The first thorough comparison of evidence for natural gas system leaks confirms that organizations including the Environmental Protection Agency (EPA) have underestimated U.S. methane emissions generally, as well as those from the natural gas industry specifically. Natural gas consists predominantly of methane. Even small leaks from the natural gas system are important because methane is a potent greenhouse gas -- about 30 times more potent than carbon dioxide. A study, "Methane Leakage from North American Natural Gas Systems," published in the Feb. 14 issue of the journal Science, synthesizes diverse findings from more than 200 studies ranging in scope from local gas processing plants to total emissions from the United States and Canada.

"People who go out and actually measure methane pretty consistently find more emissions than we expect," said the lead author of the new analysis, Adam Brandt, an assistant professor of energy resources engineering at Stanford University. "Atmospheric tests covering the entire country indicate emissions around 50 percent more than EPA estimates," said Brandt. "And that's a moderate estimate."

The standard approach to estimating total methane emissions is to multiply the amount of methane thought to be emitted by a particular kind of source, such as leaks at natural gas processing plants or belching（嗳气） cattle, by the number of that source type in a region or country. The products are then totaled to estimate all emissions. The EPA does not include natural methane sources, like wetlands and geologic seeps.

The national natural gas infrastructure has a combination of intentional leaks, often for safety purposes, and unintentional emissions, like faulty valves and cracks in pipelines. In the United States, the emission rates of particular gas industry components -- from wells to burner tips -- were established by the EPA in the 1990s.

Since then, many studies have tested gas industry components to determine whether the EPA's emission rates are accurate, and a majority of these have found the EPA's rates too low. The new analysis does not try to attribute percentages of the excess emissions to natural gas, oil, coal, agriculture, landfills, etc., because emission rates for most sources are so uncertain.

Several other studies have used airplanes and towers to measure actual methane in the air, so as to test total estimated emissions. The new analysis, which is authored by researchers from seven universities, several national laboratories and federal government bodies, and other organizations, found these atmospheric studies covering very large areas consistently indicate total U.S. methane emissions of about 25 to 75 percent higher than the EPA estimate.

Some of the difference is accounted for by the EPA's focus on emissions caused by human activity. The EPA excludes natural methane sources like geologic seeps and wetlands, which atmospheric samples unavoidably include. The EPA likewise does not include some emissions caused by human activity, such as abandoned oil and gas wells, because the amounts of associated methane are unknown.

However, the analysis also finds that some recent studies showing very high methane emissions in regions with considerable natural gas infrastructure are not representative of the entire gas system. "If these studies were representative of even 25 percent of the natural gas industry, then that would account for almost all the excess methane noted in continental-scale studies," said a co-author of the study, Eric Kort, an atmospheric science professor at the University of Michigan. "Observations have shown this to be unlikely."