US CH4 Emissions from Oil and Gas Production: Have Recent Large Increases Been Detected?

Recent studies have proposed significant increases in methane (CH4) emissions from North America over the past decade, and suggested that rapid growth in U.S. oil and gas production could be the cause. Evidence for this is based on (1) observed increases in co-emitted species such as ethane and propane, (2) a trend derived from atmospheric inversions, and (3) spatial differences across North America derived from space-based retrievals of column CH4 abundance (XCH4). We examine these claims using an ensemble of time-dependent inversions collected as part of the Global Carbon Project and best available bottom-up estimates, and we consider what long-term observations from the NOAA aircraft program tell us about U.S. emissions. We find that none of the time-dependent inversions estimate large trends in U.S. emissions, and this is true for inversions using only surface observations and for those that use XCH4. We find that short term (< 5 year) trends of up to 1.5 ppb CH4/yr can occur in spatial gradients between the Pacific "background" CH4 and continental locations due to transport variability, and that the trends in spatial differences are very sensitive to the background value used. In addition, samples must be collected frequently enough to avoid spurious trends, and located near enough to sources to be sensitive to emissions. Space-based retrievals may provide enough spatiotemporal coverage at lower latitudes, but we find that use of vertical gradients from aircraft observations increases detectability of trends. The magnitude of atmospheric variability is large in relation to the relatively small emission signal (US O&G emissions are only about 2% of global total CH4 emissions) and this makes detection of trends over short periods difficult, yet early detection of emission trends is needed for climate agreements such as the COP21. Atmospheric modeling in principle accounts for transport variability, and may be helpful in detecting trends, especially if observation density and frequency increases.