Spreading in New Directions
An overview and update concerning the role of computer modeling in air quality management, especially emergency response and planning
- By Erwin T. Prater
- Apr 01, 2004
Many stakeholders -- including federal and state regulatory agencies, citizen groups, corporations, planning groups, industrial associations and legislative committees -- have been involved in the development of air quality management strategies. In the wake of the terrorist attacks of September 2001, emergency responders and emergency planners have joined this list of stakeholders. Resulting strategies may involve a single facility or many facilities within a region. For example, a multi-facility strategy might require a regional air quality modeling study that examines potential ozone attainment.
Air quality management strategies are intended to minimize adverse health or welfare effects of air pollutants. Air quality standards have been established for many pollutants as goals or limitations on allowed concentrations. The establishment of a regulatory standard often precludes the need to assess the impact on health or welfare directly since the process of developing the standard involves consideration of these impacts. Pollutants for which there are no standards, or are otherwise unregulated, may fall into other categories, such as "nuisance" pollutants (e.g. odors). Assessments of ambient concentration levels of these pollutants are often required in developing a control strategy, although ambient standards may not be established.
Chemical reactions, meteorological transport and diffusion of pollutants in the atmosphere are the fundamental links between emissions and the resulting ambient pollutant concentrations. Air quality management involves some consideration of the physical and chemical processes that occur between the time that a pollutant leaves a source and the time it arrives at a receptor (point) of interest. In practice, models have provisions for quantifying the effects of emission height; source characteristics (e.g. storage temperature, effluent velocity, effluent temperature and phase, stack diameter and orientation); topographic effects; building geometry; and pollutant transport and removal.
Direct measurement of ambient concentrations is one way to establish the relationship between emissions and ambient pollution levels. Within the accuracy of the specific measurement process, it is the best means of determining the aggregate effects of atmospheric processes on pollutants. Direct measurements are often the focal point of intensive scientific measurement field campaigns that study meteorological processes. However, for many compliance issues, direct measurements are infeasible. With respect to emergency management, direct measurements are clearly undesirable. In these instances, the presence of a hazardous chemical is simulated, as is the common practice in emergency response drills. In other cases, the costs of measurements, especially over large regions and long periods of time are cost-prohibitive. However, the hypothetical computation of concentrations at many potential receptors under many meteorological conditions is relatively inexpensive, especially given the recent advances in computing power. For these reasons, the use of dispersion models to assess air quality is widely accepted.
Because our scientific understanding of the physical processes that affect pollutant concentrations is incomplete, there has long been controversy over the accuracy and reliability of predicted pollutant concentrations from air quality models. However, there has also been considerable effort to improve the reliability of atmospheric dispersion models due to the importance of being able to successfully estimate pollutant concentrations and the economic impacts associated with determining suitable emission controls needed to maintain air quality standards. The results of these efforts are expected to improve future models.
Acknowledging the uncertainties in predictions and the various sources of uncertainties in models, the American Meteorological Society (AMS) issued a report in 1981 that commented on the role of modeling in air quality management. 1 The report concluded:
"?to the extent that models reflect our best understanding of the relevant physical processes, they represent a scientifically sound and objective means for taking into account differences of source geometries size and types of pollutant sources and of topographic and meteorological settings in assessing air quality impacts. Thus, models provide a logical and environmentally equitable basis for decision making."
The discussion that follows will focus mainly on aspects of air quality modeling related to regulatory issues in the United States. This article focuses on the roles of dispersion modeling as applied to the Clean Air Act (CAA) of 1970 and the CAA Amendments of 1977 and 1990. Due to interest in domestic preparedness in the wake of the terrorist attacks of September 2001, however, applications of other models to emergency response and planning will also be discussed.
Legal Basis for the Use of Air Quality Modeling
The legal basis for the use of air quality models is codified in Title 40 of the Code of Federal Regulations (CFR) (40 CFR 52.10: Review of New Sources and Modifications; and 40 CFR 52.21: Prevention of Significant Deterioration of Air Quality). Under these regulations, states are required to establish procedures that prevent the construction of pollutant sources that would result in non-compliance with an approved State Implementation Plan (SIP), or would cause, or contribute to, ambient pollutant concentrations that exceed the National Ambient Air Quality Standards (NAAQS). The review procedures for a major modification or a new stationary source must include an air quality analysis that estimates the impact of the source on the ambient air quality. Therefore, a central theme in air quality management is demonstration of compliance with specified ambient air quality concentration values. In practice, this involves a combination of emissions limitations including New Source Performance Standards (NSPS) and National Emissions Standards for Hazardous Air Pollutants (NESHAPs) and limitations on the proximity of sources to each other (e.g., land use) based on modeling techniques.
