H1 24064; ESMA^^P THE WORLD BANK 11 . What Do We Know About Air PolAtionn'-India Case Study Govermltents il Souith Asia ateo urged to ad,iiess Ua ban-atlir po)(llultion as a mat( eof'higli pr iorf in iits most affected cities. In oirder- to devise effective intervention mleasures. howevej; we need to unzderstanid which sources are responsible fir thie high exposure of'the genera1l public to air pollution. Tota Eniergy Research In stitute (Teri) in Deliii recently reviewetl the informcation available sintce 1990 in Inidia to answer this questioni [1]. Their rep)ort showvs Ihat gap)s ini datal atnd analysis are 'ili/i 1', lm l /arge lon mkue answering this impoJrtanlt question dif ficult. This br-iefing note sunmnarizes kev findings in1 thzat report. articulate matter is the most serious pollutant in The available data. in India show that pollutant large cities in South Asia. There are mainy sources concentrationis are typically withini the national ambient of particulate pollution: large industrial plants, air quality standards [2] with the exception of particles. medium- and small-scale industries, refuse burning, A recent case study [3] monitored RSPM twice a week households burning biomass for cooking and heating, at 10 stations in Delhi over a 13-month period between vehicular exhaust, re-suspended road dust, construction. July 2000 and July 20I1I. The mean RSPM concentrations particles migrating from other regions, and naturally averaged 204 pg/r1n3, coinsiderably above the US atnual occurring dust. These sources emit particles of varying PM1l standard of _5( pg/M3. The correlations between sizes-small particles affect public health much more than NO, and RSPM concentrations were extremely weak, large particles. It is important to have a good suggesting that sources other than road traffic were understanding of the level of exposure of the general public contributing significantly to ambient RSPM. to particulate air pollution, and of the relative contributions There is essentially no information on background of these different sources (referred to as source apportionmiffer ent).purces (referred to a r tic u particulate concentrations. This information would be important for devising effective mitigation measures and setting realistic targets, because it matters a great deal whether high ambient concentrations are a result of human Ambient air quality has been monitored in India since activities in the city, or as a result of naturally occurring 1967. There were 204 monitoring stations in operation in particles, or even particles migrating from other regions. 2001. Sullur dioxide (SO,). nitrogen dioxide (NO,) and If background concentrations are high, imposing strict total suspended particles (TSP) have been historically controls on human activities in an attempt to reduce air monitored on a regular basis. Respiratory suspended polluion may yield much smaller benefits than anticipated. particulate matter (RSPM), with the particle diameter Discussions with practitioners in air quality monitoring cut-off somewhere near 10 miicrons (pmii), hlas been added recently at a numniber of monitoring stations, totaling suggest that quality assurance anid quality control needs t, ' 5 C strengthening. The areas that call for attention include 77 by the end of 2000. TSP is much less relevant to the not only the accuracy and reproducibility of the health impact of air pollution than RSPM and smaller measurements but also data analvsis; timely publication particles. For this reason the trend worldwide is to focus of and access to, raw data: re-examination of site- increasingly on measuring the concentration of smialler ofanacest,-wdtar-xmitonfst- increasingly on measuring the concentration of smaller selection in light of changes in land use patterns- the actual particles which can penetrate deeper into human (as opposed to stipulated) monitoring frcquency; and the respiratory systemis. For examiple. the US Eniviroriinental.. Protection Agency (EPA) discontinued monitoring of TSP positioning of the instruments at a given site. in 1987 in favor of PM,, (particles smaller thian 10 gm), With respect to how data are used, there is often a and achieved its first year of nation-wide monitoring of tendency to collect data at "hot spots," and to base city- PM25 (smaller than 2.5 gm) in 1999. wide policies on the data collected at sites that rank among the most polluted. However, such an approach is America or Europe are used after some adjustments, not optimal for addressing air pollution at the least cost but they could seriously under-estimate emission levels to socicty. in South Asia. One consequence is that emission Identifying Sources factors used vary markedly from study to study, sometimes differing several-fold. Broadlyspeaking, there arc two approaches to quantifying * Industrial emissions depend on a number of factors, the contributions of pollution sources to human exposure. including how the plants are run and maintained. When 1. The first approach, dispersion modeling, starts with the emission factors are based on the data provided emissions from different sources (emissions inventory) by manufacturers (of boilers, for example) who tend and calculates ambient concentrations in the vicinity to assume very good equipment maintenance, the of the "receptor" (where ambient concentrations are factors may be seriously under-cstimated. measured). The final results should match the mneasured ambient concentrations, but are often * The numbers multiplying the emission factors, such significantly below them. Ambient concentrations are as the amount of fuel used, can be estimated only used to calibrate the models for running future roughly in many cases. When the amounts of transport scenarios, fuels sold in a city are compared to those calculated from the vehicle fleet data, for examplc they have 2. The second approach, receptor modeling, analyzes been known to differ markedly. particles in the atmosphere at a given location and matches their charactristics with those of chemically * For