em • feature by Panos Georgopoulos, UMDNJ – RWJ Medical School and Environmental and Occupational Health Sciences Institute (EOHSI); Sastry Isukapalli, UMDNJ – RWJ Medical School; Janet Burke, U.S. Environmental Protection Agency National Exposure Research Laboratory; Sergey Napelenok, U.S. Environmental Protection Agency Atmospheric Modeling Division; Ted Palma and John Langstaff, U.S. Environmental Protection Agency Office of Air Quality Planning and Standards; Mohammed Majeed, State of Delaware Department of Natural Resources and Environmental Control; Shan He, New Jersey Department of Environmental Protection Division of Air Quality; Daewon Byun and Mark Cohen, National Oceanic and Atmospheric Administration; and Robert Vautard, Laboratory of Climate and Environment Sciences (CEA/CNRS/UVSQ). Contact Panos Georgopoulos: panosg@fidelio.rutgers.edu. 26 em october 2009 Air Quality Modeling Needs for Exposure Assessment from the Source-to-Outcome Perspective Humans are exposed continuously to mixtures of air pollutants. The compositions of these mixtures vary with time and location and their components originate from many types of sources, both local and distant, including industrial facilities, vehicles, consumer products, and more. Exposure characterization is often the weakest link in the “source-to-outcome” sequence of processes and events affecting human and ecological health risks from environmental pollutants (see Figure 1). It is recognized1 that it is generally easier to characterize exposures for ecosystems than for human populations, as in the latter case exposures can be particularly sensitive to high-resolution spatial and temporal variations in ambient concentrations and the “micoenvironmental” adjustments imposed by a variety of indoor and outdoor settings (occupational, residential, recreational, commuting, etc.). Ultimately, quantifying inhalation exposures of humans to atmospheric contaminants, such as criteria air pollutants and air toxics, requires characterization of the air flow that enters the human respiratory tract, i.e. “personal air”. Assessing personal air concentrations, in turn, requires characterization of concentrations in residential and occupational microenvironments as well as at local neighborhood ambient scales. The constituents of the local outdoor air may originate from a variety of distances, from regional to continental and beyond. Exposure modeling, therefore, involves processes spanning a wide range of spatial and temporal scales. The last decade has seen an evolution in the practice of exposure assessment, with the focus changing from considering potential exposures to a single pollutant that would occur outdoors at a given location or across an area of concern, to “person oriented” multi-pollutant exposure assessments. Current assessments take into account the behavioral and physiological dynamics of contact with various contaminants, as individuals (actual or “virtual” in the case of computer simulations) move in different indoor and outdoor “microenvironments” while engaged in activities that determine rates of contact and uptake of multiple pollutants. To characterize inhalation intakes, airborne concentrations of co-occurring pollutants have to be determined for each individual at the spatial and temporal scales defined by the microenvironments and exposure activity events. This progress was made possible by the availability of enhanced computational modeling resources, widespread GIS applications, new databases on human activities, demographics, microenvironmental attributes, Copyright 2009 Air & Waste M anagement Association awma.org Figure 1. An overview of the sequence of processes and spatial scales involved in the source-to-dose-to-outcome sequence for human health risks due to inhalation exposures. emissions, etc.2 However, the use of more comprehensive realistic frameworks for exposure assessment raises additional needs for air quality models. These needs are discussed here, following a brief summary of the current status of inhalation exposure models. It should be pointed out that this article focuses on modeling needs related to inhalation exposures of airborne contaminants and not on multi-pathway exposures from contaminants in multiple media. Information Needs for Exposure Assessment Exposure is defined as the “contact of a biological receptor with the contaminants of concern,” resulting in the intake and subsequent (systemic) uptake of these contaminants. Information necessary for calculating health-relevant metrics of exposure includes time-course profiles of concentration levels, frequency and duration of the contact. The environmental (physicochemical) and biological properties of the contaminants determine the relevant time scales and corresponding temporal and spatial averaging and sampling practices relevant to assessments of exposure and associated health risks. Potential health effects associated with the contaminants of concern determine the types of exposures that need to be considered (acute, subchronic, chronic) and subsequently the types of environmental, behavioral, and demographic information that needs to be collected. Spatial and temporal scales and resolutions can differ widely in exposure assessments, depending on the situation of concern. In general, the resolutions of the analysis should be consistent with the exposure events, awma.org while the time scale should be relevant to the health effects of concern. In situations involving acute inhalation exposures, such as cases of accidental releases of chemicals, the criteria for limiting potential outdoor contacts are provided through the Acute Exposure Guideline Levels (AEGLs), which combine concentration levels and contact durations to relate exposures to different harmful effects.3 The National Advisory Committee for AEGLs has been developing these guidelines to help both federal and local authorities, as well as private companies, deal with emergencies involving spills, or other accidental releases of hazardous chemicals. The AEGL values are