Research Projects

Uncertainties in both individual emission estimates and a compiled emission inventory as a whole are not well characterized. Since most emission estimates are not based on repeated measurements of the quantities, the conventional statistical formula for a confidence interval around the estimated mean of reported measurements does not apply. Therefore, an alternative formulation for quantifying uncertainty of an estimated quantity must be devised. To this end, an extensive review of earlier uncertainty studies on emission estimates and other studies on subjectively evaluated uncertainties has been conducted.

The review revealed that all earlier studies failed to address bias uncertainties which are considered to be very important for quantifying overall uncertainty in emission estimates, because of the use of models and engineering analyses in those estimates instead of repeated measurements. Another finding is that there exists no established method of quantifying subjectively evaluated uncertainties in emission estimates. The absence of an adequate method for quantifying uncertainties in emission estimates led the authors to explore and develop a new method that will facilitate assessing uncertainties in individual emission estimates and in emission inventory as a whole. In exploring such a method, the necessary attributes for an "Ideal Method" were first identified. Although an "Ideal Method" does not exists, its attributes are definable from the lessons learned from the failures and flaws found in the earlier studies.