THE WORLD BANK South Asia Urban Air Quality Management Briefing Note No. 14 What is Causing Particulate Air Pollution? Evidence from Delhi, Kolkata, and Mumbai Very little information is available on sources of fine particulate air pollution in South Asia. This study represents one of the first detailed fine particulate matter source apportionment studies carried out in the region. The results indicate that there is no single dominant source, but sources differ by location and season among the three Indian cities examined. This finding would suggest that vigorously pursuing control measures in one sector, while leaving other sectors largely untouched, is less likely to result in a marked improvement in urban air quality than if a multi-pronged approach addressing a number of sources and sectors is adopted. I dentifying what air pollution sources are major countries. However, detailed compositions of local sources contributors to elevated ambient concentrations of are expensive to generate and this study relied on data critical pollutants is the first step toward designing an from countries outside of South Asia. Markers are effective policy package for air quality management. In sufficiently specific to different sources and are not South Asia, this question is especially relevant for sources expected to vary from region to region. However, the of airborne PM (particles smaller than 2.5 microns, also precise compositions of sources have been shown to vary 2.5 called fine particulate matter) which is a major health with the mode of operation (for example, speed and concern. The Georgia Institute of Technology, United frequency of stopping and starting in the case of vehicle States, in collaboration with the National Physical operation), location (especially for road dust), weather Laboratory, the Indian Institute of Technology (Mumbai), conditions, and other parameters. Therefore, chemical and the National Environmental Engineering Research mass balance receptor modeling, as with other source Institute carried out an analysis of ambient PM in Delhi, apportionment modeling, should be regarded as a tool that 2.5 Kolkata, and Mumbai using a technique called chemical provides a semi-quantitative understanding of the mass balance receptor modeling (See [1] for more details importance of different sources. More detailed information and references). on sources of uncertainties are given in [1]. Chemical Mass Balance Receptor Sampling Sites Modeling One urban residential site was selected in each city. Care Receptor modeling has been widely used as a technique was taken to avoid undue influence from heavy city-traffic inairpollutionsourceapportionmentstudies[2].Chemical or industrial emissions. mass balance receptor models analyze the chemical Mumbai, the largest city in India, is located on the Arabian compositions of PM samples and compare them to 2.5 Sea. Because ocean air is typically cleaner than continental "source profiles," the chemical compositions of emissions air, proximity to the ocean and the influence of diurnal from different sources such as vehicles, road dust land and sea breezes aid the dilution of PM resuspension, and wood burning. The comparison allows 2.5 concentration. In contrast, Delhi is located inland. Analysis estimation of the contributions of different sources to the of wind trajectories from 1995 to 1999 shows that collected ambient PM samples. 2.5 62 percent of all trajectories arriving in Delhi during that This study used compounds found in the organic carbon period experienced stagnation [1]. Air stagnation keeps fraction of PM as molecular markers for several key particles suspended over the city for an extended period 2.5 sources. Molecular markers can be highly specific for of time and worsens air quality. In addition, lower different sources and there exists a good understanding temperatures in the winter months lead to atmospheric of organic tracers for a wide range of sources in industrial inversion which traps pollutants close to the ground, and 1 this, combined with low rainfall, increases PM and road dust tend to have a low ratio. It is important 2.5 concentrations further. to emphasize that the definition of EC and OC is procedural. There are at least 15 internationally The weather conditions differ markedly among the accepted procedures for EC and OC determination, three cities [3]. Delhi has a moderate monsoon season and the ratio of EC to OC differs for any between July and September and has the least rainfall given sample depending on the procedure, although of the three cities. Kolkata has a longer monsoon all of them should give the same total carbon content. season, lasting from May to October, and much more rainfall. Mumbai has a severe monsoon season with The molecular markers used in this study included: the heaviest rainfall of the three. These differences in the weather and geographical conditions partially M Hopanes and steranes, present in lubricating oil and consequently in the exhaust emissions of gasoline and explain much higher PM concentrations in Delhi 2.5 diesel-powered motor vehicles than in other cities. M Levoglucosan, a major component of particulate matter Sampling of PM2.5 from wood combustion PM samples were collected over consecutive 24- 2.5 M Picene, a marker for coal combustion hour periods between March 2001 and January 2002. The