Comparative Health Risks of Domestic Waste Combustion in Urban

Aug 28, 2007 - A major source of uncertainty is that of the fraction of waste burned in the open. The analysis presented here assumed 10%. At this lev...
0 downloads 0 Views 195KB Size
Environ. Sci. Technol. 2007, 41, 6847-6853

Comparative Health Risks of Domestic Waste Combustion in Urban and Rural Slovakia JANA KRAJC ˇ OVIC ˇ O V AÄ * Slovak Hydrometeorological Institute, Bratislava, Jese´niova 17, 833 15 Bratislava, Slovakia ALAN Q. ESCHENROEDER Harvard School of Public Health, Department of Environmental Health, Lincoln, Massachusetts

This paper addresses the health risk incurred by two alternative waste management schemes: open burning of household waste in barrels practiced in rural Slovakia and controlled municipal waste combustion in the city of Bratislava. Using agricultural land use data and village population data we formulate three prototype villages, each representing about one-third of the rural population. The two configurations of the controlled combustion are an outdated municipal waste incinerator (MWI) and a modern waste-to-energy (WTE) plant equipped with modern air pollution control devices. These configurations actually exist(ed) in Bratislava, Slovakia at the same site, but in different time frames. The CALPUFF model provides direct exposure data and the EMERAM software (developed in this paper) computes indirect exposure. A major source of uncertainty is that of the fraction of waste burned in the open. The analysis presented here assumed 10%. At this level, the cancer risk from open burning ranges from 10 to 80 times the commonly regarded de minimus value of one in a million. This means that under the U.S. contemporary regulatory culture, some regulatory action to control or enforce the burning ban would be expected. Cancer risks from the incinerator ranged from 7 to 371 in a million while the WTE risks were below 1 in a million. Cancer risks from open burning are higher than those of the WTE plant and at the same time affect a larger portion of concerned population.

1. Introduction Methods of waste management most widely used in the developed world are landfilling and controlled combustion. A portion of the global waste stream is eliminated by uncontrolled burning, which is a common practice of waste disposal in rural areas of many countries. The work reported in this paper compares the health risks of open burning with those incurred by incineration. The fraction of the waste stream involved with this practice is rather uncertain as it probably varies from one country to another, depending on many factors, e.g., population density, rural/urban ratio, behavioral habits, education, and existing legislation and its enforcement. Recent surveys show that 16-48% of rural population practice open burning in the United States, and * Corresponding author tel: +421 2 59 415 208; fax: +421 2 54 774 374; e-mail: [email protected]. 10.1021/es0627186 CCC: $37.00 Published on Web 08/28/2007

 2007 American Chemical Society

54% of all population does so in Mexico (1). The European Union Dioxin Emission Inventory document (2) assumes that only 0.25% of total domestic waste is burned illegally in former EU countries. At the same time these authorities admit that there are no data on the frequency with which this activity is performed. The 2005 EU dioxin inventory (3) assumes an average of 10 kg of waste per person illegally burned annually, which for Slovakia means approximately 5%. We believe that this estimate understates the problem as it appears based on personal observation in rural Slovakia. United States Environmental Protection Agency (U.S. EPA) household waste barrel burning simulation tests demonstrated that 2-40 households burning their trash daily in barrels can produce average polychlorinated dibenzo(p)dioxins and dibenzofurans (PCDD/F) emissions comparable to those of a 200 ton/day municipal waste combustion facility equipped with modern air pollution control devices (4). The fact that emissions from open burning are released at the ground level, possibly resulting in decreased dispersion, raises concerns about the human exposure to a range of pollutants: organic, inorganic, and particulate matter. The rural population of Slovakia is not evenly dispersed over the farmlands, but rather it is concentrated in densely populated villages made up of family houses with kitchen gardens. Villages are surrounded by farmlands. While open-burning scenarios are designed analytically to represent the statistical distribution of population in Slovak rural areas, the municipal waste combustors are modeled in the real environment of the city of Bratislava. Actually, we were able to make two comparisons: one with the municipal waste incinerator (MWI) which had been used until the year 2002, and the other with the modern waste-to-energy facility (WTE) which replaced the MWI after that year. The following sections describe the process of risk assessment applied to each of the above-mentioned waste management practices which is in line with the U.S. EPA methodology (5).

