations. Atmospheric Environment, 36, S7–S18.

MCI (Ministry of Commerce and Industry) (2001). Handbookof industrial policy and statistics 2001. Office of theEconomic advisor, Ministry of Commerce and Industry,Govt. of India, New Delhi. URL: http://eaindustry.nic.in/handout.htm.

Ministry of Transport website: http://transport.delhigovt.nic.in/default.html.

Mohan, M., & Dube, M. (1998). Ambient air qualityassessment of sulphur dioxide over Delhi. Indian Journalof Environmental Protection, 720–728. August.

Monforti, F., & Pederzoli, A. (2005). THOSCANE: A tool todetail CORINAIR emission inventories. EnvironmentalModelling & Software, 20(5), 505–508. New Delhi.URL: http://delhiplanning.nic.in/Economic%20Survey/Economic%20Survey%201999–2000.htm

Okabe, A., Boots, B., & Sugihara, K. (1992). Spatial tesselationsconcepts and applications of Voronoi diagrams. New York:Wiley.

Orthofer, R., & Winiwarter, W. (1998). Spatial and temporaldisaggregation of emission inventories. In H. Power &J. M. Baldasano (Eds.), Air pollution emissions inventory.In J. W. S. Longhurst & H. Power (Eds.), Advances in AirPollution (pp. 51–70), vol. 3. Southampton: Computation-al Mechanics Publications.

Ritter, M., Ohr, B., & Gugele, B. (1999). Luftschadstoff-trends inOsterreich 1980–1998, BE-165. Vienna: Umweltbundesamt.

Sturm, P. J., Blank, P., Bohler, T., Lopes, M., Mensink, C., Volta,M., et al. (2000). In P. M. Borrell & P. Borrell (Eds.),Proceedings of the EUROTRAC Symposium ’98 vol. 2(pp. 708–709). Southampton: WIT Press.

TERI (The Energy and Resources Institute) (1997). Environ-mental aspects of energy use in large Indian metropolises(p. 207). New Delhi: Tata Energy Research Institute.[TERI Project Report No. 94EM53]. Summary andConclusions. URL: http://www.teriin.org/reports/rep06/rep0602.htm.

Xia, L., & Shao, Y. (2005). Modelling of traffic flow and airpollution emission with application to Hong Kong Island.Environmental Modelling & Software, 20, 1175–1188.

Environ Monit Assess

All in-text references underlined in blue are linked to publications on ResearchGate, letting you access and read them immediately.