Applications of Modeling to Air Quality Management
For consistency in regulatory compliance and enforcement, it is desirable that standard models be specified. Model consistency is difficult to achieve in practice, however, since differences in source characteristics, meteorology and topography often suggest that site-specific or specialized models would be more reliable than standard models. Therefore, the selection of a modeling approach to address a regulatory issue is critical. Conservative models that overestimate concentrations may result in overly stringent controls that unnecessarily limit industrial growth and impose unnecessary costs. Conversely, less conservative models that underestimate the impact of sources could result in insufficient control technologies and force costly retrofitting of emission control systems.
A screening model is a conservative model that may simplify the analyses required in compliance demonstrations. Screening models avoid more costly and unnecessary full-scale modeling if the results indicate compliance. They are fast and easy to use, especially with current personal computers (PCs). If a screening model indicates that a source is not in compliance, a more detailed, expensive and refined model is required. The expense arises largely from additional data requirements and the labor involved in setting up, testing and running more complex models. This was especially true in the past when computers were relatively slow and had less memory than current PCs. Geographical information systems (GIS) and Internet distribution has since made refined models much easier to set up and run. This allows analysts to spend more time on air quality management tasks and less time on the "nuts-and-bolts" task of setting up refined models.
Model Input and Output Requirements
As suggested in the proceeding section, enhanced computing technology has greatly improved the modeling process. Two aspects that have been greatly improved are meeting the model input and output requirements, especially with respect to how technology allows input data to be efficiently assembled and output data easily displayed. Improvements in computing technologies have also allowed more complex models to be developed and run times to be significantly reduced.
Air quality models require the following input data: source characteristics, geometry of buildings and other structures that affect wind flow; detailed terrain data; geometry of roads and other linear pollutant sources; inventories of other pollutant sources that contribute to background pollution sources; hourly measurements of wind speed and direction; and atmospheric dispersion characteristics. For continuous emissions modeling, meteorological information is required for time periods ranging from one to five years in length. Until recently, air dispersion modelers had to manually compile most of this information from paper records, photos and maps. For example, terrain data was manually digitized from U.S. Geological Survey topographic maps. This was a painstaking and expensive task with great potential for error. Similarly, the geometry of roads and other linear or area sources was determined by manually analyzing airborne photos, maps and other data sources. This was also a laborious and expensive task. Meteorological data is another input that has been greatly simplified. High-quality, pre-processed (model-ready) meteorological data sets can be readily downloaded over the Internet, decreasing the time required to run an air quality model.
Increased computing power has also improved the data-display capabilities of dispersion models. Two-dimensional mapping capabilities have improved, and PCs have become sufficiently fast to show smooth animation of predicted concentration "clouds." Three-dimensional rendering techniques have also been applied to model output. This is becoming increasingly important for applications of models in emergency response and in homeland security and military applications.
The Future Role of Computer Modeling in Air Quality Management
Computing resources for air quality modeling will continue to improve as PCs become faster and less expensive. The Internet will become an increasingly efficient channel for compiling input data for air dispersion models. Increased computing power will allow more complex air quality models, such as AERMOD2 and CALPUFF3, to be feasibly run. AERMOD is a state-of-the-art model that will eventually replace the current ISC (Industrial Source Complex) models in regulatory compliance demonstrations. AERMOD allows analysts to easily generate reports, and it incorporates GIS display capabilities. AERMOD also works with other programs and allows smooth integration of information on pollutant sources and geometries; particulate and tailpipe emissions from road and vehicle traffic; and motor vehicle emissions characteristics.
CALPUFF is another state-of-the-art model that simulates the effects of time- and space-varying meteorological conditions. CALPUFF can use both three-dimensional meteorological fields from dedicated meteorological models and wind data from a single observation point as input. CALPUFF contains algorithms for building downwash and allows modeling of long-range effects, such as wet and dry pollutant removal and chemical transformations.
Emergency Response and Planning
The terrorist attacks of September 2001 have increased interest in air dispersion models suitable for emergency response and planning. Two applicable models are DEGADIS4 and AFTOX5. DEGADIS is a dense-gas air dispersion model that predicts downwind hydrocarbon flammability levels and toxic gas concentrations. DEGADIS is designed to simulate accidental releases of many pressurized gases, such as chlorine. AFTOX, a model developed by the U.S. Air Force, also predicts downwind concentrations. However, AFTOX is designed specifically for releases that do not involve dense gases.
