RNS
Browse Projects > Detailed View

Characterizing Uncertainty in Transportation Air Quality Modeling

Description:

Although a high level of uncertainty is known to exist in air quality modeling for transportation projects, it is rarely assessed to enable decision-makers to make more informed decisions. For example, when “worst case” or conservative modeling assumptions are systematically applied, model results (e.g. forecasts) may overstate the potential for air quality impacts and harm to public health from transportation projects, thus failing to provide an accurate assessment of the true potential impacts. In another example, differences in potential air quality impacts between project alternatives may be considered significant or substantive whereas in reality they may be within the margin of error if uncertainty was assessed. Ultimately, addressing uncertainty for air quality modeling for transportation projects is about making good decisions, using funds wisely, and maintaining public trust and confidence.

Determining and communicating uncertainty in modeling results is a long-standing challenge in mobile source air quality modeling. Modeling results are typically the product of an integrated chain of models that include travel demand, vehicle emissions, and dispersion models, each with their own inherent level of uncertainty. Errors and uncertainty propagate from modeling inputs through each step of these modeling chains, but are rarely evaluated before the results are used in decision-making. Common sources of potential error or uncertainty in mobile source air quality modeling include: (1) Travel demand modeling - traffic volume, travel speeds; current and forecasted; (2) Vehicle emission modeling - emission factors by vehicle type, vehicle fleet turnover rates, fleet age, vehicle mix, driving cycles, future emission factor estimates; (3) Dispersion modeling - background concentrations, meteorology, choice of model formulation, chemical reactions, near roadway barriers (trees, walls, buildings, etc.). (4) Algorithms developed and used in models which are the underlying principles for the model results (e.g. calculating emission concentrations).

Land-use and traffic simulation modeling add additional sources of uncertainty, as does the exposure analysis step in a Health Risk Assessment (e.g., activity patterns, indoor versus outdoor concentrations, inhalation rates). Additional uncertainty may result from the complexity of these models, which increases the risk of user errors in selecting appropriate parameters and model setup. Further, models are often not subject to the type of rigorous review and update processes that are needed to improve them and reduce uncertainty over time. A rigorous assessment of uncertainty in transportation air quality analyses can help analysts understand how much confidence to place in the various steps of a required air quality analysis and support the preparation and review of these required evaluations by state departments of transportation (DOT) staff. This proposed study will comprehensively address the topic of uncertainty in mobile source air quality modeling chains by identifying critical sources of uncertainty and providing guidance on how to measure uncertainty and appropriately characterize and communicate model results.

Objective:

The primary objective is to determine how to best characterize and communicate the uncertainty associated with mobile source air quality modeling results for each model input (i.e. travel demand, vehicle emissions, dispersion models, etc.) and the whole modeling chain in order to appropriately inform decision-makers and the public on the potential impacts of a project, program, or plan. This includes identifying inputs that have the greatest impact on modeled outcomes and recommending methods to cost-effectively reduce uncertainty and analysis costs.

Benefits:

This research will support more informed decision-making both at the regional planning and project level, which currently proceeds with limited to no information on uncertainty. It will relay information on uncertainty to the public in a simple and concise manner. It will also help focus modeling efforts on inputs that provide the most benefit for the least cost, resulting in reduced overall costs and streamlined project development. Over the long-term, it will also lead to the development of improved models and input data that are needed for the development of more accurate and less uncertain modeling results (e.g. forecasts).

Related Research:

No research was found that looks at uncertainty for the whole transportation air quality modeling chain. The National Research Council (NRC) highlighted the need to address uncertainty in regulatory modeling for environmental applications in a landmark 2007 report, "Models in Environmental Regulatory Decision Making”. The report assessed EPA’s use of computer models in developing regulations and recommended a series of guidelines and principles for improving agency models and decision-making processes for considering uncertainty. Given the long duration in service (decades) of EPA models, the NRC report identified a critical need for a periodic and transparent review and update process for each model. This process would ensure that the models continue to meet regulatory needs with requisite accuracy (for which uncertainty must be assessed and appropriately communicated) and are efficient and cost-effective to apply. The findings and recommendations of the 2007 NRC study though critical to good modeling practices have not yet been implemented for the specific application of transportation air quality analyses. This proposed study would address that gap for the transportation air quality modeling chain.

