The reckoning: How observation data is upending our understanding of methane emissions

The oil and gas sector is at the heart of global efforts to reduce methane emissions. Companies, regulators and investors agree that methane leakage, outside of safety measures, rarely benefits anyone. At the same time, finding leaks can be painstaking and costly, and oil and gas-producing assets exhibit significant variations in emission rates due to operational practices, equipment configuration, infrastructure access and other variables.

Given rising climate change concerns, the potency of methane’s warming potential — 28 times more than CO₂ on a 100-year basis and 84 times more on a 20-year basis — has made it the priority for mitigation in the oil and gas sector, which has concentrated sources of methane and viable pathways for selling it.

Over the past decade, oil and gas companies have made significant strides in addressing methane emissions. They have set goals such as the Global Methane Pledge, built teams of experts and piloted an array of technologies to monitor this invisible substance. Importantly, they began systematically estimating and reporting greenhouse gas emissions to various authorities.

Methane emissions reported to the US Environmental Protection Agency since 2012 form the most important record available, especially as US gas supply and LNG exports have surged. When the record was promulgated, the EPA sought to create a straightforward system that all companies could implement with available information. A proscriptive and rigid system was necessary to ensure comparable results and feasible calculations. Unfortunately, unlike CO₂, methane emissions result from sporadic operations or unintentional releases due to malfunction. No formula can reliably forecast the rate or volume of these releases.

Producers have since honed internal processes to report annually, but even this simplified method requires significant resources. Furthermore, the public record that the EPA’s system generated has had little value to companies beyond compliance.

Alongside this regulatory time series, technologies to directly detect methane have rapidly matured. Companies have migrated from generic factors to observation, augmenting internal datasets exponentially and providing new, relevant information. Appearing more frequently in company sustainability reports, studies and third-party releases — including S&P Global Commodity Insights benchmarks — this data paints a different picture of emission levels. Critically, and controversially, the observation data almost always exceeds corresponding figures from the mandated, static calculations. This begs the question: Who is right?

The truth is that no one knows exact methane emissions outside of laboratory settings. Assumptions, calculation and judgment still retain a significant role in even the best observed figures. And for sizable oil and gas assets over a notable period, it is not cost-efficient to obtain precision. If confronted with five estimates of emissions for a facility, it is helpful to posit the existence of a sixth — the actual number.

S&P Global Commodity Insights has extensive expertise in emissions across the energy chain. The S&P Global Commodity Insights Center for Emissions Excellence partners with data scientists and domain experts to evaluate data from external sources as well as proprietary estimates derived from core energy databases. Strengths and weaknesses exist in every quantification approach, including remote sensing using satellites and overflights. The margins of error for remote sensing remain wide at +/- 50% for any individual leak estimate. However, regular calibration of remote sensing against known releases at test facilities has allowed improvement that provides meaningful certainty. When interpreted properly and placed in the right context, we find that estimates derived from observation are consistently and significantly more credible than the methane emissions historically calculated for regulatory purposes. They are in the ballpark of actual emissions.

Methane emissions reported to the US Environmental Protection Agency since 2012 form the most important record available, especially as US gas supply and LNG exports have surged. When the record was promulgated, the EPA sought to create a straightforward system that all companies could implement with available information. A proscriptive and rigid system was necessary to ensure comparable results and feasible calculations. Unfortunately, unlike CO₂, methane emissions result from sporadic operations or unintentional releases due to malfunction. No formula can reliably forecast the rate or volume of these releases.

Producers have since honed internal processes to report annually, but even this simplified method requires significant resources. Furthermore, the public record that the EPA’s system generated has had little value to companies beyond compliance.

Alongside this regulatory time series, technologies to directly detect methane have rapidly matured. Companies have migrated from generic factors to observation, augmenting internal datasets exponentially and providing new, relevant information. Appearing more frequently in company sustainability reports, studies and third-party releases — including S&P Global Commodity Insights benchmarks — this data paints a different picture of emission levels. Critically, and controversially, the observation data almost always exceeds corresponding figures from the mandated, static calculations. This begs the question: Who is right?