In these and other emergency-response models, the display of the results is very important since the users are generally experts in areas other than air quality management (e.g., emergency management or public safety). A simple, "clean," graphical user interface (GUI) is critical. Predicted concentrations are often displayed on standard topographic maps.
Some models of interest to emergency response and planning are not traditional air quality models. Three of these models are HEXDAM6 (High Explosive Damage Assessment Model), VEXDAM7 (Vapor Cloud Explosion Damage Assessment Model) and VASDIP8 (Vulnerability Assessment of Structurally Damaging Impulses and Pressures.) HEXDAM evaluates structural damage from explosions based on building geometry and construction materials. HEXDAM also predicts injuries to building occupants, which is important in emergency response planning. VEXDAM evaluates structural damage -- with the exception of explosions originating from vapor clouds -- and predicts injuries, like HEXDAM.
VASDIP complements HEXDAM and VEXDAM by estimating the vulnerability of occupants to blast over-pressure from an explosion.
Although emergency managers typically do not have time to run a model during a response, these models can help with developing response plans "on paper" before an actual emergency. For example, HEXDAM or VEXDAM could show that an explosion would block a critical hallway. Using this information, an emergency manager would not want to use that hallway as an evacuation route after an explosion.
Air dispersion models have become widely accepted in air quality management. A typical air dispersion model incorporates detailed knowledge of pollutant sources and physical mechanisms that govern the resulting pollutant concentrations. Though our scientific understanding of these processes is incomplete, models provide a logical and environmentally equitable basis for decision-making. Advances in inexpensive PCs and "user friendly" displays allow decision-makers to efficiently use information. The same advances also allow air quality analysts to tailor models for specific facilities more efficiently, reducing the cost of producing modeling results.
An increase in computer power will also increase the complexity of problems that can be addressed as air quality managers and other decision makers are faced with different problems. An example of this is the application of complicated explosion models in emergency planning and response. Until recently, models of this complexity could only feasibly be run on expensive workstations. These models can now be run on PCs.
The Internet will also increase the ease with which data can be gathered for input into models. Additionally, modeling results will be more easily distributed. As a result, computer modeling is expected to become even more useful in the future.
1. American Meteorological Society (1981). "Air Quality Modeling and the Clean Air Act: Recommendations to EPA on Dispersion Modeling for Regulatory Applications." Boston, American Meteorological Society.
2. Cimorelli, A. J., S. G. Perry, A. Venkatram, J.C. Weil, R. J. Paine, R. B. Wilson, R. F. Lee, and W.D. Peters,1998: "AERMOD: Description of Model," EPA Web site (
3. Scire, J.S., D.G. Strimaitis and R.J. Yarmartino, 1999. "A User's Guide for the CALPUFF Dispersion Model," Earth Tech Inc., Concord, Mass.
4. Havens, J. A. and T. O. Spicer. 1985. "Development of an Atmospheric Dispersion Model for Heavier-than-Air Gas Mixtures," Vol. I. Report CG-D-22-85 to U.S. Coast Guard. Washington, D. C: Office of Research and Development, U.S. Coast Guard, U.S. Department of Transportation.
5. Kunkel, B. A., 1991: AFTOX 4.0 -- "The Air Force Toxic Chemical Dispersion Model -- A
User's Guide. PL-TR-91-2119," Environmental Research Papers No. 1083, Phillips Laboratory,
Directorate of Geophysics, Air Force Systems Command, Hanscom AFB, Mass.
6. Tatom, F.B., Roberts, M.D., Tatom, J.W., and Miller, B.A., "High Explosive Damage Assessment Model Advanced Industrial Version (HEXDAM-D+)," Twenty-Fifth Department of Defense Explosive Safety Seminar, Anaheim, Calif., August 18-20, 1992.
7. Tatom, F.B., Tatom, J.W., and Miller, B.A., "Vapor Cloud Explosion Damage Assessment Model Second Industrial Version (VEXDAM 5.0)," Vapor Cloud Explosion Subcommittee Meeting, Center for Chemical Process Safety, AICHE, January 31, 1994.
8. Tatom, F.B., and Tatom, J.W., "Vulnerability Assessment of Structurally Damaging Impulses and Pressures (VASDIP)," Twenty-Fourth Department of Defense Explosive Safety Seminar, St. Louis, Mo., August 30, 1990.
This article originally appeared in the 04/01/2004 issue of Environmental Protection.