The analysis also said that methane discharges in Texas and Oklahoma, where oil and gas production was concentrated at the time, were 2.7 times greater than conventional estimates. Emissions from oil and gas activity alone could be five times greater than the prevailing estimate, the report said.
The study relies on nearly 12,700 measurements of atmospheric methane in 2007 and 2008. Its conclusions are sharply at odds with the two most comprehensive estimates of methane emissions, by the Environmental Protection Agency and an alliance of the Netherlands and the European Commission.
The E.P.A. has stated that all emissions of methane, from both man-made and natural sources, have been slowly but steadily declining since the mid-1990s. In April, the agency reduced its estimate of methane discharges from 1990 through 2010 by 8 to 12 percent, largely citing sharp decreases in discharges from gas production and transmission, landfills and coal mines.
The new analysis calls that reduction into question, saying that two sources of methane emissions in particular — from oil and gas production and from cattle and other livestock — appear to have been markedly larger than the E.P.A. estimated during 2007 and 2008.
One of the study’s principal authors, Scot M. Miller of Harvard University’s department of earth and planetary sciences, said its higher estimates underscore methane’s significant contribution to rising temperatures.
“These are pretty substantial numbers we’re dealing with, and an important part of greenhouse gas emissions,” he said on Monday. “Our study shows that there could be large greenhouse gas emissions in places in the country where we may not necessarily have accounted for them.”
Methane made up only about 9 percent of greenhouse gas emissions in 2011, the E.P.A. said; carbon dioxide is easily the most prevalent gas. But methane is much more potent. Even though it rapidly breaks down in the atmosphere, its contribution to global warming is 21 times greater than carbon dioxide’s over a 100-year period.
The E.P.A. and Europe’s Emissions Database for Global Atmospheric Research largely agree on how much methane is discharged annually in the United States. At the most basic level, both arrive at estimates by assigning an average discharge to each category of methane emission, such as landfills, and multiplying the average by the number of sources in each category.
The latest analysis differs from those estimates because it relies on actual measurement of methane concentrations. Nearly 5,000 air samples were collected from 10 huge communications towers spread across the country — some on mountaintops, others more than 1,000 feet high — and some 7,700 more from an aircraft monitoring program, both programs run by the National Oceanic and Atmospheric Administration and the Department of Energy.
The data did not directly identify the sources of methane discharges. But the researchers were able to infer those sources through a range of techniques. In areas associated with oil and gas production, for example, the amount of airborne methane could be correlated with measurements of propane, another gas that serves as a sort of marker for oil and gas activity.
The study concluded that livestock produced roughly twice as much methane during the reporting period as the European database estimated. Most striking, the analysis reported that oil and gas operations in a north-south swath of Kansas, Oklahoma and Texas may have produced five times more methane — and, combining all sources of discharge, the three states may have been responsible for a quarter of all man-made methane discharges in the United States.
Mr. Miller cautioned that both estimates were subject to large margins of uncertainty; the methane from oil and gas activity could be as small as 2.3 times the European estimates, or as great as 7.5 times. The reason, he said, is that the potential for inaccuracy rises as the area being surveyed or the category of emissions grows smaller.
The same caveat applies to the few regions where the study found that methane discharges were smaller than European estimates: the Appalachian coal belt, southern Illinois and western Kentucky, and New York City, for example. Some of those spots were also in areas where monitoring of airborne methane was infrequent or absent.
That said, the study’s overall conclusion that methane emissions were 1.5 times E.P.A.’s latest estimates is statistically accurate to within about 5 percent, Mr. Miller said.