Impact of Emissions Estimation Uncertainty on Methane Monitoring Reporting and Verification (MMRV) Programs

Recent emphasis on decreasing methane emissions from oil and gas production and transport has stimulated the development of multiple regulatory and voluntary reporting programs. These programs include monitoring and verification requirements, commonly known as MMRV (methane monitoring, reporting and verification). An underlying assumption in these programs is the use of advanced methods to estimate emissions, including continuously installed sensors at facilities and aerial, satellite and driving survey methods. These methods provide emission estimates at the scale of major sources or entire facilities. Regulatory programs in the USA and EU are increasingly tying these estimates to substantial financial risks while encouraging anonymous 3rd party measurements, raising the stakes for using these methods. This abstract reviews recent studies of these methods, and reviews four areas of concern. First, multiple commonly used methods display accuracy problems which are likely to be present in most methods. Recent studies study assessed methods at production and midstream facilities onshore in the USA. Two methods deployed simultaneously at 14 midstream facilities disagreed by 2:1 averaged across all facilities and by more than 2:1 at 6 of the facilities. Other studies in U.S.A production and European midstream have identified similar accuracy issues. Second, measurement informed inventory methods, which use full-facility estimates to update emissions reporting, remain poorly developed and unevenly implemented. While one study found that survey methods identified large emitters and operators corrected reporting, another study found that most aerial detections did not result in effective corrections to inventory estimates. Third, methods used to extrapolate facility-scale estimates to basin scale have unaddressed uncertainties. Recent work indicates that 9-49% of plumes detected by aircraft methods are due to maintenance emissions, which are poorly characterized by anonymous aerial sampling. Additionally, extrapolation methods poorly estimate short emission events, resulting in a significant potential to over-estimate of emissions. Conversely, random non-detects of exhaust emissions likely under-estimates emissions from engines and combustors by a factor of 2 or more. These errors both shift emission between sectors and may result in significant bias. Additional control inputs, better GIS data, and improved methods are required to better estimate regional emissions. Finally, recent studies of continuous emissions monitors in both controlled and field tests indicate poor quantification accuracy in controlled testing, and poorer accuracy in field conditions. While advanced methods show promise for improving emissions detection and mitigation, consumers of these data need to be aware of the performance of these methods and account for bias, uncertainty, and variability of emissions estimates when constructing programs that utilize these estimates.