An examination of the intra-SMSA distribution of carbon monoxide exposure

Although fixed-site monitoring data have been used to estimate the spatial pattern of human exposure, the intra-urban distribution of actual exposure has not been documented. This paper used the data collected during the Environmental Protection Agency's (EPA) field investigation of personal exposure to carbon monoxide (CO) to investigate the nature of the distribution of CO with respect to residential location in the Washington, D.C. SMSA. Dot-distribution maps and analysis of variance were used to document the spatial pattern of individual-level in-home CO concentrations. The results show sampled individuals living in the SMSA center are exposed to statistically significantly higher levels of CO than are those living in the suburbs. The most important implications of this work are for exposure modeling. Further investigation is needed to determine whether incorporation of a geographic component will improve exposure prediction.     

Evaluation of a Probabilistic Exposure Model Applied

The probabilistic National Ambient Air Quality Standards (NAAQS) Exposure Model applied to carbon monoxide (pNEM/CO) was developed by the U.S. Environmental Protection Agency (EPA) to estimate frequency distributions of population exposure to carbon monoxide (CO) and the resulting carboxyhemoglobin (COHb) levels. To evaluate pNEM/CO, the model was set up to simulate CO exposure data collected during a Denver Personal Exposure Monitoring Study (PEM) conducted during the winter of 1982-1983.

This paper compares computer-simulated exposure distributions obtained by pNEM/CO with the observed cumulative

relative frequency distributions of population exposure to CO from 779 people in the Denver PEM study. The subjects were disaggregated into two categories depending upon whether they lived in a home with a gas stove or an electric stove. The observed and predicted population exposure frequency distributions were compared in terms of 1-hr daily maximum exposure (1DME) and 8-hr daily maximum moving average exposure (8DME) for people living in homes with gas stove or an electric stove. For 1DME, the computer-simulated results from pNEM/CO agreed most closely within the range of 6-13 ppm, but overestimated occurrences at low exposure (<6 ppm) and underestimated occurrences at high exposure (>13 ppm). For 8DME, the predicted exposures agreed best with observed exposures in the range of CO concentration between 5.5 and 7 ppm, and over-predicted occurrences below 5.5 ppm and under-predicted occurrences above 7 ppm.     

Evaluation of a Probabilistic Exposure Model Applied to Carbon Monoxide (pNEM/CO) Using Denver Personal Exposure Monitoring Data

The probabilistic National Ambient Air Quality Standards (NAAQS) Exposure Model applied to carbon monoxide (pNEM/CO) was developed by the U.S. Environmental Protection Agency (EPA) to estimate frequency distributions of population exposure to carbon monoxide (CO) and the resulting carboxyhemoglobin (COHb) levels. To evaluate pNEM/CO, the model was set up to simulate CO exposure data collected during a Denver Personal Exposure Monitoring Study (PEM) conducted during the winter of 1982-1983. This paper compares computer-simulated exposure distributions obtained by pNEM/CO with the observed cumulative relative frequency distributions of population exposure to CO from 779 people in the Denver PEM study.

The subjects were disaggregated into two categories depending upon whether they lived in a home with a gas stove or an electric stove. The observed and predicted population exposure frequency distributions were compared in terms of 1-hr daily maximum exposure (1DME) and 8-hr daily maximum moving average exposure (8DME) for people living in homes with gas stove or an electric stove. For 1DME, the

computer-simulated results from pNEM/CO agreed most closely within the range of 6-13 ppm, but overestimated occurrences at low exposure (<6 ppm) and underestimated occurrences at high exposure (>13 ppm). For 8DME, the predicted exposures agreed best with observed exposures in the range of CO concentration between 5.5 and 7 ppm, and over-predicted occurrences below 5.5 ppm and under-predicted occurrences above 7 ppm.     