intended to serve for alert and emergency response planning as well as for disaster control. AEGL values represent toxicologically relevant “ceiling” exposure levels for different exposure periods (10 min, 30 min, 1 hr, 4 hr, 8 hr) and for three different degrees of severity of toxic effects: a threshold for notable discomfort (AEGL-1); a threshold for serious, longlasting effects or an impaired ability to escape the area of exposure (AEGL-2); and a threshold for lethal effects (AEGL-3). In situations involving sub-chronic and chronic inhalation exposures it was common practice—up to the recent past—to use ambient levels of concentrations of criteria pollutants and air toxics as inputs to epidemiological studies addressing questions on human health risks. Specifically, outdoor concentrations, either from ambient central fixed monitors or from numerical simulation models have been used as surrogates for personal exposure. Copyright 2009 Air & Waste M anagement Association october 2009 em 27 However, a “key problem in using modeled or monitored ambient concentration data to estimate exposures is the fact that people in most societies spend most of their time indoors”4. The fact that actual human exposures are dominated by personal activities and the attributes of microenvironments is now gaining universal recognition (see Figure 2). Figure 3 further illustrates this point by presenting the distributions of simultaneous measurements of background, local outdoor, indoor, and personal air concentrations for three pollutants with different physicochemical properties: benzene, formaldehyde, and fine particulate matter (PM2.5). The pollutants were collected from over 300 homes of non-smokers in three U.S. cities representing different climatological and demographic settings (Elizabeth, NJ, Houston, TX, and Los Angeles, CA). It is clear that personal exposures cannot be adequately represented by ambient concentration levels. Figure 4 presents predicted concentrations vis a vis inhaled doses of PM2.5 in Philadelphia, PA and provides an example of a modeling analysis consistent with this fact. It should be noted that exposure science is a rapidly evolving field and the development of a standard and commonly accepted terminology is an ongoing process. Very often, procedures that are called exposure modeling, exposure estimation, or exposure assessment may in fact only refer to a sub-set of the components required for exposure characterization, such as modeling local dispersion patterns of a contaminant. Though they are not complete exposure studies per se, such efforts have value, as they potentially improve individual components of a comprehensive assessment. Nevertheless, in the following, the terms “exposure model” or “modeling” will refer specifically to formulations that explicitly describe contact with contaminants by considering (a) microenvironmental attributes such as concentration levels; (b) behavioral attributes such as activities of individuals in a given microenvironment; and (c) biological attributes such as gender, age, weight, body mass index, etc. Evolution and Current Practice of Exposure Modeling Existing comprehensive inhalation exposure models consider the movement of individual human subjects (actual or virtual), or of appropriately defined cohorts, in space and time, as sequences of activity or exposure events. In these sequences, each event is defined by time, a geographic location, a microenvironment, and the activity of the subject. The U.S. Environmental Protection Agency (EPA) has supported comprehensive efforts in developing models implementing this general concept, and these efforts have resulted in the National Exposure Model and Probabilistic National Exposure Model5 (NEM/pNEM), Hazardous Air Pollutant Exposure Model6 (HAPEM), Simulation of Human Exposure and Dose System7 (SHEDS), Air Pollutants Exposure Model8 (APEX), and Modeling Environment for Total Risk studies9 (MENTOR) families of models. Table 1 summarizes essential attributes of these inhalation exposure models that currently also represent the state-of-the-art in the field. The overall approach followed in these models in general consists of the following components: 1. Estimation of the background or ambient levels of the pollutants of concern via (a) spatio-temporal Figure 2. For most people the majority of exposures to airborne contaminants takes place through contact and inhalation of chemicals in indoor (residential or occupational) microenvironments. The air in these microenvironments contains a complex mixture of contaminants including those entrained from outdoor (ambient) air, those emitted indoors, and those formed via chemical transformations in indoor air (e.g. ultrafine particles formed from the interaction of entrained ozone with emissions from household air fresheners and solvents). 28 em october 2009 Copyright 2009 Air & Waste M anagement Association awma.org analysis of fixed monitor data; (b) emissionsbased air quality modeling applied in a coarse resolution mode at the regional-to-urban scales; or (c) a combination of model output and observations. 2. Estimation of local outdoor levels of the pollutants of concern. These levels would typically characterize the ambient air of either an administrative unit (such as a census tract) or a conveniently defined neighborhood. This component can also involve: (a) spatio-temporal statistical analysis of local monitor data; (b) application of gridbased air quality models at their highest resolution (typically around 2-4 km); (c) application of “sub-grid” scale modeling employing local-scale dispersion models; or (d) “refinement” of the estimates of a regional model using schemes that take into account subgrid variation in topography and land-use combined with considerations of subgrid transformation and mixing processes. 