Inversion Estimates of Lognormally Distributed Methane Emission Rates From the Haynesville‐Bossier Oil and Gas Production Region Using Airborne Measurements. Issue 6 (18th March 2019)

• Record Type: Journal Article
• Authors: Cui, Yu Yan; Henze, Daven K.; Brioude, Jerome; Angevine, Wayne M.; Liu, Zhen; Bousserez, Nicolas; Guerrette, Jonathan; McKeen, Stuart A.; Peischl, Jeff; Yuan, Bin; Ryerson, Thomas; Frost, Gregory; Trainer, Michael
• Is Part Of: Journal of geophysical research. Volume 124:Issue 6(2019)
• Page Start: 3520
• Page End: 3531
• Publication Date: 2019-03-18
• Subjects: methane -- oil and gas production -- inverse modeling -- SENEX; Atmospheric physics -- Periodicals; Geophysics -- Periodicals; 551.5
• DOI: 10.1029/2018JD029489
• Languages: English
• ISSNs: 2169-897X

Abstract

Abstract: Quantifying methane (CH4 ) emissions from the oil and natural gas (O/NG) production sector is an important regulatory challenge in the United States. In this study, we conduct a set of inversion calculations using different methods to quantify lognormal distributed CH4 surface fluxes in the Haynesville‐Bossier O/NG production basin in Texas and Louisiana, combining three statistical cost functions, four meteorological configurations, and two days of aircraft measurements from a 2013 field campaign. We aggregate our posterior flux estimates to derive our best estimate of the basin‐wide CH4 emissions, 76 metric tons/hr, with a 95% highest density interval of 51–104 metric tons/hr, in agreement with previous estimates using mass balance and eddy covariance approaches with the same aircraft measurements. Our inversion estimate of basin‐wide CH4 emissions is 133% (89%–182%, 95% highest density interval) of a gridded Environmental Protection Agency's inventory for 2012, and the largest discrepancies between our study and this inventory are located in the northeastern quadrant of the basin containing active unconventional O/NG wells. Our inversion approach provides a new spatiotemporal characterization of CH4 emissions in this O/NG production region and shows the usefulness of inverse modeling for improving O/NG CH4 emission estimates. Plain Language Summary: Oil and natural gas (O/NG)‐related methane (CH4 ) emission estimates have drawn great concern because activity in this industry has increased dramatically over the past decade. However, estimating CH4 emissions from O/NG production regions is very challenging because the emission rates are highly heterogeneous. To properly characterize the CH4 emissions in the Haynesville oil and gas production region, we develop an inverse modeling system to handle different ways of characterizing the highly skewed (i.e., lognormally distributed) CH4 sources in Haynesville. The inverse model calculations are driven by high‐frequency, high‐precision CH4 mixing ratios measured on a National Oceanic and Atmospheric Administration aircraft during a field study in the summer of 2013. We use a variety of meteorological simulations to define the transport errors in our inversions, and we take advantage of a resampling method to characterize the posterior uncertainties. Our results suggest that Haynesville's CH4 emissions are likely underestimated in the U.S. Environmental Protection Agency's national CH4 inventory and particularly in the subdomain where many active unconventional wells are located; day‐to‐day variability in Haynesville's overall CH4 emissions likely exists. Our work offers an extensive characterization of inversions' uncertainties and demonstrate the feasibility of improving CH4 emission estimates from O/NG production regions using high quality aircraft observations. Key Points: A set of inversion calculations provide estimates of lognormally distributed methane emissions that are confirmed by two other top‐down methods Multiple meteorological configurations and inversion approaches are used to capture transport errors and inversion biases The best estimate from our inversions shows that CH4 emissions are likely underestimated by an EPA inventory for this oil and natural gas basin