Impact of the High-Emission Event Duration and Sampling Frequency on the Uncertainty in Emission Estimates

Abstract

Short duration remote sensing measurements of methane emissions from oil and gas operations are being deployed at a large scale, but interpretation of these snapshot measurements is complex due to the spatial and temporal variability of methane emissions. The accuracy and precision of annual emission estimates extrapolated from short duration measurements depend on the measurement frequency and complexity of temporal emission patterns. This work examines sampling uncertainties associated with short duration measurements of varying frequencies for methane emissions from a group of sites representing the Barnett Shale region. Routine, frequent emissions are accurately captured with minimal bias through semiannual sampling; however, infrequent high-emission rate events increase the error associated with annual emission estimates, even under the assumption of no measurement uncertainty. If emission events have a duration of ≤1 week, monthly sampling has an estimated sampling error of >15%. For quarterly sampling with emission events that persist for ≤1 month, the sampling error is >30%. There is also negative bias associated with quarterly, semiannual, and annual sampling, which suggests infrequent campaigns may be systemically underestimating emissions. The sampling error increases as the duration of the high-emission events becomes shorter, making the temporal persistence of emission events an important factor in designing measurement protocols.

Assessing Detection Efficiencies for Continuous Methane Emission Monitoring Systems at Oil and Gas Production Sites

“…For example, with monthly sampling, a high emission event could persist for 30 days before being detected by the next series of measurements; in comparison, the continuous monitoring network examined in this work for an idealized site configuration would detect an infinite-duration emission event between 48 and 551 min, on average. Both continuously operating monitoring networks and periodic monitoring with short-duration measurements have the potential to miss short-duration events, but the analysis framework described in this work can be used to correct for nondetections based on dispersion modeling. It would be much more challenging to correct for sampling errors for periodic short-duration measurements.…”

“…Schissel and Allen 14 have assessed the impact of emission intermittency on the ability of annual, semiannual, quarterly, monthly, and weekly measurements to accurately detect emissions, including large, short-duration events. Simulating emissions that were representative of the Barnett Shale production region, they found that routine, relatively continuous emission events were accurately captured with a semiannual sampling frequency.…”

Towards multi-scale measurement-informed methane inventories: reconciling bottom-up inventories with top-down measurements using continuous monitoring systems

“…Drone and aerial measurements have reasonably well characterized quantification accuracy 31,32 but are subject to sampling error, which has been found to introduce negative bias on cumulative emission estimates. 15 CMS provide high frequency measurements, but localization and quantification capabilities are still in development and require additional testing before being fully trusted by industry and regulators. 22 OGI with Hi-Flow for quantification is poorly suited for site-level quantification [24][25][26] but can be used to check if specific components are emitting and differentiate between nearby sources (e.g., as a follow-up to top-down measurements).…”

“…Schissel and Allen 15 quantify the sampling error introduced by top-down measurements, finding that monthly measurements of sites with emissions lasting for one hour result in an average absolute percent error of 23% in annualized emission estimates. This error assumes zero measurement error and is solely a result of infrequent sampling.…”

Role of Intermittency on the Effectiveness of Methane Emissions Detection at Oil and Gas Facilities: Comparative Assessment of Optical Gas Imaging and Site-wide Surveys

“…Because data on the statistics of large emitter events such as their frequency and duration are not publicly available, EPA assumed that these large emitters are persistent in their FEAST model. However, several recent studies have demonstrated the importance of intermittent large emitter events at oil and gas facilities (5,8,19).…”