Monte Carlo Analysis Of Uncertainty In Greenhouse Gas Emission Estimates: Carbon Dioxide, Methane and Nitrous Oxide Emissions in the United States

摘要

In phase one of this study, only carbon dioxide emissions were addressed. This paper extends the analysis to two other greenhouse gases: methane and nitrous oxide. In addition, updated carbon dioxide emission estimates are provided. The Energy Information Administration (EIA)' publishes an annual report "Emissions of Greenhouse Gases in the United States" and provides a qualitative analysis of the methodology used to produce those estimates. The "good practices" guidelines developed by the Intergovernmental Panel on Climate Change (IPCC), suggest a "Tier 2" procedure for uncertainty analysis to be applied using Monte Carlo simulations. A two-step method is used in this study to include variability and relative uncertainties occurring in the collected data. Monte Carlo techniques are applied to define density functions based on expert opinion. This approach allows for the combination of normal and skewed probability distribution functions (PDFs) among other distributions that reflect, as realistically as possible, the nature of the various parameters. PDFs are developed and implemented for activity data (e.g. energy consumption) and emission coefficients by fuel category and by appropriate sector (e.g. fuels used for electricity generation versus other sectors of the economy). Results from 1990 through 1999 are presented, and trends in the data are provided. Uncertainties are evaluated and represented by resulting standard deviations. A sensitivity analysis is conducted to determine the greatest impacts on uncertainty with regard to the total emission estimates with a focus on effective allocation of agency resources for reducing uncertainties. Further work on the allocated uncertainties as well as a more detailed sensitivity analysis is currently under consideration.