Grantee Research Project Results
Final Report: Rapid Mapping for Clean Air in Commerce City
EPA Grant Number: R828577Title: Rapid Mapping for Clean Air in Commerce City
Investigators: Vogt, Richard L. , Wagner, Lynn Robbio
Institution: Tri-County Health Department, CO , University of Colorado Health Sciences Center , University of Colorado at Denver , Metropolitan State College of Denver , University of Colorado at Boulder
Current Institution: Tri-County Health Department, CO , Metropolitan State College of Denver , University of Colorado Health Sciences Center , University of Colorado at Boulder , University of Colorado at Denver
EPA Project Officer: Packard, Benjamin H
Project Period: January 1, 2001 through December 31, 2002
Project Amount: $400,000
RFA: Environmental Monitoring for Public Access and Community Tracking (EMPACT) (2000) RFA Text | Recipients Lists
Research Category: Water , Air , Ecological Indicators/Assessment/Restoration
Objective:
The objectives of this research project were to: (1) develop methods for integrating an atmospheric dispersion model with a geographic information system (GIS) for the purpose of describing the spatial and temporal distribution of selected air pollutants in the industrialized urban community of Commerce City, Colorado, located in the Denver metropolitan area; and (2) develop a Web site for providing public access to maps and animations of the dispersion model data, data from other monitoring programs, and spatially referenced data that are relevant to air pollution and public health.
Summary/Accomplishments (Outputs/Outcomes):
To develop a resource for rapid mapping and reporting of air pollutants, we developed an atmospheric dispersion model for the Commerce City domain with source-term data for benzene. We have evaluated the model through comparisons with other models and with atmospheric monitoring data. We have linked the model with real-time meteorological data and a GIS that can be accessed through the Internet. The model and Internet resources demonstrate the potential for near real-time models for educational and research purposes.
Faculty and graduate students from the University of Colorado Health Sciences Center, Boulder campus and Denver campus, participated in a project to link local universities with communities. We used university resources to update the information system used by the Tri-County Health Department (TCHD) for broadcasting real-time monitoring data of meteorological conditions over the Internet and to develop an atmospheric dispersion model and a GIS. University collaborators also have interacted with local business and government groups.
To achieve our goal of providing access to atmospheric modeling, we have developed a Web site for accessing the near real-time output of the atmospheric dispersion model for benzene air concentrations in Commerce City and for providing additional information to the public.
We also explored collaborative relationships to maintain this research project. We have applied for a number of research grants to obtain support for expanding the atmospheric dispersion model to include different air pollutants and to serve as a source for exposure data in epidemiologic studies of asthma in metropolitan Denver. To date, efforts to obtain additional funding have been unsuccessful.
We are continuing to discuss sponsorship of the real-time atmospheric dispersion model application with the Air Pollution Control Division of the Colorado Department of Public Health and Environment (CDPHE). Also, we are looking for funding to expand the model to include other pollutants and a larger spatial domain.
Technical Aspects
Development of an Atmospheric Dispersion Model. We used the U.S. Environmental Protection Agency’s (EPA’s) AERMOD as the project’s atmospheric dispersion model. This model is a steady-state plume model that is designed to evaluate the short-range dispersion of air pollutants emitted from industrial facilities. The input data requirements for AERMOD include emissions rates and source geometry, meteorological data, and terrain characteristics. The AERMET component of AERMOD is used to process meteorological data for use in the dispersion model.
Development of a Source-Term Database. We obtained benzene point source and area emission data from the CDPHE for the year 2000. We used the GIS developed for Commerce City to aggregate mobile and area source emissions at the census tract level and to format these data for AERMOD in 10 km x 10 km grid cells. We used U.S. Census population data to weigh, by population size, the area sources of wood-burning, highway, and nonroad emissions, and miscellaneous sources (Sakulyanontvittaya, 2003). Data from the EPA National Emission Inventory were used to characterize releases and structure heights of point sources.
Industry representatives from the Northeast Metro Industrial Council (NEMIC) reviewed the benzene emissions for major point sources in the Commerce City area. Emissions data, which included the six-county metropolitan Denver area, then were aggregated using GIS and added to the dispersion model. The 10 largest point sources of benzene have been located on a map layer for the Web-based pollution-mapping resource.