EPA addressed uncertainty in a 1998 report on “Quantification of Uncertainty in Air Quality Models Used for Analysis of Ozone Control Strategies”, which demonstrated the use of systematic methods to quantify uncertainty (Bayesian Monte Carlo) and identified influential sources of uncertainty in photochemical air quality models. The EPA report emphasized the importance of utilizing the results to help decision makers gauge the reliability of models used for assessing the effect of proposed control measures and set priorities for further research to improve predictive capabilities. Early studies on estimating uncertainties of dispersion models include Lamb and Hati (1987), Venkatram (1988) Lewellen and Sykes (1989), and Weil et al. (1992). Pielke (1998) points out the need to assess the uncertainties of linked meteorological and dispersion models. A study by Hanna (2007) provided a framework for uncertainty analysis of atmospheric models used to estimate transport and dispersion, and to tie this framework to what can be used in other environmental and risk assessment disciplines. A study by Borrego et al. (2008) presented a review of the existing methodologies to estimate air quality modeling uncertainty and produced guidelines for modeling uncertainty estimation, which can be used by local and regional authorities for air quality management. They suggested three types of uncertainty analysis depending on the level of complexity: qualitative analysis (analysis of model results against measured), total model uncertainty (quantitative analysis of results using a set of statistical parameters) and total model uncertainty by components (sensitivity analysis and/or model inter-comparison to evaluate different model modules). It is important to note that most of these studies have focused on a specific modeling component. However, it is essential to look at uncertainty analysis for the entire modeling chain, as uncertainties propagate through the entire chain.

The literature is limited in terms of quantifying the range or degree of uncertainty introduced by the dispersion modeling step (e.g., using AERMOD or CAL3QHCR). While travel demand and emission modeling are well established and widely used analytical methods, air dispersion modeling is only now emerging as a method for mobile source air quality analysis. This is an important gap since U.S. EPA conformity requirements have recently been extended to cover modeling near-road particulate matter concentrations using dispersion models. Streamlining compliance with these complicated and challenging new modeling requirements would be of significant benefit to State DOTs and other project sponsors.

A number of studies have shown that there is a critical need to address uncertainties in health risk assessments (HRA) due to incomplete and unavailable information (OEHHA, 2003; Claggett and Houk 2006, Claggett and Miller 2005, FAA 2001, FAA 2003, FAA2005a, and FHWA 2006). A U.S. Environmental Protection Agency guidance document indicates that health risk assessments of national and state policies should include sensitivity and uncertainty analyses (EPA, 2001). Nonetheless, current HRA methodologies do not report variability and uncertainty in their health impact estimates. Variability and uncertainty in estimated health risks of traffic-related air pollution arises from multiple sources that includes population, weather patterns, traffic, geographic features, accuracy of models involved in the modeling chain, etc. (Frey and Burmaster, 1999). For example, EPA conducted risk assessments near 63 schools throughout the U.S. as part of their study “Assessing Outdoor Air Near Schools”. While the risk assessment indicated 14 schools were above short or long-term health thresholds for mobile source related pollutants, the actual monitoring data showed all 14 schools were well below any risk thresholds. After this study was completed, EPA requested several state DOTs to conduct HRAs under NEPA for large transportation projects. However, FHWA indicated the outcome of such an assessment, adverse or not, would be influenced more by the uncertainty introduced into the process through assumption and speculation rather than any genuine insight into the actual health impacts directly attributable to emission exposure associated with a proposed action. Documenting HRA uncertainties can be used, as needed, in NEPA administrative records.

Related Studies:

Beckx, C., L. Int Panis, K. Van De Vel, T. Arentze, W. Lefebvre, D. Janssens, and G. Wets. 2009. The contribution of activity-based transport models to air quality modelling: A validation of the ALBATROSS–AURORA model chain. Science of The Total Environment, Vol. 407, No. 12, pp. 3814–3822.

Borrego, C., Monteiro, A., Ferreira, J., Miranda, A.M., Costa, A.M., Carvalho, A.C., Lopes, M. 2008. Procedures for Estimation of Modelling Uncertainty in Air Quality Assessment. Environment International, Vol 34(5), pp. 613–620.

California Office of Environmental Health Hazard Assessment (OEHHA). 2003. Air Toxics Hot Spots Program Risk Assessment Guidelines. The Air Toxics Hot Spots Program Guidance Manual for Preparation of Health Risk Assessments.

Claggett and Miller 2005. A Method for Evaluating Mobile Source Air Toxic Emissions Among Transportation Project Alternatives. Paper presented at Air and Waste Management Association 98th Annual Conference and Exhibition. Minneapolis, MN. June 2005.

Claggett and Houk 2006. Claggett, Michael, Ph.D., and Jeffrey Houk. FHWA Workshop on Project-Level Mobile Source Air Toxics. Phoenix, AZ. March 7, 2006.

Chen, H., S. Bai, D. Eisinger, D. Niemeier, and M. Claggett. 2009. Predicting Near-Road PM2.5 Concentrations. Transportation Research Record: Journal of the Transportation Research Board, Vol. 2123, pp. 26–37.