3. Characterization of relevant attributes of the individuals or populations under study include residence and work locations, occupation, housing data, income, education, age, gender, race, weight, body mass index, and other physiological characteristics. These factors may affect physical activities and intake rates and subsequent systemic uptakes of chemicals. This component of the exposure analysis can be pursued by (a) assembling the required information for each actual person to be considered in the study, (b) selecting a fixed-size sample population of “virtual individuals” in a way that statistically reproduces essential demographics of the administrative unit used in the assessment or (c) dividing the population-of-interest into a set of cohorts representing selected subpopulations and defining one or more representative individuals for each cohort. 4. Development of activity event or exposure event sequences for each member of the sample population, or for each cohort, using study-specific information or sampling from available timeactivity databases (such as EPA’s Consolidated Human Activity Database10) with appropriate matching criteria for the virtual subjects. 5. Estimation of levels and temporal profiles of the pollutants in various specific outdoor and indoor microenvironments such as street canyons, roadway intersections, gas stations, school yards, parks, residences, offices, schools, restaurants, vehicles, etc. This is typically done through either awma.org (a) statistical analyses of observational datasets (typically for indoor/outdoor relationships or for vehicle/outdoor relationships) that produce microenvironmental factors; (b) simple microenvironmental mass balance models with or without atmospheric chemistry; (c) simplified microenvironment-specific dispersion models, such as street canyon models; or (d) detailed Computational Fluid Dynamics (CFD) models, possibly combined with chemistry, for indoor and outdoor microenvironments. 6. Calculation of appropriate inhalation rates for the members of the sample population, combining the physiological attributes of the actual or virtual study subjects and the activities pursued during the individual exposure events. 7. Calculation of target tissue dose through respiratory dosimetry modeling, if sufficient information is available. Figure 3. Background, local outdoor, indoor, and personal concentration levels of three common air pollutants across diverse geographical areas. Shown are distributions of 48hour integrated indoor, local outdoor, background outdoor (concentrations from the nearest ambient air quality monitor) and personal air samples of target compounds collected simultaneously from approximately 100 homes of non-smokers (and with no attached garages), each in Elizabeth, NJ, Houston, TX, and Los Angeles, CA between 1999 and 2001 (data from 46). The three contaminants shown are benzene (representing a non-reactive gas), formaldehyde (representing a highly reactive gas that is both emitted and formed through atmospheric photochemistry) and PM2.5. The above components can be implemented in a nested manner in order to characterize both uncertainty and variability (intra-individual and inter-individual) involved in the exposure and dose Copyright 2009 Air & Waste M anagement Association october 2009 em 29 Table 1. Comparison of typical features/attributes of comprehensive inhalation exposure models pNEM HAPEM REHEX APEX SHEDS MENTOR Typical Spatial Extent/Resolution Urban areas; Census tract Urban/National; Census tract Urban areas; Census tract Urban areas; Census tract Urban areas; Census tract Urban areas; Census tract Typical Temporal Extent/Resolution Year: Hour Season; 3 Hours Season; Hour Weeks to Year; Hour/Minutes Weeks to Year; Hour/Minutes Weeks to Year; Hour/Minutes Population Activity Pattern Simulation Top-down analysis Top-down analysis Top-down analysis Bottom-up analysis Bottom-up analysis Bottom-up “person-oriented” Estimation of Concentrations in Microenvironments (µE’s) Mass-balance Empirical indoor/ outdoor ratios Mass balance (steady-state); dynamically generated µE’s Mass balance; empirical ratios; fixed µE’s Mass balance (steady-state); empirical ratios; dynamically generated µE’s Mass balance; empirical ratios; indoor chemistry; dynamically generated µE’s Indoor Source Emis- Limited sources; sions & Micro-envi- Probability ronmental factors distributions Additive terms for sources; Probability distributions Additive terms for sources; Probability distributions User-defined sources; Probability distributions Limited sources; Probability distributions User-defined sources; Probability distributions Uptake and Doses NA NA Potential doses and uptake Potential doses and uptake Physiologically based doses Exposure Character- Population distribuization; Temporal tions; Hourly Resolution Limited population distributions; Annual Population distributions; Hourly Population distributions; Hourly/ Minutes Population distributions; Hourly/ Minutes Population distributions; Hourly/ Minutes Notes Has been employed Proprietary in developing NATA implementation estimates Follows “One Atmosphere” approach for mixtures of chemicals Different implementations for specific groups of chemicals Follows “One Atmosphere” approach for mixtures of chemicals Potential doses Cohort-based model; precursor to APEX assessment. In order to characterize the uncertainty in estimates of distributions of exposures, these calculations can be performed multiple times by sampling from distributions of corresponding inputs and parameters representing the uncertainties. In practice, the majority of past exposure modeling studies have either incorporated only subsets of these components, or treated some of them in a simplified manner, often focusing on selected factors affecting exposure. Of course, depending on the objective of a particular modeling study, implementation of a selected subset may in fact be adequate. For example, outdoor levels of pollutants, in conjunction with basic demographic information such as residential and occupational locations and commuting patterns of individuals, can be used to calculate metrics of potential population exposures associated with ambient air. Such metrics can be useful in comparing alternative