2. Methods 2.1 Village Burn Barrels Hazard Identification. 2.1.1 Formulation of Prototype Villages. As we want to quantify the problem of open burning on a larger than local scale, this is not feasible to do for each particular village in a region. The assessment of risks will be based on a more general concept of synthetic villages representing a statistical size distribution of the settlement clusters. Data from the population and housing census of the Bratislava region, Slovak Republic (6) formed the basis of the frequency distribution of village populations. We trisected the cumulative population distribution and established midpoint village populations. We assign the populations of the three to represent the first, second, and third tertiles of the regional demographics. Each of the three prototypes (named according to their size as MIN, MOD, and MAX) represents about 30,000 persons in the region (for more details on the method see Supporting Information). The use of a distribution of village size is intended to give an approximate characterization of variability. Since village population densities are about constant, land areas (needed for source descriptions) are nearly proportional to populations (with R2 ) 0.845). 2.1.2 Emission Sources Formulation. For modeling purposes, some simplifying assumptions are necessarily adopted about the spatial and temporal configuration of sources. Based on observations, our study assumes burning of waste during daytime hours on dry days throughout the whole year VOL. 41, NO. 19, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6847

TABLE 1. Synthetic Villages: Summary of Parameters

village

population

area (m2)

MIN MOD MAX

959 1630 3601

582,000 986,000 2,170,000

radius (m)

number of households

number of sources

sources spacing (m)

total waste production (t year-1)

430.5 560.2 831.0

239 407 900

13 24 52

200 200 200

199.5 339.0 749.0

FIGURE 1. Design of synthetic villages: MAX village modeling domain with grouped sources (stars) and receptors (dots within village border). as the most probable scenario. The villages are of circular shapes. Sources are grouped as if approximately each 17 houses share one burn-barrel to which they bring their waste for burning. This will reduce the original number of sources 239, 407, and 900 (for MIN, MOD, and MAX respectively) to 13, 24, and 52. The sources are distributed uniformly throughout the villages on a 200 × 200 m grid. The emissions are uniformly distributed over the 144 dry days of the year 2001 during the daytime period of 9 to 18 h in winter and 8 to 20 h in summer half of the year. To estimate how much waste is burned in each of the villages, we link the total waste stream to population by application of the Slovak national average generation factor of 208 kg per person per year (7) averaged over 1992-1997. We assume that 10% of the waste generated in the villages is burned in barrels as a first approximation. The parameters of the prototype villages described above are summarized in Table 1. A burn-barrel emission study performed by U.S. EPA (4, 8) used a 208 L steel barrel with 24 ventilation holes placed near the ground. The mass of waste burned ranged from 6.4 to 13.6 kg. Mass balances of combustion products derived from experimental results form the basis for determining effective “stack” parameters for plume rise calculations. Combustion stoichiometry was employed to determine this 6848

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 19, 2007

mass balance based on the chemical composition of the air and fuel. The composition of waste was based on the typical percentages of various materials in the household waste of New York State citizens. From the information on the composition of household waste in Slovakia (9) we assume that the composition of waste can be considered similar to the non-recycler scenario used in the above-mentioned U.S. EPA study. 2.1.3 Receptor Configuration. There are two kinds of receptors considered in the village simulation: village receptors inside the village area where the burning takes place and agricultural receptors in the outer circle of the village where the croplands and pasture lands are located. In the air dispersion modeling, the agricultural receptor spacing is determined by the 500 m grid size while village receptors spacing is 80 m. The receptor configurations are illustrated in Figure 1 showing how population density determines the different spacing depending on village geographics in the case of the MAX village as an example. The dotted circle delineates the outer limit of agricultural receptors. 2.2 Municipal Waste Combustor Hazard Identification. 2.2.1 Emission Sources. The public solid waste management company, Odvoz a likvida´cia odpadu (OLO) operates the municipal waste combustor located in Bratislava about 2.5

TABLE 2. Summary of Exposure Pathways for Combustors and Burn Barrels population subgroup (scenario) exposure pathwaysa inhalation soil ingestion maternal breast milk consumption homegrown fruit & vegetables homegown poultry and eggs homegrown pork commercial fruit & vegetables commercial poultry and eggs commercial pork commercial beef and dairy products consumption of water consumption of fish a

urban resident & child

suburban resident & child

village resident & child

village fishing enthusiast

X X X

X X X X

X X X X X X

X X X X X X

O O O O

O O O O

O

O X

X - pathway from local contamination; O - inter-zone transfer of risks.