Acquisition of Meteorological Data. The meteorological data acquired for dispersion
modeling include surface wind speed and direction, surface temperature, cloud
cover, and the temperature profile from the morning vertical sounding. For
the real-time implementation of AERMOD, local meteorological data are obtained,
via the Internet, from a network of six towers maintained by the TCHD. These
data are reported at 15-minute intervals. Upper-air soundings are obtained
twice daily from the National Oceanic and Atmospheric Administration
(http://raob.fsl.noaa.gov/ Exit ) and hourly surface observations are obtained from
the National Weather Service via the University Center for Atmospheric Research
(http://www.rap.ucar.edu/weather/surface/ Exit ).
Displaying Model Results With a Web-Based GIS. Figure 1 shows a simplified flow diagram of the system created to report in near real time the estimations of benzene concentrations generated by the AERMOD model. This figure illustrates the process from taking the tables of concentrations generated by AERMOD to the displaying of these results on a Web-based GIS (steps 4 through 6).
The AERMOD model outputs estimations of benzene concentrations for points separated every 500 meters in tables using a text file format. This output is generated four times a day (06:00, 12:00, 18:00, and 24:00 hours). The tables are processed by a Visual Basic program that performs two functions: (1) It trims the headers and eliminates other information, leaving only a file with coordinates of each point (X and Y values) and a benzene concentration estimated value (Z value). (2) It converts the X, Y, and Z text files into an ArcView shape file (Environmental Systems Research Institute [ESRI]; step 5 in Figure 1) that is used in the GIS.
Figure 1. Simplified Flow Diagram of the System to Report Estimations of Benzene Concentrations
The GIS-enabled Web Site at http://air.uchsc.edu/ Exit was created using ArcIMS (Internet map server) version 4.0 (ESRI, Redland, CA). We also created an application that allows the dynamic addition of layers to the map server. The ArcIMS application displays several geographic layers (see Figure 2). When the user clicks the “Pollution Map” button, an interface is displayed, allowing the selection of the year, month, day, and time for an estimated benzene concentration map generated by AERMOD (see Figure 3).
Figure 2. GIS-Enabled Web Site
Figure 3. Estimated Annual Average Benzene Concentrations for 1990
Technical Difficulties. We have overcome a number of technical problems in the process of implementing the real-time version of the model. The most difficult problems have been the acquisition of a stable data stream from the local meteorological stations and providing alternative data displays when appropriate real-time data are temporarily unavailable. We also have experienced problems in keeping the ArcIMS software running on the computer system that our server accesses.
Future implementations of real-time dispersion models for the Denver airshed will require obtaining stable sources of local data. The National Weather Service station at Denver International Airport is the only reliable source that can be accessed easily via the Internet, but the meteorological conditions at this site are not representative of those near the city’s center. The reliance upon ArcIMS software requires the Web site sponsor to have a site license and technical staff that can keep the software running, which may be too expensive for some organizations. The problems with expense and system stability might be solved in the future by developing solutions with open-source software.
Outreach
Business Community. Throughout this project, updates and presentations were made to the NEMIC. The group encourages comments from the community and local businesses concerning environmental issues. The NEMIC has been instrumental in supporting the meteorological station network. They have provided funding for the original installation for the stations and have maintained the quality assurance for the stations located on their properties. The TCHD will continue to work with the NEMIC to respond to citizens’ concerns and provide the continued operation of the meteorological stations.
Citizens’ Groups. The TCHD maintains working relationships with many community groups in the Commerce City area. Representatives of the TCHD routinely attend meetings of the Conoco Citizens’ Advisory Council, Front Range Earth Force, Ground Work Denver, and the Colorado People’s Environmental and Economic Network concerning environmental issues. We have made presentations to these groups and have solicited comments. In the future, we plan to continue working on ways to use real-time models and the Internet to clarify problems of air quality to residents of Commerce City and metropolitan Denver.
Adams County Schools. Over the course of this project, we met with many representatives of the Adams County School District to discuss the ways in which project collaborators could contribute to educational activities for the school system. Our initial plans have been disrupted by substantial budget cuts and faculty turnover. Representatives of the TCHD have made presentations on the Environmental Monitoring for Public Access and Community Tracking project and public health aspects of air pollution to six classes at the Adams County High School in 2003. Currently, the TCHD is working with Ground Work Denver to develop educational materials in Spanish. University of Colorado faculty members will continue to offer support for educational projects in environmental science and health. We are concerned, however, that the school system has little interest in using our resources.