scenarios related to different meteorology, emissions, etc. Though derived metrics would not be quantitative indicators of total human exposures, they can serve as surrogates of population exposures associated with outdoor air, and thus aid in regulatory decision making concerning pollutant standards and in studying the efficacy of emission control strategies. This approach has been used in comparative evaluations of regional and local emissions reduction strategies in 30 em october 2009 the United States.11 Air quality modeling plays a critical role in the steps associated with characterizing background, neighborhood, and microenvironmental concentration levels. Table 2 summarizes essential attributes of widely used air quality models in relation to different types of exposure characterizations. The next three sections identify specific air quality modeling needs from an exposure assessment perspective for regional, urban, local and microenvironmental scales. Air Quality Modeling Needs: Regional and Urban Scales Model/Data Assimilation Ambient (outdoor) concentrations of pollutants over a regional domain may be estimated either through emissions-based mechanistic modeling, ambient-data-based modeling, or a combination of both. Monitored ambient air pollutant levels generally reflect point measurements of atmospheric concentrations at individual locations near the ground. Air quality models provide spatially-resolved descriptions of pollutant concentrations across a geographical domain, but the model outputs reflect volume-average concentrations at each grid cell and can be inaccurate due to inadequate physics/chemistry of the model and errors in input data. In practice, the availability and accuracy of Copyright 2009 Air & Waste M anagement Association awma.org emissions inventories are often the limiting factors in applying mechanistic atmospheric fate and transport models on regional levels. In addition, for pollutants capable of long-range transport, either a global simulation must be done or boundary conditions must be supplied to account for the effects of sources outside of the region. Depending on the pollutant of interest, uncertainties in atmospheric chemistry, phase partitioning, and/or deposition phenomena can also limit the accuracy of regional/ urban models. Ambient-data-based models typically calculate spatial or spatiotemporal distributions of the pollutant through the use of interpolation schemes, which include various Bayesian approaches. These schemes employ deterministic or stochastic models for allocating monitor observations to the nodes of a virtual regular grid covering the region of interest. Geostatistical techniques such as kriging provide various procedures for generating an interpolated spatial distribution for a given time, from data at a set of discrete points. However, kriging techniques are severely limited by the non-stationary character of the atmosphere and the resulting concentration patterns of air pollutants and this lack of stationarity requires semivariograms that change with time.12 Alternative approaches that interpolate monitor data simultaneously in space and time should be more applicable to air quality assessments.13, 14 Fusion of model outputs and ambient measurements can be performed to improve the estimates of the spatial distribution of pollutant concentrations.15 Such approaches include data assimilation methods16 or simpler hybridization approaches, using a method such as kriging of the model error.17 Expansion of the One-Atmosphere Approach As human (and ecological) exposure studies further recognize the need for multipollutant assessments that take into account synergistic effects of cooccurring contaminants, air quality models need to provide information on an increasing number of air pollutants, accounting for their emissions, transport, and fate, as well as their patterns of co-occurrence and interactions (i.e. a “one-atmosphere” approach). Various steps have taken place towards this objective, including the incorporation of new versions of atmospheric chemistry mechanisms in models such as CMAQ,18 EPA's national air quality model, to consider a larger number of organic compounds in addition to those affecting the levels of criteria air pollutants. Furthermore, as biogenic emissions of allergens such as pollen, fungal spores, etc., are associated with similar endpoints (such as asthma) as photochemical oxidants,19 it is necessary to include them in future versions of regional air quality models. Table 2. Attributes of various air quality models, including Eulerian grid-based and Lagrangian/Gaussian dispersion models AERMOD HYSPLIT Calpuff HYPACT HPAC CAMx CMAQ Developer AMS/USEPA NOAA-ARL EarthTech Alliant Inc DTRA Environ USEPA Model Type and Solution Scheme Gaussian plume; Analytical Lagrangian puff and particle dispersion; Eulerian option; Discrete Lagrangian puff dispersion; Discrete Lagrangian particle dispersion; Eulerian option; Discrete Lagrangian puff dispersion; Discrete Eulerian grid; 3D Finite Difference Eulerian grid; 3D Finite Difference Chemical Species/ Contaminants Individual chemicals (nonreactive) Individual chemicals or specific mixtures of chemicals† Individual chemicals† Individual chemicals Individual chemicals Mixtures of chemicals‡ Mixtures of chemicals‡ Chemical Interactions Linear decay of single chemical Optional chemistry modules† Optional chemistry modules† Linear chemical transformations Linear decay Complex mechanisms (CB4, RADM, SAPRC) Complex mechanisms (CB4, RADM, SAPRC) Typical inhalation exposure scenarios where model is most applicable Hourly to longterm assessments; local to urban scales Short- to longterm assessments; urban to regional scales; regional emergency response Hourly to longterm assessments; urban to regional scales Short-term highresolution assessments; local to regional scales; computationally demanding Short-term highresolution assessments; local to urban scales; local emergency response Episodic (few days) to annual assessments for multiple pollutants; urban to