km from the largest and densely populated residential city district of Petrzalka. After 23 years of operation as an incinerator, it was upgraded to a modern WTE facility in 2003. In addition to the steam supply to nearby greenhouses it generates 5.4 MW of electrical power and, operating for 7500 h per year, can process 134,000 metric tons (tonnes) of municipal solid waste (10). A design objective of the new facility is to achieve emissions that comply with European Union limits (11) governing new plants built after December 28, 2004 (i.e., 0.1 ng/m3 toxic equivalent dioxins and furans averaged between 6 and 8 h). U.S. EPA’s AP-42 documentation (12) provides data on pollutant emissions and particle size distributions for waste combustion. 2.2.2 Receptor Configuration. In addition to discrete receptors which are important for the construction of population risk profiles, a rectangular mesh of gridded receptors with 1 km spacing on the domain of 30 × 33 km is also included. Grid-based receptor results provide insights as to the distributions of concentration and deposition fields. Since computations are carried out at each receptor point, interpolation is unnecessary. 2.3 Exposure Scenarios and Environmental Pathways. The ultimate fate and transport of the atmospheric emissions depends upon environmental partitioning of persistent pollutants. Quantitative descriptions necessitate the inclusion of a rich assortment of human exposure pathways. Accordingly, the three routes of human intake must be considered; namely, inhalation, ingestion, and dermal absorption. 2.3.1 Village Burn Barrels. Adult and child resident scenarios will be considered for the average village population by accounting for differences in body weight, intake rates, and absorption efficiency. Although there may be people earning their living as farmers in villages, their diet and the origin of food items hardly differ from those of ordinary villagers who grow fruit and vegetables in their kitchen gardens and poultry and pigs in their backyards. As some of the real Slovak villages are situated in the vicinity of ponds or other kinds of water bodies, a fishing enthusiast will be evaluated as a worst case scenario. 2.3.2 Municipal Waste Combustor. All population is exposed to the combustor emissions via four pathways: direct inhalation, soil ingestion, soil dermal contact, and maternal breast milk consumption. However, as the soil dermal contact contribution to the total exposure is very small, it is neglected due to the uncertainties associated with this pathway (5). We assume that a portion of fruit and vegetables consumed by suburban residents is grown in the kitchen gardens, but no meat, eggs, or milk are supposed to be produced at the sites where the deposition occurs. Due to lack of exposure

data on fishing water bodies we excluded the fishing enthusiast scenario from the calculations. 2.3.3 Interzone Risk Transfers. Human consumption of animal products and plants which are not homegrown, including poultry, eggs, pork, beef, and dairy produced commercially and sold in markets and shops contribute to inter-zone risk transfers. It means that they transfer the risk from the rural areas, where they are likely to be produced on the agricultural land, to the cities and villages outside these areas. Thus the open burning not only produces the incremental cancer risk in rural areas, but it also contributes to the risk in urban centers. Table 2 summarizes the environmental pathways assigned to MWC and village burn barrels scenarios, including the inter-zone transfers. Water ingestion is not considered to be an exposure pathway because the preponderance of drinking water originates from groundwater reservoirs. More detailed information about the rationale behind the pathways included in different scenarios is available in the Supporting Information. 2.4 Exposure Assessment Tools. The modeling starts with implementation of the air dispersion model, which calculates two-dimensional fields of ground-level annual air concentration, dry and wet deposition. Inputs consist of source/ receptor characteristics, emissions, and meteorology. The output parameters are then used as inputs to the next chain of models calculating the contaminant concentrations in the environmental media. Dose-response data based on toxicology or epidemiological research are then used to characterize the risks associated with human exposure to those concentrations. Two computer models are used to perform the above calculations: CALPUFF atmospheric dispersion model and EMERAM (environmental media exposure and risk assessment model). CALPUFF (13, 14) is a U.S. EPA model capable of treating the time varying emissions in case of open burning. EMERAM is a computer software based on a system of models of fate and transport of chemicals in different environmental media. For the indirect pathways it uses expressions embodied in the protocol developed by Region VI of the U.S. EPA a decade ago (5). Over the years the U.S. EPA work has been widely circulated for peer review and applied to multimedia fate problems. Our utilization of this methodology substitutes CALPUFF for ISC as the air model. The total model system calculates the exposures to chemicals of potential concern (COPC) for all receptors through all relevant environmental pathways. 2.5. Chemicals of Potential Concern (COPC). COPC to be studied were selected using two criteria. As our interest lies in the long-term health effects, chemicals which are VOL. 41, NO. 19, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6849