Governmental Agencies. We made presentations to a number of governmental agencies and working groups including the Rocky Mountain Arsenal, the Northeast Pollution Prevention Alliance, the Regional Air Quality Council, the Denver Department of Health, EPA Region 8, and the CDPHE. In the future, we plan to demonstrate the Web site to many of these agencies with the hope that we can identify partners and future funding for expanding the capabilities of the model and the educational components of the Web site.
Recently, the Denver Department of Health and Environment (DDHE) developed an atmospheric dispersion model using the Industrial Source Code-3-ST (ISCST3) model and emissions data from stationary, area, and mobile sources. They used a GIS to validate stationary source coordinates, spatially allocate county-level emissions to census block groups, and display air dispersion model results. Emissions from the six-county region were included in the model, but annual average air concentrations of selected pollutants were predicted only for Denver County and surrounding areas including Commerce City, Aurora, Highlands Ranch, and Lakewood. Although the DDHE currently produces only annual average estimates of selected pollutants, its source-term data could be used for producing air pollutant concentrations for the entire six-county metropolitan area with AERMOD. We have discussed future collaborations with the DDHE and have applied jointly for funding; to date, we have been unsuccessful.
Graduate Student Involvement
This project provided the opportunity for two graduate students to conduct research projects for their Master’s degrees. Tanarit Sakulyanontvittaya produced a thesis from his work on AERMOD, and Phuntip Wigraisakda produced a Master’s report on ambient air sampling for benzene in Commerce City. A number of other graduate students from all three University of Colorado campuses have worked on various aspects of this project.
Evaluation of Technical Effectiveness
We have shown the technical feasibility of the rapid mapping of air pollutants with an Internet-based system of an atmospheric dispersion model and a GIS. The development of the GIS and source-term database is not complicated, but can be done most economically if the work is performed in cooperation with the state and local agencies that collect these data for regulatory purposes. If the system is implemented by an organization that has affordable access to ESRI software and technical staff who are familiar with ArcIMS, the system can be implemented and maintained with reasonable costs. The system can enhance the public’s understanding of the spatial and temporal distribution of air pollutants in an urban airshed. It also can provide exposure data for epidemiologic studies and be used to help verify the accuracy of reports of routine pollutant releases to regulatory agencies.
Description of the Atmospheric Dispersion Model
Model Description. AERMOD was developed by the American Meteorological Society and EPA (EPA, 1998a) as a steady-state plume model designed to evaluate the short-range dispersion of air pollutants emitted from industrial facilities. The input data requirements for AERMOD include emissions rates and source geometry, meteorological data, and terrain characteristics. We employed version 98314 of AERMOD and AERMET (the meteorological preprocessor that develops and organizes meteorological parameters for the model), and version 98022 of AERMAP (the preprocessor for digital terrain data).
Performance Criteria. Our implementation of AERMOD should predict concentrations of benzene in air that are comparable to those predicted by the ISCST3 model using the same source-term and meteorological data. The AERMOD predictions also should be consistent with measurements made at locations within the model’s spatial domain.
Model Testing. For comparisons between models for average monthly and annual air concentrations of benzene, we used historical meteorological data for the year 1990 from the Denver Stapleton International Airport and 1999 emissions estimates for benzene. We compared the results from AERMOD (using two parameter sets for surface characteristics, ParS1 and ParS2) and the ISCST3 at four receptor locations.
The receptor-by-receptor comparisons for all hours in 1990 at four receptors showed that the 1-hour concentration ratios of AERMOD to ISCST3 ranged from 0.5 to 2.0 for ParS1 and from 1.0 to 3.0 for ParS2 (Sakulyanontvittaya, 2003). AERMOD, with the two parameter sets, produced higher concentrations than ISCST3 for stable conditions and lower concentrations for convective conditions. The variation in concentration ratios, therefore, is believed to be a result of the different model assumptions of meteorological conditions. AERMOD uses a bi-Gaussian dispersion equation in the vertical direction for convective conditions, causing more dispersion than ISCST3. That is, AERMOD produces higher variation and lower concentrations for convective conditions than ISCST3.