certain source categories-re-suspended road distinct source types (finger-printing), dust, refuse ancd leaf burning, generators, to mention a few-data are typically not available. As a result, they Thesc two approaches should give the same results, but mabeudrstaedooitdalgth,an examples of conducting and comparing studies using the correspndingly mte fromio thersr ar - two approaches are rare even in developed country cities estimated in relative percentage terms. where much more detailed data are available (see Box- 1). The two approaches-constructing an emissions Two important points are worth flagging. The first is that inventory followed by dispersion modeling in the first, and an emissions inventory, however accurate, should not be chemical mass balance receptor modeling in the second- the basis of policy formulation. What ultimately should and their applications to cities in India are discussed below. drive policy is not which source is emitting more, but which source is likely to lead to greater exposure to health- Approach 1 damnaging pollutants. A coal-fired power plant at the edge Emissions inventory of a city with a tall stack may in absolute tonnage be the Emissions inventories have been developed in large cities largest emitter of particles, but may be contributing less- such as Delhi and Mumbai. However, they are limited by from the point of view of overall human cxposure-than, the lack of availability of necded data and are hence for example, all the households burning biomass. sketchy for the following reasons. A commnon mistake is not only to rank different sources * Emission factors suited to Indian cities are often not based on an emissions inventory, but to add up all the available. In their absence, the factors from North pollutants (regardless of their toxicity to human health) in Box 1. Confounding conventional wisdom: Lessons from the United States One of the most extensive comparisons of the two approaches to source apportionment is a study in Colorado, USA [41 which examined source contributions to PM25. The available emissions inventory indicated that diesel accounted for two-thirds of on-road vehicle PM25 emissions and gasoline the remaining one-third. However, the use of the chemical mass balance model suggested that diesel actually accounted for only a third and gasoline two-thirds, and that PM2,5 emissions from gasoline vehicles were seriously under-estimated, both with respect to diesel and on an absolute basis. The discrepancy was due mainly to the presence of gasoline "smokers" and high emissions during cold start. A recent study conducted in southern California [51 found that some gasoline-fueled passenger cars emit as much as 1.5 grains per kilometer, an emission level normally associated with heavy-duty diesel vehicles. Comprising only 1 to 2 percent of the light-duty vehicle fleet, these gross polluters were estimated to contribute as much as one- third to the total light-duty particulate emissions. It is possible that the proportion of "smoking" gasoline vehicles is much larger in South Asia. U weight before doing so. This almost always leads to the Carbon analysis, in contrast, has not been conducted until conclusion that road traffic is by far the largest contributor recently, the first study being carried out only in 2000- to urban air pollution, because in absolute tonnage, carbon 2001. Carbon analysis is useful for estimating combustion- monoxide (CO) dominates all other pollutants, and the generated particles. Comparison of black (also called majority of CO is from vehicles. But the toxicity of CO is elemental) carbon and organic carbon may help to much lower on a weight basis than those of other pollutants, distinguish between the combustion of biomass and fossil so that these results cannot be correlated withhealth effects. fuels. Trace organic analysis (identifying key The second point is that the science of the health impact hydrocarbons) is an important tool in receptor modeling, of particulate air pollution increasingly points to the but this has not been carried out in India for small particles. importance of ultra-fine particles, and significant Carbon analysis of 15 RSPM samples collected in Delhi contributions of combustion processes to the size fractions between August 2000 and February 2001 [ 1] showed that now considered most damaging to public health. A study total carbon constituted 36% of RSPM. Total carbon in the United Kingdom reported that road traffic nationally averaged 122 gg/m3, and black carbon 72 gg/M3. These contributed 25% of primary PM,0 emissions, but the figures are high by any measure, and the high proportion relative importance of road traffic emissions increased of black carbon points to significant contributions from with decreasing particle size and road transport accounted the combustion of fossil fuels. for an estimated 60% of PM.1 [6]. The question of which Summary of source apportionment of sources are contributing most to public health damage depends critically on the particle size range for which P source apportionment studies are carimed out. No more than a dozen source apportionment studies appear to have been conducted in India, and most of them Dispersion modeling identify major sources without quantification. The two A limited number of studies have carried out dispersion main approaches to source apportionment mentioned modeling in India. Most have not looked at the chemical above have been utilized. The majority have concentrated transformation of pollutants (such as secondary particulate on TSP. The selection of TSP, which includes a large formation from sulfates and nitrates). The majority have fraction of coarse particles, tends to highlight the examined pollutant concentrations-typically CO, NO2 importance of wind-blown dust and other natural or lead when leaded gasoline was still used in India- sources of particles at the expense of anthropogenic along traffic corridors. The objective of these studies, sources, although the latter are much more damaging to which met with