TABLE 3. Maximum and Median Values of ILCR and HI for All Settings and Scenarios Modeled in the Studye ILCRa (cases per million) settings OLO -

scenarios

max

HIb: adult (unitless)

HIb: child (unitless)

median

max

median

max

median

106 2

0.0 0.0

0.0 0.0

0.0 0.0

0.0 0.0

0.0 0.0

0.0 0.0

0.0 0.0

0.0 0.0

16 76

0.8 2.2

0.7 2.0

1.2 2.0

0.9 1.8

13 59

10 57

0.5 1.6

0.4 1.5

0.7 1.4

0.6 1.3

10 37

6 34

0.4 1.0

0.3 0.9

0.6 1.0

0.4 0.8

MWIc

suburban resident urban resident

OLO - WTEd

suburban resident urban resident

MAX village

village resident village fishing enthusiast

20 80

MOD village

village resident village fishing enthusiast

MIN village

village resident village fishing enthusiast

371 7 0.8 0.0

0.2 0.0

a Incremental lifetime cancer risk. b Hazard index. c Municipal waste incinerator (outdated). d Waste-to-energy plant. e 0.0 values indicate HIs which were very low. Numbers in bold show the hazard indices with values equal or greater than 1.

supposed to induce carcinogenic effects or to cause chronic functional disorders were selected. In the regulatory context this means the chemicals emitted have reference doses/ concentrations and/or cancer slope factors that are referenced in scientific or regulatory literature. The second criterion was the availability of combustor emission data, and emission factors (burn barrels) for those chemicals. Ref 8 reports very detailed emission factors for burn barrels, so the first criterion was applied to shorten the COPC list. COPC included in the study were thus reduced to PCDD/F, PCB, PAH, and a selection of SVOC and heavy metals. Due to its potential fatal effects, increased mortality due to inhalation exposure to PM2.5 was also evaluated separately from the carcinogenic and non-carcinogenic risk assessment. The details are discussed in the Supporting Information.

3. Results This section presents the risk characterization results in somewhat of an aggregated form. Condensation of the results becomes necessary because of the large volumes of data distributed across chemical species, space, and time. To conform with journal space limitations, the discussions below only highlight the significant results; therefore, the reader desiring further detail is referred to the tables and figures summarizing the findings of the calculations, and to more detailed tables in the Supporting Information. To provide some rational frame of reference we compare the risks from open burning with actual cases of combustors in Bratislava. 3.1 Cancer and Non-cancer Risks. We are most interested in the maximum risks and hazards in both cases, the distribution of risks and hazards throughout the affected populations, the contributions of different pathways and chemicals to the total risks, and the degree of risk transfer from the rural agricultural areas to both populations. Tables 3 and 4 present incremental lifetime cancer risks (ILCR), and hazard indices (HI) expressing non-cancer health effects for each village and combustor scenario. The ILCR expresses the probability of generating a tumor, whereas the HI is the ratio of exposure to some health-based exposure criterion such as the reference dose. The incinerator has produced maximum ILCR of 371 cases per million. Technology innovation brought by the upgrade to WTE reduces the risk below 1 per million. The maximally exposed receptor is the suburban municipality of Rusovce, situated southeast of the source. Maximum ILCRs in villages associated with the open burning of waste reach 10-80 per million, with the largest risk associated with the MAX village fishing enthusiast scenario. 6850