The study of the effects from meteorological parameters also showed that the percentage of AERMOD (ParS1) to ISCST3 concentration ratios that were greater than 1 in stable conditions ranged from 70 percent to 81 percent, and the percentage of ratios less than 1 in convective conditions ranged from 67 percent to 86 percent. For the ParS2 conditions, the percentage of ratios greater than 1 in stable conditions ranged from 95 percent to 98 percent, and the percentage of ratios less than 1 in convective conditions ranged from 43 percent to 74 percent (Sakulyanontvittaya, 2003).
Comparing the two parameter sets for AERMOD, ParS2 (with lower surface roughness lengths [Zo]) resulted in a higher concentration and larger variation than ParS1 (with higher Zo). The average concentrations and values for the robust highest concentration statistic were identified, and the results from AERMOD showed ParS2 to be the highest for all four receptor locations (Sakulyanontvittaya, 2003). The average monthly concentrations plotted against time show that concentrations during winter and autumn were higher than during summer and spring. This result can be explained by the meteorology for each season. Inversion heights are lower and stable conditions occur more often during winter and autumn. Inversion heights are higher and convective conditions occur more often during summer and spring.
Compared to single receptor evaluation, examination of concentrations across all receptors in the study area showed that the variation in AERMOD to ISCST3 concentration ratios across receptors was very small compared to the variation with time. This means that the terrain algorithms of the two models did not cause much difference for the relatively flat terrain of the model domain. As a result, AERMOD and ISCST3 had an excellent spatial correlation, approximately 0.971 (Sakulyanontvittaya, 2003).
The study of the effects from meteorological parameters showed that AERMOD is strongly sensitive to wind speed, wind direction, Monin-Obukhov length, and surface roughness length, and ISCST3 is strongly sensitive to wind speed, wind direction, and Pasquill stability class. The Monin-Obukhov length used in AERMOD and the Pasquill stability class used in ISCST3 are indicators of atmospheric stability. The value of the Pasquill stability class is discrete and therefore limits the predictive accuracy of ISCST3. The two models are affected by wind speed in a similar manner; concentrations of benzene are higher at lower wind speeds. The concentrations predicted by both models decreased non-linearly with wind speed. The variation in benzene concentrations between models was noticeable at low wind speeds but decreased when the wind speed increased. The evaluation of the effect of surface roughness length showed that lower length caused higher concentration because of greater mixing. The lower surface roughness length also caused greater variations in benzene concentrations. Our evaluations indicated that compared with ISCST3, AERMOD has more flexibility for use on any type of surface.
In the comparison between computational results and measurements, AERMOD and ParS2 were used to estimate the concentrations at 10 sites where benzene concentrations had been monitored within the model domain. The entire data average concentrations show that five sites—AQ3, AQ4, AQ5, MBHS, and SWCO—had good agreement between the model results and observations. The average computational results of the other five sites were less than the average measurements but only by a factor of two or smaller.
The contour plots of computed concentrations with AERMOD and ParS2 conditions indicate high concentrations along I-25, I-270, and I-70; and at Denver International Airport (Sakulyanontvittaya, 2003). Emissions from mobile sources are most likely responsible for these comparatively high concentrations. In addition, the area along I-270 is affected by many high emitting point sources.
We also have collected our own air samples that have been analyzed for concentrations of benzene and other volatile organic compounds (Wigraisakda, 2003). These samples were collected for 2-hour time periods in the morning and afternoon. We plan to compare these measurements with model predictions for these specific time periods to assess the quality of short-term predictions.
Theory Behind the Model. AERMOD consists of two preprocessors and the dispersion model. These preprocessors are the AERMIC meteorological preprocessor (AERMET) and the AERMIC terrain preprocessor (AERMAP). AERMET is the meteorological data analyzer and provides planetary boundary layer information to AERMOD. AERMAP is the surface analyzer that reads digital terrain data and generates terrain/receptor information for AERMOD. Details about the meteorological data and terrain data will be discussed in later sections. Like ISCST3, AERMOD estimates pollutant concentrations by using a normal Gaussian distribution in both the vertical and horizontal directions in the case of stable conditions. The AERMOD, on the other hand, uses a bi-Gaussian distribution for vertical dispersion and a normal Gaussian distribution for horizontal dispersion for convective conditions. In convective conditions, updraft and downdraft dominate the dispersion of the plume so that the dispersion is higher in the vertical direction. For this reason, AERMOD uses a bi-Gaussian dispersion equation to explain this phenomenon.
