Grantee Research Project Results
2000 Progress Report: Development and Demonstration of a Methodology for Characterizing and Managing Uncertainties in Emission Inventories
EPA Grant Number: R826766Title: Development and Demonstration of a Methodology for Characterizing and Managing Uncertainties in Emission Inventories
Investigators: Frey, H. Christopher , Loughlin, Daniel , Houyoux, Marc
Institution: North Carolina State University , MCNC / North Carolina Supercomputing Center
EPA Project Officer: Hahn, Intaek
Project Period: October 1, 1998 through September 30, 2001
Project Period Covered by this Report: October 1, 1999 through September 30, 2000
Project Amount: $553,298
RFA: Air Pollution Chemistry and Physics (1998) RFA Text | Recipients Lists
Research Category: Air Quality and Air Toxics , Air , Safer Chemicals
Objective:
We hypothesize that quantification of uncertainty in Emission Inventories (EIs) will lead to new insights regarding the quality of EIs, the best resource allocation to improve EIs, and decision-making for air quality management. The objectives of this work are to: (1) develop and refine methods for quantitative analysis of variability and uncertainty in EIs; (2) demonstrate the methods via application to a detailed case study of an EI; and (3) characterize the benefits of the techniques for environmental and research management.Progress Summary:
A key aspect of this project is to develop and demonstrate new methods for probabilistic analysis of emission inventories. As part of this task, methods for quantifying both variability and uncertainty have been developed and demonstrated using case studies involving data for selected emission sources. The methods include developing data sets, fitting parametric probability distributions to the data to represent inter-unit variability, estimating sampling distributions for the statistics of the fitted distributions as a means for quantifying uncertainty (e.g., for a population average emission rate), identifying and quantifying dependencies among emissions estimates, developing emissions estimates for averaging times appropriate for a particular emission inventory, analysis methods for cases in which some data are censored (e.g., when there are "non-detects"), and analysis methods for cases in which there are significant errors in each individual data point.As part of the methods development task, a systematic framework has been developed for fitting distributions to data using at least two alternative statistical estimation methods: method of matching moments (MoMM), and maximum likelihood estimation (MLE). Alternative parametric distributions may be fit to a data set using either of the estimation methods. Bootstrap simulation is used to quantify confidence intervals of the fitted cumulative distribution function. Comparisons of the fitted distribution and confidence intervals with the data enable a visual evaluation of goodness-of-fit even in situations where sample sizes are too small to use standard statistical goodness-of-fit techniques. Methods have been demonstrated for fitting distributions to censored data and for fitting mixture distributions to data to obtain improved goodness of fit. For both of these techniques, bootstrap simulation methods also have been demonstrated. For example, when fitting a distribution to censored data, it is possible to quantify the confidence interval for the fitted distribution both in the range of observed data and in the extrapolated region where there are non-detected measurements.
The methods described here have been applied to a variety of case studies. The case studies include the following source categories:
? Coal-fired power plants.
? Lawn and garden equipment.
? Construction,
farm, and industrial equipment.
? Stationary natural gas-fueled engines
(e.g., compressor stations).
? Bulk gasoline terminals.
For some source categories, work also has focused on development of probabilistic models. This type of work has primarily been related to highway vehicle emissions, for which probabilistic analysis of the existing mobile emission factor models has been updated, as in the case of light duty gasoline vehicles. In addition, we are in the process of evaluating statistical models for use in predicting hourly variability in emissions for coal-fired power plants.
A benefit of probabilistic analysis is the identification of new ways to find errors in emission inventories. One finding is that, by evaluating variability in data, such as for activity and emission factors for power plants, it is often possible to identify some data that are inconsistent with others. For example, the distribution of heat rates among power plant units of a given technology configuration are expected to be within a particular range of values. If a power plant heat rate is found to be low, which reflects an unrealistically high efficiency, then there is reason to suspect that there may be an error in the data. The possible reasons for such errors can be investigated and, based upon informed judgment, a decision can be made regarding whether to remove the suspect data from the database.
As part of preliminary work to develop a probabilistic emission inventory system, MCNC has done planning and initial designs of updates to the SMOKE model for propagation of uncertainties through the emissions processing step of modeling.
The quality of emission estimates has been evaluated for individual source categories by comparing probabilistic estimates of emissions with the traditional point estimates previously and currently employed in practice. A key benefit of probabilistic analysis include the capability to characterize the quality of emissions estimates using quantitative estimates of the range of variability and uncertainty in the estimates. Knowledge of the range of uncertainty in an emission estimate can be used, for example, to determine whether additional data are needed in order to reduce uncertainty and to help in setting program goals for emission inventory improvement. Another key benefit of probabilistic emission estimates is to be able to evaluate the impact of uncertainty in emissions for individual source categories with respect to uncertainty in a total emission inventory.
Future Activities:
Planned future activities include: additional methods development (primarily regarding measurement errors), additional probabilistic case studies of emission inventories for specific source categories, probabilistic emission inventories for specific source categories and for a total inventory, case studies of identification of key sources of uncertainty in emission inventories, an example case study of propagation of uncertainty in an emission inventory through an air quality model, refinement of the identification of key benefits of probabilistic analysis, additional publications and presentations based upon project work.Journal Articles:
No journal articles submitted with this report: View all 45 publications for this projectSupplemental Keywords:
air, mobile sources, nitrogen oxides, VOC, public policy, engineering, modeling, agriculture, business, transportation, industry., Scientific Discipline, Air, Mathematics, Physics, Chemistry, Engineering, Chemistry, & Physics, environmental monitoring, air quality standards, innovative emissions estimation models, air modeling, decision making, variability, emissions inventory, propagation of uncetainty, quatitative analysis, characterizing uncetaintiesRelevant Websites:
http://www4.ncsu.edu/~frey/
http://www.mcnc.org
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.