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 19, 2007

TABLE 4. Incremental Lifetime Cancer Risk (Cases per Million) for MAX Village for a Maximally Impacted Fishing Enthusiast Receptor: Contributions of Each Chemical of Potential Concern (COPC) through Particular Exposure Pathwaysa COPC

inh

sing

plant

Anim

fish

bmilk

total

PCB PAHs as BaP benzene acetaldehyde formaldehyde Cd As Cr VI PCDD/F total

0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

0.0 0.5 0.0 16.9 0.0 0.0 0.5 0.0 1.2 19.0

0.0 0.4 0.0 0.0 0.0 0.0 0.0 0.0 0.1 0.5

48.5 1.4 0.0 0.0 0.0 0.0 0.0 0.0 0.0 49.9

10.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.3 10.3

58.5 2.3 0.0 16.9 0.0 0.0 0.5 0.0 1.6 79.9

a Pathways included are abbreviated as follows: inh, inhalation; sing, soil ingestion; plant, plant ingestion; anim, meat and animal products ingestion; fish, fish ingestion; bmilk, maternal breast milk ingestion.

Table 4 shows the distribution of the total ILCRs for the maximally impacted MAX village receptor into the detailed contributions of each COPC through each pathway. It can be seen that ILCR in villages is mainly caused by polychlorinated biphenyls (PCB) through fish and breast milk ingestion pathways and acetaldehyde through plant ingestion. The remaining risks can be attributed to PCDD/F and PAHs. Similar tables have been created for each village and combustor settings ILCRs and HIs (see Supporting Information). They show that for the combustor cases the most important exposure pathway is plant ingestion. The largest contribution to the ILCR is associated with the PCDD/Fs and PAHs for the incinerator and PCDD/F and arsenic for WTE. While HIs are negligible for both municipal waste combustor cases, maximum HI values exceed 1 in cases of fishing enthusiast scenario in all village sizes and in the case of village resident child in MAX village. They are primarily caused by PCB and methylated Hg exposure through the fish pathway and acetaldehyde exposure via plant ingestion. HI and ILCR profiles plotted in Figures 2 and 3 illustrate the distribution of total risks throughout the populations. The ILCR profile curve for WTE case approximates zero, while for MWI it is skewed to the right reflecting large nonuniformities in exposure, associated with the geographical distribution of urban and suburban populations of Bratislava. Approximately half of the Bratislava population was facing the risk lower than 4 cases per million when the incinerator was in operation, while now all the population is at risks near zero.

FIGURE 2. Hazard index (HI) profiles for villages and municipal waste combustors (*All municipal waste combustor (MWC) scenarios result in HI close to zero).

FIGURE 3. ILCR profiles for villages and combustors (MWI, municipal waste incinerator; WTE, waste-to energy plant). On the other hand, the distribution of risk is almost linear for the villages, giving an “average” cancer risk of about 15 per million for village resident scenario and 70 per million for fishing enthusiast. The distribution is given by the nature of the emission sources which are evenly distributed throughout evenly populated villages. The stepwise shape of risk profiles for fishing enthusiast scenario is caused by the relatively high non-cancer risks arising from fish consumption, the contribution of which is constant for all receptors in the village of a particular size. The cancer risk transferred to Bratislava population through the consumption of commercially produced agricultural products in rural areas can reach values up to an additional 4.5 cases per million.

3.2 Increased Mortality due to the Atmospheric Exposure to PM2.5. Several epidemiologic studies have been performed in the past relating PM2.5 exposure to premature mortality; some of them are evaluated in ref 15. According to an American Cancer Society study, which appeared most reliable, relative risk of 1.12 has been calculated for a 24.5 µg‚m-3 increase in annual mean PM2.5 concentrations, corresponding to 0.5% increase in premature mortality per 1 µg‚m-3. Applying this number to the peak annual PM2.5 concentrations we obtained elevation of the annual mortality from all causes of 0.0027%, 0.0017%, 0.0013%, 0.0006%, and 0.0004% attributable to MAX, MOD, MIN, incinerator, and WTE plant, respectively. Tto place these results in the same frame of reference as the ILCRs, we convert percentage into cases in a million and multiply by years per lifetime (70 years). VOL. 41, NO. 19, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6851

This gives maximal equivalent risks of 1900, 1200, 900, 400, and 29 for the five scenarios enumerated respectively above. It is important to note that these risks are greater than the corresponding chemically induced cancer risks, and that they are not chemically connected since they depend on bulk mass of PM2.5 alone.