Effects of Instrument Precision and Spatial Variability on the Assessment of the Temporal Variation of Ambient Air Pollution in Atlanta,Georgia

Abstract

Data from the U.S. Environmental Protection Agency Air Quality System, the Southeastern Aerosol Research and Characterization database, and the Assessment of Spatial Aerosol Composition in Atlanta database for 1999 through 2002 have been used to characterize error associated with instrument precision and spatial variability on the assessment of the temporal variation of ambient air pollution in Atlanta, GA. These data are being used in time series epidemiologic studies in which associations of acute respiratory and cardiovascular health outcomes and daily ambient air pollutant levels are assessed. Modified semivariograms are used to quantify the effects of instrument precision and spatial variability on the assessment of daily metrics of ambient gaseous pollutants (SO₂, CO, NO_x, and O₃) and fine particulate matter ([PM_2.5] PM_2.5 mass, sulfate, nitrate, ammonium, elemental carbon [EC], and organic carbon [OC]). Variation because of instrument imprecision represented 7–40% of the temporal variation in the daily pollutant measures and was largest for the PM_2.5 EC and OC. Spatial variability was greatest for primary pollutants (SO₂, CO, NO_x, and EC). Population–weighted variation in daily ambient air pollutant levels because of both instrument imprecision and spatial variability ranged from 20% of the temporal variation for O₃ to 70% of the temporal variation for SO₂ and EC. Wind rose plots, corrected for diurnal and seasonal pattern effects, are used to demonstrate the impacts of local sources on monitoring station data. The results presented are being used to quantify the impacts of instrument precision and spatial variability on the assessment of health effects of ambient air pollution in Atlanta and are relevant to the interpretation of results from time series health studies that use data from fixed monitors.     

Personal carbon monoxide exposure in five European cities and its determinants

Studies involving carbon monoxide (CO) exposure assessment are mainly based on measurements at outdoor fixed sites or in various indoor micro-environments. Few studies have been based on personal exposure measurements. In this paper, we report results on personal measurements of CO in five European cities and we investigate determinants which may influence this personal exposure.Within the multi-centre European EXPOLIS study, personal exposure to CO, measured every minute for 48 h, of 401 randomly selected study participants (mainly non-smokers) was monitored in Athens, Basle, Helsinki, Milan and Prague. Each participant also completed a time-microenvironment-activity diary and an extended questionnaire. In addition, for the same time period, ambient levels of CO from fixed site stations were collected.There are significant differences in both personal exposure and ambient levels within the five cities, ranging from high values in Milan and Athens to low in Helsinki. Ambient levels are a significant correlate and determinant of CO 48-h personal exposure in all cities. From the other determinants studied (time spent in street traffic, time of exposure to ETS and time of exposure to gas burning devices) none was consistently significant for all cities. Change of the ambient CO levels from the 25th to the 75th percentile of its distribution resulted in a 1.5–2 fold increase of 48-h personal exposure. Short time personal exposure was also studied in order to assess the influence of specific sources. Exposure levels were significantly higher when participants were in street traffic and in indoor locations in the presence of smokers.Personal 48-h exposure of non-smokers to CO varies among urban populations depending primarily on the ambient levels. For a CO source to be a significant determinant of the personal 48-h CO exposure, it has to affect the levels of CO in the person's proximity for an adequate length of time. Activities of individuals affect shorter term personal exposure.     

Modeling of the long-term tropospheric trends of hydroxyl radical for the Northern Hemisphere

The MGO 2D (altitude–longitude) channel photochemical transport model has been applied to elucidate the spatial and temporal behavior of the hydroxyl radical in the troposphere of the northern temperate belt for the pre-industrial (1850) period and the last few decades (1960 and 1995). The relation between the tropospheric OH content and the carbon monoxide, methane, nitrogen oxides emissions during 1850–1995 is studied. The distribution of the carbon monoxide concentration is calculated and validated using the observational data collected in the different locations because of the geographical non-homogeneity of its emissions. The response of the hydroxyl radical concentrations to the non-homogeneity of the CO and other atmospheric species distribution is estimated. The carbon monoxide and methane contributions to the hydroxyl photochemical sink are also evaluated. Because the changes of OH in the troposphere alternate the intensity of methane and carbon monoxide oxidation, the CO, CH₄ and OH lifetime evolution due to the increase of anthropogenic pollution intensity is analyzed and discussed.     