regional scales Episodic (few days) to annual assessments for multiple pollutants; urban to regional scales †Custom modules exist for specific groups of chemicals (e.g. mercury, dioxins, sulfate and photochemistry in HYSPLIT; and photochemistry in Calpuff). ‡CMAQ and CAMx consider complex photochemistry for multiple contaminants simultaneously. Terminology (a) for spatial scales: local (0-5 km), urban (5 km to 50 km), and regional (> 50 km); (b) for time scales: short-term (minutes to days), long-term (annual to multi-year). Acronyms: CB4, Carbon Bond IV; RADM, Regional Acid Deposition Model; SAPRC: Statewide Air Pollution Research Center. awma.org Copyright 2009 Air & Waste M anagement Association october 2009 em 31 Figure 4. (a) 24-h averaged local outdoor concentrations on July 19, 1999 for 482 urban Philadelphia, PA and Camden, NJ census tracts, derived from hourly PM2.5 predictions of the Community Multiscale Air Quality (CMAQ) model that were “downscaled” at census tract level, using the Bayesian Maximum Entropy (BME) method; (b) Corresponding 95th percentiles, calculated for each census tract, of 24-h aggregated total inhalation doses from both outdoor and indoor sources. The color scheme shows quantiles of the concentrations and dose distributions (source: 2 reproduced with kind permission from Springer Science+Business Media). 32 em october 2009 Expansion of Multimedia Linkages of Air Quality Models The role of air quality modeling has been expanding to support combined assessment of ecological and human health risks. A representative example is the application of models such as CMAQ (a gridbased photochemical air quality model (PAQM)) and HYSPLIT to study regional-scale transport of mercury and its contribution to watershed loading.20, 21 An example of a tool linking CMAQ with a watershed model is provided in Schwede et al.22 Linkages may be dynamic (on-line) or static (offline) depending on the pollutant, the phenomena of interest, or the different media involved. The relevant spatial and temporal scales in different media can vary dramatically and the linking of models must consider those differences. The linkage of models for various media is in fact an emerging science. Lammel et al.23 compared the predictions of multimedia models of different levels of detail for a set of non-ionic and moderately polar organic chemicals and found that simpler mass balance models tend to overestimate substance sinks in air and to underestimate atmospheric transport velocity due to neglect of spatiotemporal variability. The need for spatially resolved models is especially important for semivolatile chemicals that have intermediate lifetimes and are therefore distributed regionally rather than locally or globally while cycling among various environmental media. efficient characterizations of source–receptor relationships. Several techniques exist for developing source attribution for modeled concentrations and deposition, including approaches based on linear superposition,20, 24 source elimination,25 and emissions tagging.26 The suitability, efficiency and intercomparability of different approaches depend on the pollutant involved. Furthermore, several methods for “inverse modeling” and diagnostic sensitivity analysis can be applied to regional air quality models; these include direct techniques such as the Direct Decoupled Method (DDM),27 model adjoints,28 and surrogate techniques such as the High Dimensional Model Representation (HDMR).29 These provide alternative means for assessing model response without requiring simulations for each combination of different variables. The DDM requires the addition of equations for sensitivity calculations to the original model source code, and calculates local derivatives with respect to perturbations in input parameters such as emissions, chemical reaction rates, and initial/boundary conditions. Similarly, model adjoints propagate perturbations, but for receptor-based metrics and backward in time. HDMR techniques are applied without alteration to the computational code model and rely on a global response of the model to changes in inputs. Implementation of Diagnostic Tools In an accountability framework, assessment and management of exposures should be ultimately related to source activities, whether the sources are proximal or remote. Regional/urban air quality models employed in this context can be enhanced by diagnostic methods that allow computationally Local (neighborhood or subgrid) spatial variability is a major issue with respect to characterizing local concentrations of contaminants in most exposurerelevant settings. For example, the fast rates of reactions in photochemical systems result in significant concentration gradients in the vicinity of sources (e.g. titration of ozone by NOx emissions Air Quality Modeling Needs: Local and Neighborhood Scales Copyright 2009 Air & Waste M anagement Association awma.org in the immediate vicinity of roadways). These gradients are not resolved directly by currently available grid-based PAQMs such as CMAQ and CAMx. “Plume-in-grid” options have been developed for both these models and they can be used for large point sources (such as smokestacks) that exist within a grid cell. Nevertheless, plume-in-grid formulations will mostly resolve gradients in upper atmospheric layers and thus are not necessarily relevant to human exposure calculations, which are affected by gradients caused by a multiplicity of smaller ground-level or near-ground-level combustion sources. Currently PAQMs are typically applied with horizontal resolutions ranging from 2 km to 36 km and a surface layer thickness that is of the order of 30 m. Though, computationally, it is possible to further increase the resolution of these modeling grids, there are theoretical limits imposed by assumptions inherent in the formulation of governing equations for transport and transformation in these models; therefore, a resolution of the order of 2 km is the highest allowable by current gridbased PAQMs.30 Application of PAQMs to urban domains is