4. Discussion There is a number of uncertainties associated with our assumptions; we are able to quantitatively assess their influence on the final risk for few of them and the direction of their influence for some. For some of them, however, we only know they exist, but, unfortunately, we do not have enough knowledge of each case to assess either their magnitude or the direction of their influence. 4.1 Waste Statistics. 4.1.1 Amount of Waste Open-Burned. This source of uncertainty may be the most influencing influential factor. As open burning of waste is illegal, there are no statistical data which would give us guidance to the assessment of the portion of household waste which is subject to open burning. The figure chosen for this study is 10%, but a value of 100% of the officially reported household waste production can be equally real in some places, bringing the open-burning risk figures up to 10-fold the values presented in Section 3. Since the emission rate, fate-transport, and the dose-response equations are linear in concentration, the sensitivity of the numerical results to this percentage follows a simple direct proportionality. 4.1.2 Composition of Waste. So far no official studies have been done in Slovakia which would assess the composition of household waste. Some information concerning the composition of waste in a Slovak city of Kosice can be found on the World Wide Web (9). This source reports rather high percentage of food waste, which is not likely to be burned in rural areas. Taking into account this fact, the composition of rural domestic waste may be considered similar to that of the non-recycler scenario of ref 8. However, rural domestic waste can contain additional items such as agricultural foils, gardening refuse, construction refuse, tires, and other ad hoc flammable material. Most of these will probably increase the volume of waste burned as well as the production of harmful COPCs. They may even introduce COPCs which were excluded from the study due to low emission levels during the U.S. EPA tests. 4.1.3 Burning Conditions. In contrast to the highly controlled conditions during the U.S. EPA tests, the actual burning is a very wild activity; a portion of open burning practitioners do use metal barrels similar to those used in the U.S. EPA study, but frequently we can observe burning material in heaps on the ground. It is not unusual to see household waste burned together with still wet gardening refuse. 4.1.4 Emission Rates. WTE plant emission rates were computed based on AP-42 (12). This publication does not list emission factors for PCB, PAH, or other potentially hazardous organics other than PCDD/F. Due to these missing emissions data, the risks and non-cancer hazards for the WTE plant are likely to be underestimated. Analysis of a previous risk assessment for a dry scrubber fabric filter equipped combustor (16) gives some insights as to the potential sizes of the underestimates in the present study. It included risks of PAHs and PCBs. For various exposure scenarios the ILCRs of PAH + PCB represented 50-70% of the total, and the HIs of PAH + PCB represented 0.3-10% of the totals. Thus the omission of other semivolatile organic compound emissions could be significant in the calculation of ILCRs, but is likely to be considerably less so in the case of HIs. Although this may seem counterintuitive, they represent the findings quoted in ref 16. 6852

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 19, 2007

4.2 Dispersion and Deposition Modeling. 4.2.1 Meteorological Factors. For Bratislava modeling, the most important factors are probably those related to the length of the modeling period. If we were to model a period of several years, we would probably get smoother isolines of concentration and deposition fields which would also be reflected in smother spatial distribution of risks and hazards. For the open burning simulations we used flat terrain with no obstacles, as the modeled villages were based on statistical approximations. In the real world, many villages are situated in valleys or near lakes and rivers with terraininduced circulation patterns very different from those used in the study. 4.2.2 Sources and Receptors Configuration. Apart from the meteorology which directly influences the dispersion, other factors bringing uncertainty to the open burning simulations relate to the configuration of barrel sources. In order to save computer time, sources were “grouped” as described in Section 2, and regularly distributed on the village residential area with the spacing of 200 m. As we are interested in the concentrations and depositions between the barrel sources, a dense mesh of receptors is needed to capture the concentration and deposition fields which, at such small distances from the source, are characterized by large variability. Again, we are restricted by the computational time, so the receptor spacing used in simulations is 80 m. This is still a very coarse mesh, so, naturally, we were interested in the degree to which the annual values of concentration and deposition are sensitive to the receptor spacing as well as to the grouping of sources. We made two additional simulations in which only a smaller, zoomed area of the size of spatial coverage of a single grouped source was modeled. This area was represented by a domain of 350 × 350 m, covered with the receptor mesh of 10 m resolution. The first simulation involved one grouped source in the center of the domain and was performed to get the basis for comparison with the second simulation, involving 16 sources of the strength of 1 /16 of the grouped one (grouping close enough to that used for MIN, MOD, and MAX simulations), with the spacing of 50 m. Resulting annual concentration and deposition medians were 30-40% higher in the case of “ungrouped” sources, which better represent the real settings. The maximum values are, quite logically, higher in the grouped source case, and so is the spread of the values. These findings indicate that the inhalation exposure of individuals near barrels (e.g., those performing burning) relatively strongly depends on the receptor mesh resolution. Though the inhalation pathway forms a minor contribution to the cancer and non-cancer risks, using a finer receptor mesh may have a large positive influence of the PM2.5 increased mortality results. 4.2.3 Temporal Distribution of Burning. As it is impossible to specify exact times and lengths of burns, an alternative method described in Section 2 has been applied, assuming that burning is happening most probably during dry days in daytime hours. Actually, it is not so unusual to observe burning before, after, or in between rains, or in late evenings. 4.2.4 Dry Deposition Velocity. Although CALPUFF is able to implement a resistance model for gas dry deposition, the species parameters needed for computations are unavailable for most of the chemicals involved in the study. Therefore, a user-specified dry deposition velocity of 3 cm s-1 has been used in the computations. This velocity is recommended by U.S. EPA (5) as an extremely conservative value for the human health risk assessment. As the dry deposition velocity depends on many factors, currently we have no way to assess the degree of overestimation expressed by the recommended value, or whether it is an overestimation at all. As the indirect exposure pathways are dominant risk factors and media concentrations directly depend on deposition rates, the