Methodological trends in the design of recent microenvironmental studies of personal CO exposure

This paper describes the designs of three recent microenvironmental studies of personal exposure to carbon monoxide (CO) from motor vehicle exhaust. These studies were conducted sequentially, first in four California cities (Los Angeles, Mountain View, Palo Alto, and San Francisco), then in Honolulu, and, most recently, in metropolitan Washington, D.C. Though study purposes differed, each study faced common methodological issues related to personal exposure monitors (PEMs), quality assurance and data collection procedures, and the selection of microenvironments for study.Two major objectives of the California cities study were to determine the CO concentrations typically found in commercial settings and to define and classify microenvironments applicable to such settings: The Honolulu study measured merchant exposure to CO in shopping centers attached to semienclosed parking garages during business hours and commuter exposure to CO in vehicles (passenger cars and buses) on congested roadways during peak periods. The intent of the Washington study was to develop a model of commuter exposure to motor vehicle exhaust using CO as an indicator pollutant.Certain trends are discernible from reviewing the three studies. There are clearly trends in PEM development that have expanded instrument capabilities and automated data collection and storage. There are also trends towards more rigorous quality assurance procedures and more standardized protocols for collecting exposure data. Further, one can see a trend towards more elaborate indicators for identifying microenvironments for study. Finally, there is a trend towards using personal monitors in public policy review and evaluation.     

Fine particles and carbon monoxide from wood burning in 17th–19th century Danish kitchens: Measurements at two reconstructed farm houses at the Lejre Historical–Archaeological Experimental Center

Carbon monoxide (CO) and particulate matter (PM_2.5) were measured in two reconstructed Danish farmhouses (17–19th century) during two weeks of summer. During the first week intensive measurements were performed while test cooking fires were burned, during the second week the houses were monitored while occupied by guest families. A masonry hearth was located in the middle of each house for open cooking fires and with heating stoves. One house had a chimney leading to the outside over the hearth; in the other, a brickwork hood led the smoke into an attic and through holes in the roof. During the first week the concentration of PM_2.5 averaged daily between 138 and 1650 μg m^?3 inside the hearths and 21–160 μg m^?3 in adjacent living rooms. CO averaged daily between 0.21 and 1.9 ppm in living areas, and up to 12 ppm in the hearths. Highest concentrations were measured when two fires were lit at the same time, which would cause high personal exposure for someone working in the kitchens. 15 min averages of up to 25 400 μg m^?3 (PM_2.5) and 260 ppm CO were recorded. WHO air quality guidelines were occasionally exceeded for CO and constantly for PM_2.5. However, air exchange and air distribution measurements revealed a large draw in the chimney, which ensured a fast removal of wood smoke from the hearth area. The guest families were in average exposed to no more than 0.21 ppm CO during 48 h. Based on a hypothetical time-activity pattern, however, a woman living in this type of house during the 17–19th century would be exposed to daily averages of 1.1 ppm CO and 196 μg m^?3 PM_2.5, which exceeds WHO guideline for PM_2.5, and is comparable to what is today observed for women in rural areas of developing countries.     

Anthropogenic and natural CO2 emission sources in an arid urban environment

Recent research has shown the Phoenix, AZ metropolitan region to be characterized by a CO2 dome that peaks near the urban center. The CO2 levels, 50% greater than the surrounding non-urban areas, have been attributed to anthropogenic sources and the physical geography of the area. We quantified sources of CO2 emissions across the metropolitan region. Anthropogenic CO2 emission data were obtained from a variety of government and NGO sources. Soil CO2 efflux from the dominant land-use types was measured over the year. Humans and automobile activity produced more than 80% input of CO2 into the urban environment. Soil CO2 efflux from the natural desert ecosystems showed minimal emissions during hot and dry periods, but responded rapidly to moisture. Conversely, human maintained vegetation types (e.g. golf courses, lawns, irrigated agriculture) have greater efflux and are both temperature and soil moisture dependent. Landfills exhibited the most consistent rates, but were temperature and moisture independent. We estimate the annual CO2 released from the predominant land-use types in the Phoenix region and present a graphical portrayal of soil CO2 emissions and the total natural and anthropogenic CO2 emissions in the metropolitan region using a GIS-based approach. The results presented here do not mimic the spatial pattern shown in previous studies. Only, with sophisticated mixing models will we be able to address the total effect of urbanization on CO2 levels and the contribution to regional patterns.     

Generalized Rollback Modeling for Urban Air Pollution Control

A general formula is derived that can be used to calculate the reductions in emissions of inert pollutants required to achieve National Ambient Air Quality Standards (NAAQS) and to predict future urban atmospheric concentrations. The derivation incorporates the main features of atmospheric diffusion modeling and takes account of all categories of sources and their spatial distribution. In our previous paper, carbon monoxide (CO) emissions from light duty vehicles were considered separately with the approximation that emissions from other sources of CO would grow and be controlled proportionately to that of light duty vehicles.