AERMOD is written in FORTRAN and developed for IBM-compatible PCs. It is designed to operate on MS-DOS version 3.2 or higher. The executable AERMOD and user’s manual are available on the Web site of EPA’s Support Center for Regulatory Air Models. In general, AERMOD needs three setup files, a runstream file, and two meteorological data files. In some cases, the user can specify an additional meteorological data file, with onsite meteorological data. The meteorological data and the external emission data can be included in the concentration estimation. Like ISCST3, the runstream contains five major pathways. Both runstream files for AERMOD and ISCST3 are in the same format, except some options in AERMOD are added and changed.
Mathematics. AERMOD has been documented extensively and reviewed by EPA and the American Meteorological Society. Publications describing the mathematics behind the model and peer review are available on EPA’s Web site and are summarized by Sakulyanontvittaya (2003) and EPA (1998a).
Model Uncertainty. A comparison between the AERMOD and atmospheric dispersion models is provided by EPA (1999). We are not aware of other evaluations of model uncertainty.
Input Data Requirements. The input data requirements are described by EPA (1998a) and Sakulyanontvittaya (2003).
Hardware Requirements. The AERMOD model was developed on an IBM-compatible PC using the Lahey Fortran 90 Compiler (Release 4.50i) and has been designed to run on PCs with an 80386 or higher central processing unit chip, a minimum of 2 MB of RAM, a math coprocessor, and MS-DOS version 3.2 or higher. To handle the input data files (i.e., runstream setup and meteorology) and the output files, it is required that the system has a hard disk drive. The amount of storage space required on the hard disk for a particular application will depend greatly on the output options selected.
Documentation. The user’s guide for version 98314 of AERMOD is published as EPA, 1998a; for the version 98314 of AERMET, EPA, 1998b; and for version 98022 of AERMAP, EPA, 1998c.
References:
U.S. Environmental Protection Agency. Revised Draft: User’s Guide for the AMS/EPA Regulatory Model–AERMOD. Research Triangle Park, NC: U.S. Environmental Protection Agency, November 1998a, 211 pp.
U.S. Environmental Protection Agency. Revised Draft: User’s Guide for the AERMOD Meteorological Preprocessor (AERMET). Research Triangle Park, NC: U.S. Environmental Protection Agency, November 1998b, 273 pp.
U.S. Environmental Protection Agency. Revised Draft: User’s Guide for the AERMOD Terrain Preprocessor (AERMAP). Research Triangle Park, NC: U.S. Environmental Protection Agency, November 1998c, 99 pp.
Peters WD, Paine RJ, Lee RF, Wilson RB, Cimorelli AJ, Perry SG, Weil JC, Venkatram A. Comparison of Regulatory Design Concentrations: AERMOD versus ISCST3 and CTDMPLUS. U.S. Environmental Protection Agency, April 1999, 112 pp.
Journal Articles:
No journal articles submitted with this report: View all 7 publications for this projectSupplemental Keywords:
innovation technology, decisionmaking, Rocky Mountain region, air pollution, mapping, atmospheric dispersion models, benzene, chemical mixtures, chromium, urban air toxics,, RFA, Scientific Discipline, Air, Toxics, Geographic Area, Ecosystem Protection/Environmental Exposure & Risk, Air Pollution, particulate matter, air toxics, Environmental Chemistry, HAPS, State, Monitoring/Modeling, Environmental Monitoring, tropospheric ozone, Ecology and Ecosystems, 33/50, EMPACT, atmospheric dispersion models, particulates, urban air toxics, stratospheric ozone, chromium & chromium compounds, community-based approach, diesel particulates, Rocky Mountain Arsenal, public information, benzene, Sulfur dioxide, chemical mixtures, air pollution models, urban air pollutants, web site development, industrial air pollution, Chromium Compounds, public outreach, rapid mapping, Colorado (CO), air quality, atmospheric chemistryRelevant Websites:
Progress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.