assessment of the dry deposition velocity plays a critical role in final risk values. This is especially true for burn-barrel simulations, as wet deposition is not considered because we assumed no open burning during precipitation events. 4.3 Exposure and Risk Parameters. We used recommended values (5) for most general exposure parameters used in the study. Among those which are location dependent and were derived from national statistics data are human ingestion rates and crop yields. The assumption that fruit, vegetable, meat (with the exception of pork and beef), and eggs consumption rates are fully supplied from homegrown production in villages was based on subjective observation. The consumption rates may differ from the assumption either way and currently there is no way to verify it. Animal feeding habits and consumption rates in villages are also based on subjective observations. 4.3.1 Pond Parameters. As there are no statistics available regarding fish pond parameters, a prototype pond model has been used. Pond surface area divided by the volume of water determines how much the pollutant is diluted in the water. As fish consumption is rather important exposure pathway for, e.g., PAHs, PCBs, or methylated mercury, doubling the pond depth can easily reduce the fish pathway contribution to risks, or vice versa. 4.3.2 Dose Response Parameters. The toxicity values used to connect risk with exposure come from interpretations of animal experiments and from inferences drawn from human epidemiological studies. Intrinsically, these data are extremely uncertain. Animal experiments require at least two dimensions of extrapolation to compare with human responses. One degree of extrapolation involves size, and the other is called interspecies extrapolation. 4.3.3 Exclusion of the Impact of Residual Ash from Open Burning. Due to a large uncertainty associated with the treatment of residual ash, its effect has been excluded from the computation of risks. However, in cases when the ash is applied to gardens it may become an important factor increasing the total risks from open burning. The modeling results indicate that the risks rising from the open burning of waste in rural areas highly exceed those from a modern WTE facility. In absolute values, maximum ILCR from open burning might approach the order of magnitude of the MWI risk in the case of the village fishing enthusiast. Moreover, we have to realize that we are comparing two very different phenomena. A combustor represents a highly concentrated way of waste disposal emitting harmful chemicals from a tall stack. Consequently, high concentrations and deposition fluxes near the points of maximum impact rapidly decrease with the distance from the points. Median values and ILCR profiles (Table 3, Figure 4) tell us that all of the Bratislavsky Kraj rural population of 100,000 is facing ILCR higher than 4 cases/million, with mean value of 15 per million, reaching on average 70 per million for fishing enthusiasts, while none of the Bratislava inhabitants faces risk higher than 1 per million due to the operation of WTE plant. Similarly, HI profiles (Figure 3) expressing noncancer health risks indicate that all rural fishing enthusiasts and about 20% of village children face HI higher than 1, while non-cancer hazards from both combustor cases are negligible. All the ILCR and HI values for rural population were computed assuming 10% of household waste is openburned. Taking into account the uncertainties described in previous section, especially the one concerning the amount of waste open-burned, the open burning risks can conceivably reach values which are 10-fold those discussed above. Another dangerous impact of open burning is supported by the premature mortality increases which are attributable to

exposures to PM2.5. These risks are considerably higher for open burning than for MWC. These findings indicate that open burning of household waste imposes health risks to the rural population in Slovakia which are higher than those arising from the operation of a modern municipal waste combustor. It is an activity with a potential impact on human health which cannot be neglected. The risks calculated here strongly suggest that stricter enforcement of burning bans be undertaken in rural areas.

