1 March 2012
When scientists from NOAA and the Cooperative Institute for Research in Environmental Sciences (CIRES) began routinely monitoring the atmosphere's composition at the Boulder Atmospheric Observatory (BAO) tower north of Denver a few years ago, their instruments immediately sniffed something strange: plumes of air rich with chemical pollutants, including the potent greenhouse gas methane.
Some of the pollutants picked up are known to damage air quality. Another, methane, is 25 times more effective per molecule than carbon dioxide at trapping heat in the atmosphere. The scientists were concerned. None of NOAA's other air composition monitoring towers – there are eight, in total, scattered around the continental United States – had recorded anything similar.
"So we set out to understand where these chemicals were coming from, by starting at the tower measurements 1,000 feet high up, down to the ground in a mobile laboratory," said Gabrielle Pétron, an atmospheric scientist with CIRES at ESRL GMD and lead author on a paper published in the Journal of Geophysical Research this week. Petron and co-workers customized air sampling devices and atmospheric chemistry instruments and headed out to northeastern Colorado, downwind of possible sources to collect chemical "fingerprints" that would help identify the possible sources.
After taking dozens of samples and thousands of readings along rural roads, near oil and gas equipment, landfills and animal feeding operations, the research team has an answer: The unusual air pollutants seen at the Denver tower came primarily from oil and gas production in northeastern Colorado's Weld County.
"We found gas operations in the region leaked about twice as much methane into the atmosphere as previously estimated," Petron said. "And the oil and gas infrastructure was leaking other air pollutants, too, including benzene, which is regulated because of its toxicity."
In 2008, the year most of the data were collected, Weld County had nearly 14,000 operating oil and gas wells.
The research team's chemical fingerprinting work showed that oil and gas equipment and the associated activities on well pads – condensate storage tanks, pipelines, compressors and more – leaked or vented an estimated 4 percent of all natural gas produced to the atmosphere. That loss is about double the previous best-guess estimate, based on engineering calculations and industry data, of about 2 percent loss.
"We may have been significantly underestimating methane emissions by this industry in this region," Petron said. The team also found that emissions of benzene, a known carcinogen, are underestimated. Benzene is tracked and regulated by the Environmental Protection Agency (EPA).
Petron and her colleagues found evidence of at least two sources of benzene in the region: oil and gas operations and something else, most likely cars and trucks on roads. And the new study found benzene emissions from oil and gas operations in the region to be significantly higher than expected, between 385 and 2,055 metric tons in 2008, compared with earlier estimates ranging from about 60 to 145 per year.
Finally, the researchers' findings suggest that oil and gas-related emissions of more reactive volatile organic compounds, which contribute to lung-damaging ozone pollution, are also underestimated. More reactive VOCs were not directly measured in the 2008 study, but are almost certainly co-emitted with methane and larger alkanes. According to the EPA, the northern Front Range has been out of compliance with federal health-based standards in the summer since 2007.
Chemist Greg Frost, with CIRES at ESRL CSD and co-author of the new study, said the work demonstrates the value of studying emissions from several perspectives. Top-down studies (such as from the tall tower) can complement and verify bottom-up approaches (such as estimates based on average leak rates at pipe junctions).
"What Gabrielle has done is to use the mobile laboratory and tower data to make top-down estimates of emissions, which can be used to evaluate the bottom-up estimates from industry and regulatory agencies," Frost said. "This is going to inspire a lot more research."
G. Pétron, G. Frost, B. R. Miller, A. I. Hirsch, S. A. Montzka, A. Karion, M. Trainer, C. Sweeney, A. E. Andrews, L. Miller, J. Kofler, A. Bar-Ilan, E. J. Dlugokencky, L. Patrick, C. T. Moore Jr., T. B. Ryerson, C. Siso, W. Kolodzey, P. M. Lang, T. Conway, P. Novelli, K. Masarie, B. Hall, D. Guenther, D. Kitzis, J. Miller, D. Welsh, D. Wolfe, W. Neff, and P. Tans, Hydrocarbon emissions characterization in the Colorado Front Range: A pilot study, Journal of Geophysical Research, doi:10.1029/2011JD016360, 2012.
The multispecies analysis of daily air samples collected at the NOAA Boulder Atmospheric Observatory (BAO) in Weld County in northeastern Colorado since 2007 shows highly correlated alkane enhancements caused by a regionally distributed mix of sources in the Denver-Julesburg Basin. To further characterize the emissions of methane and non-methane hydrocarbons (propane, n-butane, i-pentane, n-pentane and benzene) around BAO, a pilot study involving automobile-based surveys was carried out during the summer of 2008. A mix of venting emissions (leaks) of raw natural gas and flashing emissions from condensate storage tanks can explain the alkane ratios we observe in air masses impacted by oil and gas operations in northeastern Colorado. Using the WRAP Phase III inventory of total volatile organic compound (VOC) emissions from oil and gas exploration, production and processing, together with flashing and venting emission speciation profiles provided by State agencies or the oil and gas industry, we derive a range of bottom-up speciated emissions for Weld County in 2008. We use the observed ambient molar ratios and flashing and venting emissions data to calculate top-down scenarios for the amount of natural gas leaked to the atmosphere and the associated methane and non-methane emissions. Our analysis suggests that the emissions of the species we measured are most likely underestimated in current inventories and that the uncertainties attached to these estimates can be as high as a factor of two.