The new general formula is applied to Phoenix-Tucson using EPA data. It Is found that Phoenix-Tucson will meet the NAAQS for CO by 1985 if a 12 g/mi light duty vehicle emission standard is adopted. The EPA, using the same data in a modified rollback analysis, had predicted that Phoenix-Tucson, as well as a number of other localities, would not achieve the NAAQS even if the 3.4 g/mi statutory standard went into effect on schedule.

The underlying reasons for these very different predictions can be readily identified by means of the general formula. It is essential that the data and parameters used in these predictions be internally consistent. It is also noted that the current Federal Test Procedure (CVS-CH) for vehicle emissions gives data inconsistent with that needed to predict CO air quality with a correct methodology.     

Modelling the spatial distribution of SO2 and NOx emissions in Ireland

The spatial distributions of sulphur dioxide (SO2) and nitrogen oxides (NOx) emissions are essential inputs to models of atmospheric transport and deposition. Information of this type is required for international negotiations on emission reduction through the critical load approach. High-resolution emission maps for the Republic of Ireland have been created using emission totals and a geographical information system, supported by surrogate statistics and landcover information. Data have been subsequently allocated to the EMEP 50 x 50-km grid, used in long-range transport models for the investigation of transboundary air pollution. Approximately two-thirds of SO2 emissions in Ireland emanate from two grid-squares. Over 50% of total SO2 emissions originate from one grid-square in the west of Ireland, where the largest point sources of SO2 are located. Approximately 15% of the total SO2 emissions originate from the grid-square containing Dublin. SO2 emission densities for the remaining areas are very low, < 1 t km-2 year-1 for most grid-squares. NOx emissions show a very similar distribution pattern. However, NOx emissions are more evenly spread over the country, as about 40% of total NOx emissions originate from road transport.     

Serum PCDD/F concentration distribution in residents living in the vicinity of an incinerator and its association with predicted ambient dioxin exposure

The aim of this study was to evaluate the serum polychlorinated dibenzo-p-dioxin (PCDD) and dibenzofuran (PCDF) concentration distribution in residents living in the vicinity of an incinerator and its association with annual ambient dioxin exposure predicted by an atmospheric dispersion model. A municipal waste incinerator in Northern Taiwan was chosen for this study. This incinerator had been in operation for 6 years at the time of this study. Using the incinerator site as the center, based on the simulated ambient annual average PCDD/F concentrations. Ninety-five volunteers, all live within a radius of 5 km from the incinerator for at least 5 years, who had no occupational exposure potential, were selected based on the population distribution in each district. The average serum PCDD/F concentration for these subjects living within four zones was about 14 pg I-TEQ/g lipid. The serum distribution levels of people of the four study zones, however, were not consistent with the predicted ambient levels. Results also suggest that ambient exposure might not be the most important contributor to serum concentrations when compared to other exposure sources, such as dietary intake.     

Evaluating the effects of elevated levels of atmospheric trace gases on herbs and shrubs: a prototype dual array field exposure system

In the context of global climate change, an understanding of the long-term effects of increasing concentrations of atmospheric trace gases (carbon dioxide, CO(2), ozone, O(3), oxides of nitrogen, NO(x) etc.) on both cultivated and native vegetation is of utmost importance. Over the years, under field conditions, various trace gas-vegetation exposure methodologies with differing advantages and disadvantages have been used. Because of these variable criteria, with elevated O(3) or CO(2) levels, at the present time the approach of free-air experimental-release of the gas into study plots is attracting much attention. However, in the case of CO(2), this approach (using 15 m diameter study plot with a single circular array of vent pipes) has proven to be cost prohibitive (about 59000-98000 dollars/year/replicate) due to the consumption of significant quantities of the gas to perform the experiment (CO(2) level elevated to 400 ppm above the ambient). Therefore, in this paper, we present a new approach consisting of a dual, concentric exposure array of vertical risers or vent pipes. The purpose of the outer array (17 m diameter) is to vent ambient air outward and toward the incoming wind, thus providing an air curtain to reduce the velocity of that incoming wind to simulate the mode or the most frequently occurring wind speed at the study site. The inner array (15 m diameter) vents the required elevated levels of trace gases (CO(2), O(3), etc.) into the study plot. This dual array system is designed to provide spatial homogeneity (shown through diffusion modeling) of the desired trace-gas levels within the study plot and to also reduce its consumption. As an example, while in the single-array free-air CO(2)-release system the consumption of CO(2) to elevate its ambient concentration by 400 ppm is calculated to be about 980 tons/year/replicate, it is estimated that in the dual array system it would be approximately 590 tons/year/replicate. Thus, the dual array system may provide substantial cost savings (24000-39000 dollars/year/replicate) in the CO(2) consumption (60-100 dollars/ton of CO(2)) alone. Similarly, benefits in the requirements of other trace gases (O(3), NO(x), etc.) are expected, in future multivariate studies on global climate change.     