Supporting Information Available The complexity of this study prevents publication of all the details on modeling, exposure and dose-response data, assumptions, methods, modeling tools, specific chemicals included, and detailed summary of risk characterization results in the space allowed for ES&T articles. This material is available free of charge via the Internet at http:// pubs.acs.org.

Literature Cited (1) Neurath, Ch. Open burning of domestic wastes: The single largest source of dioxin to air? Organohalogen Compd. 2003, 63, 122-125. (2) European Commission. The European Dioxin Emission Inventory Stage II; Volume 3. Assessment of dioxin emissions until 2005; European Commission Directorate General for the Environment: Brussels, Belgium, 2000. (3) Pulles, T.; Kok, H.; Quass, U.; Juery, C.; Matejovicova, J. Dioxin emissions in Candidate Countries; TNO-report R&I-A R 2005/ 054; Netherlands Organisation for Applied Scientific Research: Apeldoorn, Netherlands, 2005. (4) Lemieux, P. M.; Lutes, C. C.; Abbott, J. A.; Aldous, K. M. Emissions of Polychlorinated Dibenzo-p-dioxins, and Polychlorinated Dibenzofurans from the Open Burning of Household Waste in Barrels. Environ. Sci. Technol. 2000, 34, 377-384. (5) U.S. EPA. Human Health Risk Assessment Protocol for Hazardous Waste Combustion Facilities. Vols. I, II and III; Peer Review Draft; EPA530-D-98-001A, B and C; U.S. Environmental Protection Agency, Office of Solid Waste and Emergency Response: Washington, DC, 1998. (6) Statistical Office of Slovak Republic. Numbers of Villages within Various Population Ranges. Population and Housing Census of 2001; 2002 (e-mail communication). (7) Statistical Office of Slovak Republic. Statistical Yearbook of Slovak Republic, 2002. (8) U.S. EPA. Evaluation of Emissions from the Open Burning of Household Waste in Barrels, Volume 1: Technical Report; EPA600/R-97-134a; U.S. Environmental Protection Agency, Control Technology Center: Washington, DC, 1997. (9) Web page: http://www.biospotrebitel.sk/clanok/1250-odpadya-trvalo-udrzatelny-zivot.htm; accessed November 2006. (10) Web page: http://www.bratislava.sk; accessed December 2001. (11) European Communities. Directive 2000/76/EC of the European parliament and of the Council of 4 December 2000 on the Incineration of Waste. Off. J. Eur. Communities 2000, 28.12.2000, L332/91-L332/111. (12) U.S. EPA. Compilation of Air Pollutant Emission Factors: Volume I: Stationary Point and Area Sources; AP-42; U.S. Environmental Protection Agency, Office of Air and Radiation: Washington, DC, 1985. (13) Scire, J. S.; Robe, F. R.; Fernau, M. E.; Yamartino, R. J. A user’s guide for the CALMET meteorological model; Earth Tech, Inc.: Concord, MA, 2000. (14) Scire, J. S.; Strimaitis, D. G.; Yamartino, R. J. A User’s Guide for the CALPUFF Dispersion Model (Version 5; Earth Tech, Inc.: Concord, MA, y 2000. (15) Levy, J. I.; Spengler, J. D. Modeling the benefits of power plant emission controls in Massachusetts. J. Air Waste Manage. Assoc. 2002, 52, 5-18. (16) Eschenroeder, A.; von Stackelberg, K.; Holstein, E. C. Health risk assessment of the proposed RENOVA resource recovery facility in Arecibo, Puerto Rico; Alanova Inc: Boston, MA, 1999.

Received for review November 14, 2006. Revised manuscript received July 3, 2007. Accepted July 10, 2007. ES0627186

VOL. 41, NO. 19, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6853