Bias present in US federal agency power plant CO2 emissions data and implications for the US clean power plan

Abstract

Power plants constitute roughly 40% of carbon dioxide (CO2) emissions in the United States. Climate change science, air pollution regulation, and potential carbon trading policies rely on accurate, unbiased quantification of these large point sources. Two US federal agencies - the Department of Energy and the Environmental Protection Agency - tabulate the emissions from US power plants using two different methodological approaches. We have analyzed those two data sets and have found that when averaged over all US facilities, the median percentage difference is less than 3%. However, this small difference masks large, non-Gaussian, positive and negative differences at individual facilities. For example, over the 2001-2009 time period, nearly one-half of the facilities have monthly emission differences that exceed roughly ±6% and one-fifth exceed roughly ±13%. It is currently not possible to assess whether one, or both, of the datasets examined here are responsible for the emissions difference. Differences this large at the individual facility level raise concerns regarding the operationalization of policy within the United States such as the recently announced Clean Power Plan. This policy relies on the achievement of state-level CO2 emission rate targets. When examined at the state-level we find that one-third of the states have differences that exceed 10% of their assigned reduction amount. Such levels of uncertainty raise concerns about the ability of individual states to accurately quantify emission rates in order to meet the regulatory targets.