total number of days yielding useful results was M Silicon and aluminum, markers for road dust and the 21 in Delhi, 20 in Kolkata, and 25 in Mumbai. The only markers that were not based on OC. days on which PM samples were collected are 2.5 shown in Table 1. Twenty source profiles were tested in the chemical mass balance model. The modeling did not find Table 1 Schedule of Dates for Sample significant levels of combustion products of fuel oil. Collection This is consistent with very low consumption of fuel oil compared to diesel and gasoline in Delhi. After Season Month Delhi Kolkata Mumbai extensive analysis, five source profiles were retained: Spring Mar-01 4,10,16, 16,22,28 4,10,16, gasoline, diesel, road dust, coal, and biomass. Of the 22,28 19,22,28 five source profiles retained, regional source profiles-- Apr-01 3,9 Bangladesh in this case--were available only for biomass: coconut leaves, rice straw, cow dung, biomass Summer Jun-01 8,14,20, 8,14,20, 8,14,20, briquette, and jackfruit branches. The source profiles 26 26 26 used for gasoline, diesel, and road dust were from the Jul-01 2,8,14 2 2 United States, and that for coal was from Beijing. The Autumn Oct-01 5,11,17 11,17,23, 5,11,17, absence of local source profiles is one source of 29 23,29 modeling uncertainties. Nov-01 4 4,10 The profiles for gasoline, diesel, and coal indicate Winter Dec-01 17,23,29 5,11,17, 5,11,17, the fuel used but not how or in which sector the fuel 23,29 23,29 is combusted. In the case of gasoline, virtually all gasoline can be safely attributed to vehicles. But it Jan-02 4,10,16 4,10 is not possible to distinguish between diesel burned Chemical Analysis in vehicles and diesel burned in stationary sources (such as small diesel power generators frequently Detailed analysis of particles typically involves used by shops and small industrial establishments in chemical analysis of sulfates (SO ), nitrates (NO ), 2- - 4 3 India). That said, stationary sources are known to ammonium (NH ), and other water-soluble inorganic + 4 emit much less particulate matter per unit of fuel compounds; determination of elemental carbon (EC) burned than vehicle engines so that a significant and organic carbon (OC) as well as total carbon (the fraction of what is identified as diesel here is probably sum of EC and OC) by weight; and chemical analysis from diesel vehicle exhaust. In this study diesel of the organic compounds. actually includes kerosene used in conjunction with Carbon in particulate matter comes from combustion lubricating oil, most notably kerosene added to processes and it is relatively straightforward to automotive diesel, but not kerosene used in cooking. determine the total amount. Differentiation between Biomass and coal burned by households are EC and OC is more complex. Fine particles found in indistinguishable from those burned in bakeries and diesel engine exhaust and fuel oil and coal combustion cottage industries. Some portions of PM classified 2.5 products tend to have a high EC-to-OC ratio, while as road dust may be fugitive emissions from industry. emissions from gasoline cars not equipped with It is also not possible to trace secondary sulfates, catalytic converters, biomass combustion products, nitrates, and ammonium to different sources. 2 Results exceed 100 percent using this source apportionment methodology. These observations nevertheless A summary of the total concentrations of PM , EC, 2.5 and OC in micrograms per cubic meter (µg/m³), and suggest that the results for these samples could contain larger uncertainties than other samples and carbon (EC+OC) as a percentage by weight (wt%) should be interpreted with greater caution. For the of PM is given in Table 2. The high carbon contents 2.5 rest of this note, Kolkata spring, summer, and autumn measured indicate the importance of fossil-fuel samples are excluded from further consideration for and biomass contributions to fine particulate these reasons. air pollution. The results of chemical mass balance receptor Table 2 Seasonal Average Concentrations modeling for the remaining eight samples are shown of PM and Carbon in Figure 1. The sources shown in the figure include 2.5 fuels (gasoline, diesel, coal, and biomass), "road dust," Season Component Units Delhi Kolkata Mumbai particulate matter formed through atmospheric reactions (secondary nitrates and sulfates, which are Spring PM µg/m³ 114 55 36 formed from emissions from various combustion 2.5 EC µg/m³ 9.1 6.1 3.7 sources, and secondary ammonium, which can be OC µg/m³ 38 19 9.5 from agricultural sources in addition to combustion Total carbon wt% 41 44 37 sources), and unidentified sources (such as water and unidentified organic compounds). "Unidentified" is Summer PM µg/m³ 49 26 21 2.5 the difference between the sum of components EC µg/m³ 4 6.6 1.1 accounted for in source apportionment and the OC µg/m³ 16 7.8 1.6 measured PM level. 