EPA 2009-2013. U.S. Environmental Protection Agency. Study: Assessing Outdoor Air Near Schools; Washington D.C. March 2013. http://archive.epa.gov/schoolair/web/html/index.html

FAA 2001. U.S. Department of Transportation, Federal Aviation Administration. LAX Master Plan EIS/EIR, Technical Report14a. Human Health Risk Assessment. Los Angeles, CA. January 2001. http://www.laxmasterplan.org/docs/drafteirNE/T14a_LR.pdf.

FAA 2003. U.S. Department of Transportation, Federal Aviation Administration, Office of Environment and Energy. Select Resource Materials and Annotated Bibliography On The Topic of Hazardous Air Pollutants (HAPs) Associated With Aircraft, Airports, and Aviation. Technical Directive Memorandum D01-010. Washington, DC. July 1, 2003

FAA 2005a. U.S. Department of Transportation, Federal Aviation Administration. O’Hare Modernization Final Environmental Impact Statement, Appendix I, Hazardous Air Pollutant Discussion. Chicago, IL. July 2005. ftp://publicftp.agl.faa.gov/ORD FEIS/Appendix I.pdf.

FHWA 2006. U.S. Department of Transportation, Federal Highway Administration. Interim Guidance on Air Toxic Analysis in NEPA Documents. Memorandum from Cynthia Burbank to Division Administrators. http://www.fhwa.dot.gov/environment/airtoxic/020306guidmem.htm.

Frey, H.C and Burmaster, D.E. 1999. Methods for characterizing variability and uncertainty: comparison of bootstrap simulation and likelihood-based approaches. Risk Analysis 19 (1), 109–130.

Hanna, S.R. 2007. A Review of Uncertainty and Sensitivity Analyses of Atmospheric Transport and Dispersion Models. In Developments in Environmental Science, 6: Air Pollution Modelling and its Applications XVIII, C.Borrego and E.Renner (Editors), WIT Press, Southampton, UK, pp. 331-351.

Heist, D., V. Isakov, S. Perry, M. Snyder, A. Venkatram, C. Hood, J. Stocker, D. Carruthers, S. Arunachalam, and R. C. Owen. 2013. Estimating near-road pollutant dispersion: A model inter-comparison. Transportation Research Part D: Transport and Environment, Vol. 25, pp. 93–105.

Lamb, R.G., Hati, S.K., 1987. The representation of atmospheric motion in models of regional-scale air pollution. Journal of Applied Meteorology and Climatology. Vol 26(7), 837–846.

Lefebvre, W., J. Vercauteren, L. Schrooten, S. Janssen, B. Degraeuwe, W. Maenhaut, I. de Vlieger, J. Vankerkom, G. Cosemans, C. Mensink, N. Veldeman, F. Deutsch, S. Van Looy, W. Peelaerts, and F. Lefebre. 2011. Validation of the MIMOSA-AURORA-IFDM model chain for policy support: Modeling concentrations of elemental carbon in Flanders. Atmospheric Environment, Vol. 45, No. 37, pp. 6705–6713.

Lewellen, W.S., Sykes, R.I., 1989. Meteorological data needs for modeling air quality uncertainties. Journal of Atmospheric and Oceanic Technology Vol 6(5), pp.759–768. Lin, J. and Vallamsundar. S. 2013. Transportation Conformity Particulate Matter Hot-Spot Air QualityModeling, preparedfor Illinois Department of Transportation, Illinois Center for Transportation, Report No. FHWA-ICT-13-024.

NRC 2007. Committee on Models in the Regulatory Decision Process, National Research Council. Models in Environmental Regulatory Decision Making. National Academies Press. ISBN: 0-309-11001-7 (2007). http://www.nap.edu/catalog/11972/models-in-environmental-regulatory-decision-making

Radonjic, Z., Chambers, D.B., and Kirkaldy, J. 2003. “Modelling Line Sources (Roads) Using CAL3QHCR, ISC3, AERMOD and CALPUFF.” Presented at the Air and Waste Management Annual Conference and Exhibition, Mystic, CT, October 22–24, 2003. Schroeder, A.J., and Schewe, G.J. 2009. “Sensitivity of AERMOD to Meteorological Data Sets Based on Varying Surface Roughness.” Presented at the Air and Waste Management Annual Conference and Exhibition, Paper No. 168, Detroit, MI, June 16–19, 2009.

U.S Environmental Protection Agency. 1998. Quantification of Uncertainty in Air Quality Models Used for Analysis of Ozone Control Strategies. Accessed at http://cfpub.epa.gov/ncer_abstracts/index.cfm/fuseaction/display.highlight/abstract/50/report/F

Venkatram, A. 1983. Uncertainty in Predictions from Air Quality Models. Boundary Layer Meteorology, 27: 185-196.

Venkatram, A. 1988. Inherent Uncertainty in Air Quality Modeling. Atmospheric Environment, 22 (6): 1221-1227.