further complicated by urban topography, the urban heat island, and other factors. It is beyond the scope of the present discussion to overview the various issues relevant to urban fluid dynamics and related transport/fate processes of contaminants. Various reviews of these issues and of available approaches for modeling urban fluid mechanics and dispersion are available.31,32 One way of accounting for subgrid concentration gradients is the hybrid modeling approach,33, 34 where concentrations from a grid-based PAQM (such as CMAQ) and a local plume dispersion model (such as AERMOD) are added to provide total contributions from regional transport/chemistry and from local-scale dispersion. In such an approach, it is important to avoid double counting sources in the two models.35 Furthermore, since local dispersion models such as AERMOD cannot account for photochemical transformation of contaminants, the hybrid approach is expected to be more appropriate for less reactive (e.g. benzene) rather than highly reactive (e.g. formaldehyde) components of the urban air pollution mix. Air Quality Modeling Needs: Microenvironmental Scale In the context of modeling, exposure occurs in “microenvironments”, or the outdoor, indoor, or in-vehicle locations where individuals spend their awma.org time, where the ambient atmospheric concentration value (monitored or modeled) is modified by the immediate surroundings of the individual person/ receptor. General categories of microenvironments include residential indoor, occupational indoor, public indoor, vehicular, and outdoor. Characterizing microenvironments can involve modeling of various local sources and sinks and transport/fate processes, and interrelationships between ambient and microenvironmental concentration levels. Three general approaches have been used in the past to model microenvironmental concentrations: empirical (typically linear regression) fitting of data; parameterized mass balance modeling; and detailed Computational Fluid Dynamics (CFD) modeling. Characterization of Outdoor Microenvironments Empirical regression analyses have been used in various studies to relate specific outdoor locales defined by land-use to spatial variability of pollutant concentrations. For example, some studies36 use regression analysis to assess the associations between airborne concentrations and land-use variables such as the area of open space, traffic count and distance from nearest highway, the length of highways and roads within a certain radius, population density, and industrial land–use. However, various arguments have been made regarding the value of using landuse regression modeling to assign exposure classifications in large-scale epidemiologic studies. Parameterized mass balance models for outdoor microenvironments include various local roadway, intersection, and street canyon models.37 Nearhighway pollutant dispersion models consider vehicle wake parameterizations derived from canopy flow theory and wind tunnel measurements to adjust the atmospheric velocity and turbulence fields. Typically, in parameterized street canyon models, concentrations of exhaust gases are calculated using a combination of a plume model for the direct contribution and a box model for the recirculating part of the pollutants in the street. Parameterization of flow and dispersion conditions in these models is usually deduced from analysis of experimental data and model tests that considered different street configurations and various meteorological conditions. Various CFD-based street canyon models have also been developed in recent years (see the series of International Conferences on Harmonization- Copyright 2009 Air & Waste M anagement Association october 2009 em 33 www.harmo.org), employing various alternatives for local closure of the turbulent transport equations. Reviews and intercomparisons of such models vis-a-vis field data are available.38 Characterization of Indoor Microenvironments Numerous indoor air quality modeling studies have been reported in the literature; however, depending on the modeling scenario, only a few of them address physical and chemical processes that affect complex air pollution mixtures (e.g. photochemical oxidants) indoors39-42. It is beyond the scope of this discussion to review in detail the current status of indoor air modeling. Various studies have compared the different formulations of zonal (multicompartmental) models and of more complex, CFD, models.43 Some indoor air models have considered atmospheric chemistry, which can be especially important in the presence of indoor sources such as gas stoves,44 while others considered potential limitations of uniform mixing assumptions.45 Both can be important issues when calculating personal exposures and need to be addressed in conjunction with improving indoor emission inventories. Summary Though existing inhalation exposure modeling systems have evolved considerably in recent years, the air quality models that provide the inputs to these systems can be substantially improved from the perspective of providing exposure-relevant estimates of air quality metrics. Deriving from the discussion in the previous sections, various specific and evolving needs are summarized here: • Ambient photochemical modeling systems are not currently optimized for estimating pollutants at the neighborhood scale. Therefore, practical methods are needed for downscaling regional/ urban modeling estimates to neighborhood and microenvironmental scales, with an emphasis on consistency in linking and coupling models at different scales. • Microenvironmental modeling efforts need to balance mechanistic detail and usability by developing simplified but adequate models that take into account mixing, and local indoor or outdoor chemistry. These models can be developed either directly or as simplifications of detailed CFD methods. For population exposure assessment, there is a need for computationally efficient methods for modeling air quality dynamics in representative realistic distributions of outdoor 34 em