2.5 Total carbon wt% 40 55 13 "Road dust" (which may include fugitive industrial Autumn PM µg/m³ 159 45 64 emissions) was the largest contributor in three 2.5 samples, biomass combustion in one, and unidentified EC µg/m³ 11 9.1 5.6 sources in the remaining four. Despite large numbers OC µg/m³ 57 18 20 of two-stroke engine gasoline vehicles, known for Total carbon wt% 44 62 38 their high particulate emissions, diesel contribution exceeds that of gasoline in all cases. This is plausible Winter PM µg/m³ 231 305 89 2.5 given the high consumption of diesel compared to EC µg/m³ 17 27 8.2 gasoline in India. This finding would suggest OC µg/m³ 96 147 34 that focusing on diesel vehicles should be given Total carbon wt% 46 57 48 priority in air quality management. However, it is difficult to separate gasoline and diesel contributions accurately, so that the aggregate contribution of Samples in each city were combined by season for gasoline and diesel combustion is likely to be a organic marker analysis, giving a total of 12 combined much more accurate value than the contribution of samples for source apportionment determination. Of each fuel. the 12 samples, one (summer in Mumbai) could not be used because it gave OC that was below the detection limit for Figure 1 Receptor Modeling of PM2.5 identifying organic markers, leaving in Delhi, Kolkata, and Mumbai a total of 11. 350 Of the 11 remaining samples, 300 Kolkata in spring, summer, and meter Unidentified 250 Secondary ammonium autumn showed trends that were cubic Secondary nitrates inconsistent with all other samples. 200 Secondary sulfates per In addition, the sums of contributions Biomass 150 Coal of different components (combustion 100 Road dust products of gasoline, diesel, and so Gasoline Micrograms 50 Diesel on) exceeded the measured PM 2.5 mass in these samples, adding up to 0 107 (spring), 130 (summer), and 120 spring winter winter spring winter (autumn) percent of the measured summer autumn autumn mass. It is possible, and in fact Delhi Delhi Delhi Delhi Kolkata Mumbai Mumbai not uncommon, for the sum to Mumbai 3 If all of diesel is attributed to mobile sources, vehicle Understandably, much policy attention has concentrated exhaust becomes the largest contributor in one season: on vehicle exhaust to date in the region. However, this Mumbai in spring at 28 percent. The combined source apportionment study highlights the importance contribution of coal and biomass combustion was of addressing several sources of air pollution in parallel. greater than that of gasoline and diesel combined in all In particular, solid fuel use in industry and household seasons except summer in Delhi and spring and winter cooking as well as for heating in winter can become a in Mumbai. significant source of to airborne fine particulate matter. This is especially true in cities with cold winters that Secondaryparticulateformationcomprisedapproximately require heating--mainly in northern India, Nepal, and one-tenth to one-fifth of PM . Secondary sulfates and 2.5 Pakistan--precisely in the season when ambient nitrates arise mostly from combustion processes, so that concentrations from all sources are elevated on the actual percentage contributions of biomass, coal, and account of thermal inversion. These and other fossil fuel combustion are higher than those indicated sources would need to be tackled for air quality above. In particular, high-sulfur fuels such as fuel oil used improvement. in industry may be contributing disproportionately to secondary sulfates. A number of cities in India are currently developing action plans to improve air quality. The results of this Conclusions and Policy Implications study underscore the importance of basing, to the extent possible, strategies on city-specific data on the The results show that there is no single dominant source mix of emission sources and meteorological but rather a number of sources contribute to PM . 2.5 parameters. Broadly, the contributions of different sources vary with season and across the three cities. For example, mobile References sources and biomass combustion appear to contribute substantially and in several cases approximately in equal 1. ESMAP. 2004. Toward Cleaner Urban Air in South Asia: proportions (spring and autumn in Delhi and autumn in Tackling Transport Pollution, Understanding Sources. Mumbai). The contribution of "road dust" can also be ESMAP Report 281/04, March. Washington D.C. Available online at . Mumbai). Predictably the combined contribution of biomass and coal is the highest in winter in Delhi and 2. Chow, J.C. and J.G. Watson. 2002. "Review of PM Source 2.5 Kolkata, presumably as a result of heating. Contributions Apportionment for Fossil Fuel Combustion and Other from solid fuel combustion are also significant in non- Sources by the Chemical Mass Balance Receptor Model." heating seasons: spring and autumn in Delhi and autumn Energy & Fuels16(2): 222-260. in Mumbai, probably on account of considerable use of 3. World Bank. Forthcoming. "For a Breath of Fresh Air: Ten solid fuels in small-scale industries and by households Years of Progress and Challenges in Urban Air Quality for cooking. Management in India." This briefing note was prepared in August 2004 as part of the South Asia program on urban air quality management, funded in part by the joint UNDP/World Bank Energy Sector Management Assistance Programme (ESMAP). The objective of the program is to support the region-wide process of developing and adopting cost-effective and viable policies and efficient enforcement mechanisms to reverse the deteriorating trend in urban air. A full set of briefs and other materials are available at . For further information, contact Sameer Akbar (sakbar@worldbank.org) or Masami Kojima (mkojima@worldbank.org). Designed & Printed by: Macro Graphics Pvt. Ltd. 4