Field monitoring of volatile organic compounds using passive air samplers in an industrial city in Japan

Highly portable, sensitive, and selective passive air samplers were used to investigate ambient volatile organic compound (VOC) levels at multiple sampling sites in an industrial city, Fuji, Japan. We determined the spatial distributions of 27 species of VOCs in three campaigns: Mar (cold season), May (warm season), and Nov (mild season) of 2004. In all campaigns, toluene (geometric mean concentration, 14.0microg/m3) was the most abundant VOC, followed by acetaldehyde (4.76microg/m3), and formaldehyde (2.58microg/m3). The spatial distributions for certain VOCs showed characteristic patterns: high concentrations of benzene and formaldehyde were typically found along major roads, whereas high concentrations of toluene and tetrachloroethylene (PCE) were usually found near factories. The spatial distribution of PCE observed was extremely consistent with the diffusion pattern calculated from Pollutant Release and Transfer Register data and meteorological data, indicated that passive air samplers are useful for determining the sources and distributions of ambient VOCs.     

Simulation of the tropospheric distribution of carbon monoxide during the 1984 MAPS experiment

A global, three-dimensional tropospheric chemistry model was used to perform simulations of the tropospheric distribution of carbon monoxide (CO) coinciding with NASA's Measurement of Air Pollution from Satellites (MAPS) experiment which took place during 5–13 October 1984. Archived meteorological data for September and October, 1984, were obtained from the European Centre for Medium-Range Weather Forecasting and used to drive the offline chemical transport model simulations. Base-case CO emissions were generated by applying emission factors to compiled inventories for related or co-emitted trace species. Simulation results from September and October have been compared with a recent re-release of the 1984 MAPS data and with in situ correlative data taken during the MAPS mission. Because of unrealistically large spatial variability in N₂O mixing ratios measured concurrently by MAPS, model results were also compared with an adjusted CO data set generated by assuming that errors in N₂O measured mixing ratios were correlated with errors in the MAPS CO data. These comparisons, in conjunction with simulations probing model sensitivities, led to the conclusion that biomass burning CO emissions from central and southern Africa may have been larger during September and October, 1984, than our initial best estimate based on the CO₂ emissions data of Hao et al. (1990. Fire in the Tropical Biota; Ecosystem Processes and Global Challenges. Springer, Berlin, pp. 440–462; 1994. Global Biogeochemical Cycles 8, 495–503). This result is in disagreement with recent estimates of biomass burning emissions from Africa (Scholes et al., 1996, Journal of Geophysical Research 101, 23677–23682) which are smaller than previously thought for emissions from this region. Although unknown model deficiencies cannot be conclusively ruled out, model sensitivity studies indicate that increased CO emissions from central and southern Africa offer the best explanation for reducing observed differences between model results and MAPS data for this time period. Our results, in combination with a disparity in recent CO emission estimates from this region (Scholes et al., 1996; Hao et al., 1996, Journal of Geophysical Research 101, 23577–23584), and in light of recent indications of highly variable biomass burning activities from the tropical western Pacific (Folkins et al., 1997, Journal of Geophysical Research 102, 13291–13299), seem to suggest that biomass burning emissions exhibit significant year-to-year variability. This large variability of emissions sources makes the accurate simulation of specific time periods very difficult and suggests that biomass burning trace species inventories may have to be developed specifically for each simulated time period, employing satellite-derived information on fire coverage and flame intensity.     