october 2009 and indoor microenvironments, in ways that allow aggregation and statistical extrapolations of results across the range of such microenvironments within a regional/urban model cell. These approaches should utilize high-resolution information in urban and suburban topography, combined with detailed microinventories of local emission sources. • Comprehensive air quality models such as CMAQ can be enhanced through incorporation of, or linking with, additional modules for dynamic soil, water, and other compartments. In the future, exposure assessments can be substantially improved through the development and application of comprehensive multimedia models that address local, regional and global scales and can be coupled with multi-pathway human exposure models. • Also, in light of the synergistic effects of cooccuring pollutants, there is a need for expanding the range of airborne contaminants included in the one-atmosphere approach to include allergens and other biological agents.47 Modeling frameworks for exposure assessment in the past have typically focused on individual contaminants and on subsets of their pathways and sources of exposure, potentially neglecting significant contributions from remaining pathways and sources. In recent years, the focus of environmental human and ecological health risk analyses, pursued by both the research community and regulatory agencies, has been gradually shifting from considerations of single to multiple contaminants and pathways. In the future, integrative analyses that link environmental, behavioral, and biological considerations2 will allow increased accountability and more realistic and accurate risk assessments. em In Next Month’s Issue… Advances in Pollution Control Technology EM explores advances in control technology that will help small utility boilers control SO2 emissions, help reduce greenhouse gas emissions, add more effective controls for nitric acid plants, and explain key technologies for mercury emissions from industrial boilers. Also look for… • PM File • Competitive Strategy Copyright 2009 Air & Waste M anagement Association awma.org References 1. NARSTO Technical Challenges of Implementing Multipollutant Air Quality Management under an Accountability Framework, 2009. In Press. 2. Georgopoulos, P. A Multiscale Approach for Assessing the Interactions of Environmental and Biological Systems in a Holistic Health Risk Assessment Framework. Water. Air. Soil Pollut. Focus 2008, 8 (1), 3-21. 3. NRC Standing Operating Procedures for Developing Acute Exposure Guideline Levels for Hazardous Chemicals; Subcommittee on Acute Exposure Guideline Levels, Committee on Toxicology, Board on Environmental Studies and Toxicology, Commission on Life Sciences, National Research Council: Washington, D.C., 2001. 4. WHO Principles of Characterizing and Applying Human Exposure Models; World Health Organization: Geneva, Switzerland, 2005. 5. USEPA Air Quality Criteria for Ozone and Related Photochemical Oxidants (Final), Vol. II; EPA 600/r-05/004bF; U.S. Environmental Protection Agency: Research Triangle Park, NC, 2006. 6. Özkaynak, H.; Palma, T.; Touma, J.; Thurman, J. Modeling Population Exposures to Outdoor Sources of Hazardous Air Pollutants. J. Expos. Sci. Environ. Epidemiol. 2008, 18 (1), 45-58. 7. Burke, J. M.; Zufall, M. J.; Ozkaynak, H., A Population Exposure Model for Particulate Matter: Case Study Results for PM2.5 in Philadelphia, PA. J. Expo. Anal. Environ. Epidemiol. 2001, 11 (6), 470-489. 8. USEPA Total Risk Integrated Methodology (TRIM) Air Pollutants Exposure Model Documentation (TRIM.Expo/APEX, Version 4) - Volume I: User’s Guide; U.S. Environmental Protection Agency: Research Triangle Park, NC, July 2006. 9. Georgopoulos, P. G.; Lioy, P. J., From Theoretical Aspects of Human Exposure and Dose Assessment to computational model implementation: The MOdeling ENvironment for Total Risk Studies (MENTOR). J.Toxicol. Environ. Health B 2006, 9 (6), 457-483. 10. Stallings, C.; Tippett, J. A.; Glen, G.; Smith, L. CHAD User’s Guide - Extracting Human Activity Information from CHAD on the PC; Written for USEPA National Exposure Research Laboratory by ManTech Environmental Technologies: March 22, 2002. 11. Foley, G.; Georgopoulos, P. G.; Lioy, P. J., Examining Accountability for Changes in Population Exposures to 8-Hour Ozone Standard With Implementation of Different Control Strategies. Environ. Sci. Technol. 2003, 37 (21), 302A-309A. 12. Georgopoulos, P. G.; Purushothaman, V.; Chiou, R., Comparative Evaluation of Methods for Estimating Potential Human Exposure to Ozone: Photochemical Modeling and Ambient Monitoring. J. Expo. Anal. Environ. Epidemiol. 1997, 7 (2), 191-215. 13. Georgopoulos, P. G.; Wang, S. W.; Vyas, V. M.; Sun, Q.; Burke, J.; Vedantham, R.; McCurdy, T.; Ozkaynak, H., A Source-to-Dose Assessment of Population Exposures to fine PM and Ozone in Philadelphia, PA, During a Summer 1999 Episode. J. Expo. Anal. Environ. Epidemiol. 2005, 15 (5), 439−457. 14. Christakos, G., Modern Spatiotemporal Geostatistics. Oxford University Press: New York, NY, 2000; p 288. 15. Denby, B.; Garcia, V.; Holland, D.; Hogrefe, C., Integration of Air Quality Modeling and Monitoring Data for Enhanced Health Exposure Assessment. EM, October 2009, 46-49. 16. Elbern, H.; Schmidt, H.; Ebel, A. Variational Data Assimilation for Tropospheric Chemistry Modeling. J. Geophys. Res. Atmos. 1997, 102 (D13), 15967-15985. 17. Blond, N.; Bel, L.; Vautard, R., Three-Dimensional Ozone Data Analysis with an Air Quality Model over the Paris Area. J. Geophys. Res. 2003, 108 (D17), 8564. 18. Sarwar, G.; Luecken, D.; Yarwood, G.; Whitten, G. Z.; Carter, W. P. L., Impact of an Updated Carbon Bond Mechanism on Predictions from the CMAQ Modeling System: Preliminary Assessment. J. Appl. Meteorol. Clim. 2008, 47 (1), 3-14. 19. Gilmour, M. I.; Jaakkola, M. S.; London, S. J.; Nel, A. E.; Rogers, C. A. How Exposure to Environmental Tobacco Smoke, Outdoor Air Pollutants, and Increased Pollen Burdens Influences the Incidence of Asthma. Environ. Health Perspect. 