The Environmental Protection Agency took the first steps toward better understanding what has been a relatively inexact science to this point -- calculating the amount of methane that is released into the atmosphere from hydraulic fracturing, or fracking, sites.

The EPA released five technical papers on Tuesday assessing methane leaks during the fracking process, the compressors and pneumatic devices used in the oil-and-gas drilling method, and the ongoing emissions from completed wells. The agency now will receive feedback on those papers to help determine which actions should be taken to cut emissions of methane, the biggest unregulated greenhouse gas in the U.S.

The rate at which methane is released at fracking sites is a key to the climate change debate surrounding an energy boom that has yielded abundant supplies of natural gas. Researchers are growing increasingly interested in tracking down rare, but large, emissions events that they say account for the sizable difference in methane leak rates reported by the EPA versus outside studies.

"The challenge is, how can we be confident that we are capturing those rarer events well?" asked Steve Hamburg, chief scientist at the Environmental Defense Fund.

The EPA papers rely on a growing body of work regarding emissions of methane, a short-lived pollutant that has 21 times the climate effect of carbon dioxide and accounts for 9 percent of the nation's greenhouse gas emissions. It's part of a White House strategy outlined last month to reduce methane emissions, which could lead to new regulations on fracking.

Natural gas burns cleaner than coal, and its recent affordability has pushed electric utilities to use more of it -- the EPA said this week that such switching helped push carbon emissions down 3.4 percent in 2012 from 2011. But environmental groups say the climate-change benefits of natural gas could be erased if enough methane leaks during the fracking process.

Recent studies have shown wide variations in estimates of methane emissions from fracking sites, some of which suggest the EPA has significantly low-balled its figures. The reasons, though, are fairly simple: Until recently, there wasn't much data; and, even now, there are different approaches for estimating those emissions.

A Purdue University-led study published Monday in the Proceedings of the National Academy of Sciences surveyed a portion of the Marcellus Shale formation in southwest Pennsylvania and found methane was being released at a rate between 100 and 1,000 times greater than EPA estimates. A Harvard University-led study, published in November in the same journal, said the EPA short-changed emissions by about 50 percent.

Katie Brown, a spokeswoman with industry group Energy in Depth, said the Purdue study was questionable because it relied on work by a pair of scientists who she labeled anti-fracking activists.

Much of the difference is owed to methodology, scientists said.

The EPA uses a "bottom-up" approach to calculate methane leaks, which involves taking measurements of component parts directly from individual fracking wells.

That tactic allows the EPA to more efficiently account for sector-by-sector emissions, which helps inform policymakers, noted David McCabe, an atmospheric scientist with the Clean Air Task Force. That allowed the agency to say last April, for example, that oil-and-gas sector methane emissions fell an annual average of 41.6 million metric tons between 1990 and 2010, resulting in a 20 percent drop from earlier EPA estimates.

A joint Environmental Defense Fund and University of Texas study, which Brown touted, took that bottom-up approach. It found a methane leakage rate of 1.5 percent and an emissions estimate "comparable" with EPA's numbers. Environmentalists say 3 percent would erase the climate benefit of natural gas.

But some experts argue the bottom-up methodology might not capture rare, but large, events.

The fracking industry has a wide array of companies and players, many of which have been willing to participate in studies. Those companies, however, might not be representative of the whole, McCabe said. On top of that, sample sizes are usually small — larger ones would be too expensive for most studies.

"Statistically, there's a real chance of undercounting," McCabe said. "There are very clear systematic reasons that bottom-up studies are always going to underestimate emissions."

That's why studies that take a "top-down" approach have shown much higher emissions than what EPA has reported.

By measuring emissions from an elevated plane, researchers are able to grab a broader swath of information that might capture some of those high-emission events. Such was the case with the Harvard and Purdue studies, though they differed in sample size (the Harvard study covered the entire U.S., the Purdue one just a portion).

Still, each approach leaves holes and creates room for disparities, Hamburg said.

Extrapolating the emissions from a small set of wells across the 500,000 operating in the U.S. could exceed or undercut the total emissions estimates — a useful snapshot, perhaps, but not a panorama. Similarly, scaling down from a broader view might fail to account for regions or sites that produce significantly higher or lower emissions.

That's why the Environmental Defense Fund spearheaded a two-week study in October of Texas' Barnett Shale. It pulled together 12 teams that sampled direct measurements from wells and used specially equipped airplanes to gauge atmospheric methane levels — the goal is to track down some of the rare but high-emission events that have eluded researchers. The study's participants are meeting Friday to go over the data and plan to publish the results next year.

"It's together that the power is really there," Hamburg said of the two approaches.