The Permian Basin offers an illuminating case study to contrast the S&P Global methane benchmarks, built with partner Insight M, based on remote sensing data versus regulatory data submitted to the EPA. It is critical to note that basin-level aggregates do not apply to individual companies. A handful of operators even have actual emissions below the figure reported to the EPA. Most companies have the reverse, of course, but as in any system as diverse as the Permian, performance ranges widely and distributions matter.

The sum of methane emissions submitted to the EPA for 2022 in the Permian production segment, grossed to account for the small portion of assets not reporting, was 8.3 Bcf. Expressed as a percentage of the total barrels of oil equivalent that the basin produced, this represented an intensity of about 0.06%, rounded to the nearest hundredth. In 2023, the EPA figure increased to about 0.07% using the same rounding, a sharp rise of 26%. Both years registered an intensity below 0.2% of output, suggesting relatively low, if deteriorating, methane emission operations in the world’s biggest oil field.

Meanwhile, observed data tells a different story. The S&P Global benchmark for upstream Permian methane emissions intensity is 0.92% of barrels of oil equivalent for 2022 and 0.63% for 2023, an impressive 32% improvement. While the emissions remain large, the absolute decrease in 2023 Permian Basin upstream methane emissions — 34 Bcf — is huge. Because methane is such a potent GHG, the equivalent amount of the 18.5 million metric tons of CO₂ avoided is also outsized. In fact, the decrease in CO₂ equivalent was more than the total 2023 driving emissions avoided by every electric vehicle ever sold in the US, even if fully powered with zero-carbon electricity.

As argued above, the EPA’s proscribed methodology underestimates actual emissions for many companies. Indeed, the EPA’s proposed changes to its methodology will mostly work to close the gap with observed volumes. But for historical purposes, the observed data was roughly 15 times the level of submissions to the US government in 2022 and 8 times as large in 2023.

Thus, observation data is resetting the baseline much higher than the regulatory record, but it also shows the industry is making significant progress as it ramps up spending and deploys technology. Contributing to the public debate about methane emissions, some oil and gas producers view adopting observation-informed measures as a large setback in communicating their current emission levels and achieving their methane mitigation goals.

There are payoffs, however, for companies that leverage these new technologies and use observed estimates sooner rather than later. First, unlike static calculations, direct observation provides companies with critical data they can use to take concrete action to reduce methane and carry out root-cause analysis — it is difficult to fix a leak you do not know exists. Also, companies face mounting reputational risk as the proliferation of third-party detection, including by environmental groups, provides a new level of public transparency regarding methane emissions. It may be more prudent to recalibrate sooner and proactively control that shift rather than accept the risk of involuntary, third-party exposure.

Finally, the gain in data credibility from adopting observation-based numbers can offer a critical counterbalance to acknowledging higher emission levels. The global methane discussion today simplistically ranks performance using data of vastly different qualities. Insightful analysis should consider the spectrum of high versus low intensity tied to an assessment of the data quality. To this end, S&P Global Commodity Insights has developed and publicized a transparent scheme that scores the reliability of emissions data.

Using the Permian Basin as an example, moving to observation-informed data from reported data dramatically increases the level of emissions intensity. However, it also catapults the reliability score from a very low confidence level to much higher confidence. The regulatory data makes the basin look quite clean but is inherently unreliable. In other words, we do not know if the estimate can be trusted.

In contrast, in the Permian Basin example, the observation-informed estimates show worse performance but are more trustworthy. This has benefits as publicized third-party estimates are likely to corroborate and bolster, rather than contradict, the estimates of the operator. In addition, by including data quality, companies can provide a means to distinguish true performance over time and across companies and jurisdictions. Investors, in particular, can choose which they prefer: a cleaner but doubtful number or a higher-intensity, certain performance that they can track. In our conversations, most investors indicate a preference for the latter.

The vast expansion of data from remote emissions detection should accelerate reductions to global methane emissions from oil and gas operations. Armed with semi-real-time information, companies can address the leaks — especially the largest leaks — that previously may have lasted for weeks or months. We believe that producers will mostly meet their 2030 goals for methane emissions reduction from operated assets. Furthermore, as investors and governments come to trust the more accurate observation-informed estimates, they will likely demand and reward the greater transparency that such estimates provide. Over the next decade, the current clash of methane emissions estimates in the public and corporate spheres should give way to a convergence in the data, moving toward an accurate and shrinking emissions assessment.