Weil, J.C., Sykes, R.I., Venkatram, A., 1992. Evaluating air quality models: Review and outlook. Journal of Applied Meteorology and Climatology. Vol 31, pp. 1121–1145.

Zou, B., Zeng, Y., Liu, H., Zhang, H., Qiu, L., and Zhan, B.F. 2010. “Sensitivity Analysis of AERMOD in Modeling Local Air Quality under Different Model Options.” Presented at Bioinformatics and Biomedical Engineering, 4th International Conference, pp.1–4.

Tasks:

Potential tasks include the following:

1.Through consultation with the project panel, identify the specific models and modeling chains to be assessed in this study. Modeling systems vary from place to place. For example, some metropolitan areas maintain activity based travel demand models while others use more traditional four step transportation models. Either of these travel demand models may also be integrated with a land-use simulation model. The two approved vehicle emission models are EMission FACtor (EMFAC) for California and the US EPA Motor Vehicle Emissions Simulator (MOVES) for all other states. To accommodate variation in preference, the US EPA also lists several preferred and recommended dispersion models. Furthermore, different modeling chains are often used for project level and regional scale analyses. Given all of these inconsistencies, this task is aimed at identifying the most representative models and modeling chains to evaluate so that the project’s findings are useful for varied local conditions.

2.For each modeling step (i.e., traffic, emissions and dispersion modeling for project-level analyses):

a) Identify potential sources of uncertainties. For this purpose, a detailed literature review is expected. Data for example comparing modeling predictions to monitoring data are available from well-established near-road studies (e.g., Las Vegas). Additionally, the detailed assessment of model sensitivities (in the next subtask) would also be informed by studies addressed in the literature review

b) Determine and document model sensitivities to uncertainties in model inputs and other analysis assumptions.

c)Determine and document the costs for obtaining the necessary inputs for each model and the incremental costs for improving or refining those inputs to improve accuracy and/or reduce uncertainty.

d) Document/illustrate the modeling and data process flow, presenting the results of uncertainty within individual steps of the modeling chain and cumulatively along the modeling chain. Identify and prioritize the most important or critical modeling inputs based on consideration of: (1) their contribution to reducing modeling uncertainty for the individual model and the overall modeling chain and (2) the costs for obtaining or determining those inputs.

3.Investigate and assess alternative means of communicating uncertainty in transportation and air quality modeling results to better inform decision-makers and stakeholders on actual impacts of proposed projects, programs, and plans, as applicable for each specific regulatory application (project-level analysis, regional analysis, HRA, etc.) and considering the recommendations of the 2007 NRC report.

4.Prepare draft and final reports for panel review. Document all findings

Sponsoring Committee:AMS10, Air Quality and Greenhouse Gas Mitigation
Source Info:Ms. Jackie Ploch, Air Quality and Noise Work Leader, Texas Department of Transportation, Chair AASHTO SCOE Air Quality, Climate change and Energy Subcommittee, and TRB ADC 20 member

Mr. Christopher Voigt, Environmental Engineer Senior, Virginia Department of Transportation, Research Coordinator, AASHTO SCOE Air Quality, Climate Change and Energy Subcommittee, and Chair, TRB ADC20 Subcommittee on Project-Level Air Quality Analysis,

Mr. Kevin Black, Air Quality Specialist, FHWA Resource Center,

Mr. Jeff Houk, FHWA Resource Center,

Dr. Jane Lin, Chair, Standing Committee on Transportation and Air Quality (ADC 20), Associate Professor, University of Illinois, Chicago,

Dr. Josias Zietsman, Chair, ADC 20 Research Subcommittee, Head, Environment and Air Quality Division, Texas A&M Transportation Institute (TTI),

Dr. Douglas Eisinger, Chair, ADC 20 Subcommittee on Regional Air Quality Analysis, Principal Scientist, Transportation Policy and Planning, Sonoma Technology, Inc.,

Dr. Reza Farzaneh, Associate Research Engineer, Air Quality Program, Texas A&M Transportation Institute and

Dr. Suriya Vallamsundar, Assistant Research Scientist, Environment and Air Quality Division, Texas A&M Transportation Institute.

Dr. Gregory Rowangould, Assistant Professor, Department of Civil Engineering, University of New Mexico
Date Posted:12/17/2015
Date Modified:12/27/2015
Index Terms:Air quality management, Uncertainty, Public health, Land use models, Traffic simulation, Pollutants, Exhaust gases, Vehicle mix, Traffic volume, Travel demand,
Cosponsoring Committees: 
Subjects    
Highways
Transportation (General)
Environment

Please click here if you wish to share information or are aware of any research underway that addresses issues in this research needs statement. The information may be helpful to the sponsoring committee in keeping the statement up-to-date.