varying success, was not to quantify public health. source contributions, but to compare model outputs with There has been only one study attempting to investigate the actual concentration measurements at selected source contributions to PM13 [7]. About 20 elements were locations in order to validate the models. As such, measured in Mumbai over a year and factor analysis was dispersion modeling in India is in its early stages of policy carried out identifying four pollution sources: re-suspended relevance. dust, refuse and vegetation burning, sea-salt and road Approach 2 traffic. As expected, soil-derived dust was found to be a major source of larger size particles whereas refuse The first step in any chemical mass balance receptor burning and road traffic were identified as major sources modeling is detailed chemical analysis of receptor and of smaller size particles. source samples. The level of detail required is Similarly, only one study appears to have built upon an considerable, presenting difficulties. For source attribution, emissions inventory and dispersion modeling to quantify source profiles are needed. It would be straightforward source contributions. The Urban Air Quality Management if one compound or element served as a tracer for a single Strategy in Asia, UJRBAIR, used a multisource Gaussian source type, and that tracer was not present in any other model to estimate TSP concentration distributions in source type, but this is rarely the case. It may even be Mumbai [8]. Based on the estimates of PM,J/TSP ratios that sources cannot be "fingerprinted": similar sources for different sources, the study suggested that 30% each may not have similar profiles, or different source of was from background and vehicle exhaust, categories may have similar profiles. respectively, 20% from road dust re-suspension, 15% The majority of the studies conducted in India to date from area sources (domestic fuel combustion, small have focused on trace metal and water-soluble element industries, stone crushing and refuse burning), and 3% analysis. Because of the widespread availability of leaded from large and medium-size industrial plants. These gasolineinIndiauntil 2000,anumberof studies examined estimates contain large uncertainties and should be lead in particles along traffic corridors. interpreted with caution. U There has been one study [9] using chemical mass balance * Those findings from other countries that challenge the to examine source contributions to TSP in Mumbai. The conventional wisdom (Box I) should inform US EPA chemical mass balance receptor model was researchers and policymakers. used. Model results were unsatisfactory at highly polluted References sites, suggesting that US EPA profiles were not suitable for India. I . Tata Energy Research Institute. 2001. "Review of past Another study [10] used particle size distribution data and on-going work on urban air quality in India," report rather than chemical analysis to attribute sources. Particle submitted to the World Bank, Dcember. Available at size is sometimes indicative of emission sources, with . coarse particles typically from mechanical processes and 2. Available at . formation. The study dividedparticles ranging in size from 3. Data available at . fraction corresponding to secondary particles and products of combustion, and the largest to dust and marine aerosols. 4. Available at . Conclusions 5. T.D. Durbin, M.R. Smith, J.M. Norbeck and T.J. Truex. 1999. "Population density, particulate Data that can be used for policy formulation in urban air emission characterization and impact on the particulate quality management are scarce in India. How to make inventory of smoking vehicles in the South Coast Air best use of the available information, and how to bridge Quality Management District," Journal of Air and the gaps in data and analysis are the two challenges facing Waste Management Association, 49:28-38. researchers and policymakers. 6. Airborne Particles Expert Group. 1999. "Source * Based on epidemiological evidence, monitoring of TSP Apportionment of Airborne Paritculate Matter in isincreasingly abandonedinfavorof PM,(andPM, the United Kingdom," January. Available at elsewhere in the world. These smaller particles should . * Regular monitoring of PM,, and PM25, while being 7. V. Meenakshy, B.S. Ncgi and S. Sadasivan. 1996. an important first step, has a limited role to play in "PM- 10 aerosols at Chembur, Bombay: their elemental policy appraisal because it can merely signal that there concentratioins and sources," Proceedings of the is a problem. Monitoring needs to be supplemented by 5th National Symposium on Environment, March, studies to identify sources and assess effects on public pp. 227-230. health. No single methodology will answer all or even 8. Jitendra Shah and Nagpal Tanvi, ed. 1996. "Urban most questions, and instead a wide range of approaches Air Quality Management Strategy in Asia URBAIR: will be needed. Collaboration between governmient Greater Mumbai Report," October, Washington D.C. agencies and scientific institutes to this end should be given high priority. 9. V.K. Sharma and R.S. Patil. 1994. "Chemical mass * The most important gap in the work to date in India on balance model for source apportionment of aerosols in Bombay" Environmental Mon itoring and particulate source apportionment is the near-complete y, g lack of data on emissions from the area sources listed Assessmentl 29(1): 75-88. above underURBAIR. This gap has ledtoapotentially l0.V.K. Sharma and R.S. Patil. 1992. "Size distribution biased focus on emissions control in the transport of atmospheric aerosols and their source identification sector, and can be systematically addressed in a using factor analysis in Bombay, India," Atmospheric relatively short time. Environment 26B(l): 135-140. This briefing note was prepared in March 2002 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 . Forfurther information, contact Sameer Akbar (sakbar@worldbank.org) or Masami Kojima (mkojima@worldbank.org) about the program.