Factors influencing the ground level distribution of ozone in Europe

Ozone is a widely distributed pollutant in the atmospheric boundary layer over north west Europe. Three main sources have been identified: the stratosphere, the free troposphere and boundary layer photochemical production. The pattern of ground level ozone concentrations resulting from these three sources cannot be accurately specified. Ozone shows significant variations in space and time but because of the high cost of continuous monitoring equipment, spatial variations on a national and international basis have not been studied in detail. Variations in ozone concentrations at individual monitoring sites have been given a great deal of attention and experience gained from United Kingdom monitoring sites is described in some detail. The averaging time statistical model of Larsen is employed to relate the exposure levels measured over different averaging periods. Diurnal variations have a major influence on exposure levels at sites nominally exposed to the same regional ozone distribution. The physical and chemical mechanisms which give rise to diurnal variations are detailed so that sites can be screened for different diurnal behaviour characteristics.     

Effects of instrument precision and spatial variability on the assessment of the temporal variation of ambient air pollution in Atlanta, Georgia

Data from the U.S. Environmental Protection Agency Air Quality System, the Southeastern Aerosol Research and Characterization database, and the Assessment of Spatial Aerosol Composition in Atlanta database for 1999 through 2002 have been used to characterize error associated with instrument precision and spatial variability on the assessment of the temporal variation of ambient air pollution in Atlanta, GA. These data are being used in time series epidemiologic studies in which associations of acute respiratory and cardiovascular health outcomes and daily ambient air pollutant levels are assessed. Modified semivariograms are used to quantify the effects of instrument precision and spatial variability on the assessment of daily metrics of ambient gaseous pollutants (SO2, CO, NOx, and O3) and fine particulate matter ([PM2.5] PM2.5 mass, sulfate, nitrate, ammonium, elemental carbon [EC], and organic carbon [OC]). Variation because of instrument imprecision represented 7-40% of the temporal variation in the daily pollutant measures and was largest for the PM2.5 EC and OC. Spatial variability was greatest for primary pollutants (SO2, CO, NOx, and EC). Population-weighted variation in daily ambient air pollutant levels because of both instrument imprecision and spatial variability ranged from 20% of the temporal variation for O3 to 70% of the temporal variation for SO2 and EC. Wind     

Carbon Monoxide and Lead-An Environmental Appraisal

Because of the common source, lead and CO values in the atmosphere tend to behave in a similar manner. Thus, diurnal variations in these two pollutants show a pattern related to motor vehicle traffic flow. Also, the exposure to both vary by orders of magnitude with the highest being on the road (in the car) thus setting up special dosage situations. Community sources seem to affect background level at least based on fall-off with distance. There may be a relatively wider exposure of the general population to lead and CO. While the lead levels may not be increasing in the downtown portion of the central city proper, typical central city levels of several years ago may be more diffuse and spread out, thus occurring over increasingly large portions of the community. Similarly, there may be a wider exposure of the population to CO as the levels become more nearly uniformly high over a larger area. In addition, there may be problems of a shorter term exposure to high levels of CO in commuter traffic. This may be of consequence to selected types of drivers or passengers. Finally, it should also be noted that during air pollution episodes, CO levels appear to rise with no data currently available on changes in concomitant ambient lead levels.     

Using a chemistry transport model to account for the spatial variability of exposure concentrations in epidemiologic air pollution studies

Environmental epidemiology and more specifically time-series analysis have traditionally used area-averaged pollutant concentrations measured at central monitors as exposure surrogates to associate health outcomes with air pollution. However, spatial aggregation has been shown to contribute to the overall bias in the estimation of the exposure-response functions. This paper presents the benefit of adding features of the spatial variability of exposure by using concentration fields modeled with a chemistry transport model instead of monitor data and accounting for human activity patterns. On the basis of county-level census data for the city of Paris, France, and a Monte Carlo simulation, a simple activity model was developed accounting for the temporal variability between working and evening hours as well as during transit. By combining activity data with modeled concentrations, the downtown, suburban, and rural spatial patterns in exposure to nitrogen dioxide, ozone, and PM2.5 (particulate matter [PM] < or = 10 microm in aerodynamic diameter) were captured and parametrized. Exposures predicted with this model were used in a time-series study of the short-term effect of air pollution on total nonaccidental mortality for the 4-yr period from 2001 to 2004. It was shown that the time series of the exposure surrogates developed here are less correlated across co-pollutants than in the case of the area-averaged monitor data. This led to less biased exposure-response functions when all three co-pollutants were inserted simultaneously in the same regression model. This finding yields insight into pollutant-specific health effects that are otherwise masked by the high correlation among co-pollutants.