2006, 114 (4), 627-33. 20. Cohen, M.; Artz, R.; Draxler, R.; Miller, P.; Poissant, L.; Niemi, D.; Ratte, D.; Deslauriers, M.; Duval, R.; Laurin, R.; Slotnick, J.; Nettesheim, T.; McDonald, J., Modeling the Atmospheric Transport and Deposition of Mercury to the Great Lakes. Environ. Res. 2004, 95 (3), 247-265. 21. Pongprueksa, P.; Lin, C. J.; Lindberg, S. E.; Jang, C.; Braverman, T.; Bullock, O. R.; Ho, T. C.; Chu, H. W. Scientific Uncertainties in Atmospheric Mercury Models III: Boundary and Initial Conditions, Model Grid Resolution, and Hg(II) Reduction Mechanism. Atmos. Environ. 2008, 42 (8), 1828-1845. 22. Schwede, D. B.; Dennis, R. L.; Bitz, M. A., The Watershed Deposition Tool: A Tool for Incorporating Atmospheric Deposition in Water-Quality Analyses. J. Am. Water Resour. Assoc. 2009, 45, 973-985. 23. Lammel, G.; Klopffer, W.; Semeena, V. S.; Schmidt, E.; Leip, A., Multicompartmental Fate of Persistent Substances - Comparison of Predictions from Multi-Media Box Models and a Multicompartment Chemistry-Atmospheric Transport Model. Environ. Sci. Pollut. Res. 2007, 14 (3), 153-165. 24. Cohen, M. D.; Draxler, R. R.; Artz, R.; Commoner, B.; Bartlett, P.; Cooney, P.; Couchot, K.; Dickar, A.; Eisl, H.; Hill, C.; Quigley, J.; Rosenthal, J. E.; Niemi, D.; Ratte, D.; Deslauriers, M.; Laurin, R.; Mathewson-Brake, L.; McDonald, J. Modeling the Atmospheric Transport and Deposition of PCDD/F to the Great Lakes. Environ. Sci. Technol. 2002, 36 (22), 4831-4845. 25. Seigneur, C.; Vijayaraghavan, K.; Lohman, K.; Karamchandani, P.; Scott, C. Global source attribution for mercury deposition in the United States. Environ. Sci. Technol. 2004, 38, 555-569. 26. Liu, J.; Mauzerall, D. L.; Horowitz, L. W.; Ginoux, P.; Fiore, A. M. Evaluating Inter-Continental Transport of Fine Aerosols: (1) Methodology, Global Aerosol Distribution and Optical Depth. Atmos. Environ. 2009, doi:10.1016/j.atmosenv.2009.03.054. 27. Napelenok, S. L.; Cohan, D. S.; Hu, Y. T.; Russell, A. G. Decoupled Direct 3D Sensitivity Analysis for Particulate Matter (DDM-3D/PM). Atmos. Environ. 2006, 40 (32), 6112-6121. 28. Hakami, A.; Seinfeld, J. H.; Chai, T.; Tang, Y.; Carmichael, G. R.; Sandu, A., Adjoint Sensitivity Analysis of Ozone Nonattainment over the Continental United States. Environ. Sci. Technol. 2006, 40 (12), 3855-3864. 29. Wang, S. W.; Georgopoulos, P. G.; Li, G. Y.; Rabitz, H. Random Sampling-High Dimensional Model Representation (RS-HDMR) with Nonuniformly Distributed Variables: Application to an Integrated Multimedia/Multipathway Exposure and Dose Model for Trichloroethylene. J. Phys. Chem. 2003, 107 (23), 4707-4716. 30. Georgopoulos, P. G.; Seinfeld, J. H. Nonlocal Description of Turbulent Dispersion. Chem. Eng. Sci. 1989, 44 (9), 1995-2016. 31. Fernando, H. J. S.; Lee, S. M.; Anderson, J.; Princevac, M.; Pardyjak, E.; Grossman-Clarke, S. Urban Fluid Mechanics: Air Circulation and Contaninant Dispersion in Cities. Environ. Fluid Mechan. 2001, 1, 107-164. 32. Britter, R. E.; Hanna, S. R. Flow and Dispersion in Urban Areas. Annu. Rev. Fluid Mechan. 2003, 35, 469-496. 33. Isakov, V.; Touma, J. S.; Burke, J.; Lobdell, D. T.; Palma, T.; Rosenbaum, A.; Ozkaynak, H. Combining Regional- and Local-Scale Air Quality Models with Exposure Models for Use in Environmental Health Studies. J. Air Waste Manag. Assoc. 2009, 59 (4), 461-472. 34. Isakov, V.; Irwin, J. S.; Ching, J., Using CMAQ for Exposure Modeling and Characterizing the Subgrid Variability for Exposure Estimates. J. Appl. Meteorol. Climatol. 2007, 46 (9), 1354-1371. 35. Wesson, K. et al. Comparing Models/Methods for Estimating Multi-Pollutant Fine-Scale Air Quality Concentrations. In 30th NATO/SPS International Technical Meeting on Air Pollution Modeling and its Application, San Francisco, CA, 2009. 36. Gilbert, N. L.; Goldberg, M. S.; Beckerman, B.; Brook, J. R.; Jerrett, M., Assessing Spatial Variability of Ambient Nitrogen Dioxide in Montreal, Canada, with a Land-Use Regression Model. J. Air Waste Manag. Assoc. 2005, 55 (8), 1059-1063. 37. Pierce, T. E.; Isakov, V.; Haneke, B.; Paumier, J. Emission and Air Quality Modeling Tools for Near-Roadway Applications; EPA/600/R-09/001 (NTIS PB2009-103941); US Environmental Protection Agency: Washington, D.C., 2008. 38. Vardoulakis, S.; Fisher, B. E. A.; Pericleous, K.; Gonzalez-Flesca, N., Modeling Air Quality in Street Canyons: A Review. Atmos. Environ. 2003, 37 (2), 155-182. 39. Freijer, J. I.; Bloemen, H. J. T. Modeling Relationships Between Indoor and Outdoor Air Quality. J. Air Waste Manag. Assoc. 2000, 50 (2), 292-300. 40. Hayes, S. R. Estimating the Effect of Being Indoors on Total Personal Exposure to Outdoor Air Pollution. JAPCA 1989, 39, 1453-1461. 41. Hayes, S. R. Use of an Indoor Air Quality Model (IAQM) to Estimate Indoor Ozone Levels. J. Air Waste Manag. Assoc. 1991, 41, 161-170. 42. Nazaroff, W. W.; Cass, G. R. Mathematical Modeling of Chemically Reactive Pollutants in Indoor Air. Environ. Sci. Technol. 1986, (20), 924-934. 43. Teshome, E. J.; Haghighat, F., Zonal Models for Indoor Air Flow - A Critical Review. Int. J. Ventilat. 2004, 3 (2), 119-. 44. Georgopoulos, P. G.; Walia, A.; Roy, A.; Lioy, P. J. An Integrated Exposure and Dose Modeling and Analysis System. 1. Formulation and Testing of Microenvironmental and Pharmacokinetic Components. Environ. Sci. Technol. 1997, 31 (1), 17-27. 45. Sorensen, D. N.; Weschler, C. J., Modeling-Gas Phase Reactions in Indoor Environments Using Computational Fluid Dynamics. Atmos. Environ. 2002, 36 (1), 9-18. 46. Weisel, C. P.; Zhang, J. J.; Turpin, B. J.; Morandi, M. T.; Colome, S.; Stock, T. H.; Spektor, D. M. Relationships of Indoor, Outdoor and Personal Air (RIOPA): Part I. Collection Methods and Descriptive Analyses; 7; National Urban Air Toxics Research Center and the Health Effects Institute: November 2005. 47. Knowlton, K.; Rotkin-Ellman, M.; Solomon, G. Sneezing and Wheezing: How Global Warming Could Increase Ragweed Allergies, Air Pollution, and Asthma; Natural Resources Defense Council: 2007. awma.org Copyright 2009 Air & Waste M anagement Association october 2009 em 35