This reference compares global fuel estimates with a matched historical US study/inventory comparison and a later regional aerial study. It explains leakage denominators, confidence intervals, repeated surveys, methane versus flaring, and technical versus economic abatement. The result is a reusable source map: a writer can see what a number measures before deciding whether it supports a US, global, upstream or full-supply-chain claim. A publication date is kept separate from the period when the emissions were observed.
Key Takeaways
- 124 Mt/year is the IEA estimate for fossil fuel operations in 2025. (IEA 2026; accessed, Nov 2026).
- Oil contributes 45 Mt, coal 43 Mt and natural gas 36 Mt in that estimate. (IEA 2026; accessed, Nov 2026).
- Oil and gas together total 81 Mt/year, an AMR sum of the two reported categories. (IEA 2026; accessed, Nov 2026).
- Around 35% of human-caused methane is attributed to the fossil fuel sector by IEA. This denominator is different from fossil-fuel emissions alone. (IEA 2026; accessed, Nov 2026).
- 13 Tg/year versus 8.1 Tg/year compares the 2015 US study estimate with the inventory comparator reproduced in its Table 1. (Alvarez et al., 2018; accessed, Nov 2026).
- 4.9 Tg/year is the difference between displayed national central values, not a new independently measured quantity. (Alvarez et al., Table 1; accessed, Nov 2026).
- Production is 7.6 versus 3.5 Tg/year in the historical matched segment comparison. (Alvarez et al., Table 1; accessed, Nov 2026).
- 2.3% refers to gross US natural gas production in that historical study, not an energy-normalized global upstream rate. (Alvarez et al., 2018; accessed, Nov 2026).
- 2.43% is the six-region weighted result reported by Sherwin et al., with a 95% confidence interval of 2.27% to 2.59%. (Sherwin et al., 2024; accessed, Nov 2026).
- 52% of US onshore oil and 29% of gas production were covered by the regions in that paper, not the whole US supply chain. (Sherwin et al., abstract; accessed, Nov 2026).
- 15 aerial campaigns covered six regions, including repeat visits to some regions. Campaign totals must not be summed as unique national coverage. (Sherwin et al., supplement Table S3; accessed, Nov 2026).
- Around 70% of fossil methane is technically abatable with existing technology, while the no-net-cost quantity uses a separate economic calculation. (IEA 2026; accessed, Nov 2026).
How much methane comes from oil, gas and coal?
The IEA’s 2025 estimates put oil ahead of coal and natural gas as individual fossil-fuel categories. The three reported central values add to 124 Mt/year. AMR’s shares use that explicit denominator: 45 / 124 for oil, 43 / 124 for coal and 36 / 124 for natural gas. (IEA 2026, key findings; accessed, Nov 2026).
View chart data
2025 estimates, Mt methane/year; published total 124 Mt.
| Fuel | Mt/year |
|---|---|
| Oil | 45 |
| Coal | 43 |
| Natural gas | 36 |
Values and definitions are also readable in HTML. Source precision and scope apply.
| Fuel | 2025 emissions estimate (Mt/year) | Share of published 124 Mt total (%) |
|---|---|---|
| Oil | 45 | 36.2903 |
| Coal | 43 | 34.6774 |
| Natural gas | 36 | 29.0323 |
Why is oil associated with methane?
The classification follows the producing system, not the chemical identity of the main marketed product. Oil operations can produce associated natural gas and release methane through equipment, venting or incomplete combustion. Describing all fossil methane as emissions from the natural-gas industry would therefore erase the oil and coal categories. Conversely, describing the whole 124 Mt total as oil and gas methane would incorrectly include coal.
What does the 35% share mean?
IEA’s approximate 35% statement refers to fossil fuel operations as a share of methane from human activity. AMR’s 36.29% oil share refers to oil within the 124 Mt fossil-fuel total. Similar-looking percentages can answer different questions. A citation should retain the denominator immediately beside the statistic rather than rely on surrounding context.
Does this measure every abandoned well?
The tracker distinguishes operational sectors, large detected events and abandoned facilities in its documentation and chart notes. Do not attach an assumed abandoned-well total to these rounded fuel values or infer it from a chart category. The 124 Mt panel reproduces the aggregate categories as stated in the source. For abandoned-well counts and program obligations, use AMR’s separate orphan-well research rather than treating a count as an emissions measurement.
How did a US measurement study compare with the official inventory?
Alvarez and colleagues estimated 2015 US oil and gas supply-chain methane at 13 Tg/year. Their Table 1 reproduces an EPA inventory comparator of 8.1 Tg/year. AMR’s subtraction of displayed central values is 4.9 Tg/year; the quotient gives about 60.49% above the displayed inventory value. The paper describes the difference approximately in its abstract and uses a different rounded percentage in its body. AMR does not substitute its rounded-value arithmetic for the authors’ underlying analysis. (Alvarez et al., 2018, Table 1 and text; accessed, Nov 2026).
View chart data
2015 emissions, Tg methane/year; study and inventory values from the same table.
| Segment | Study | Inventory |
|---|---|---|
| Production | 7.6 | 3.5 |
| Gathering | 2.6 | 2.3 |
| Processing | 0.72 | 0.44 |
| Transmission/storage | 1.8 | 1.4 |
Values and definitions are also readable in HTML. Source precision and scope apply.
| Segment | Study central estimate (Tg/year) | 95% upper increment | 95% lower decrement | EPA inventory comparator (Tg/year) | Difference in displayed central values (Tg/year) | Qualification |
|---|---|---|---|---|---|---|
| Production | 7.6 | 1.9 | 1.6 | 3.5 | 4.1 | 2015 emissions as reported in 2018 paper |
| Gathering | 2.6 | 0.59 | 0.18 | 2.3 | 0.3 | 2015 emissions as reported in 2018 paper |
| Processing | 0.72 | 0.2 | 0.071 | 0.44 | 0.28 | 2015 emissions as reported in 2018 paper |
| Transmission and storage | 1.8 | 0.35 | 0.22 | 1.4 | 0.4 | 2015 emissions as reported in 2018 paper |
| Local distribution | 0.44 | 0.51 | 0.22 | 0.44 | 0.0 | Taken directly from inventory, not independent validation |
| Oil refining and transportation | 0.034 | 0.05 | 0.008 | 0.034 | 0.0 | Taken directly from inventory, not independent validation |
| Published national total | 13 | 2.1 | 1.7 | 8.1 | 4.9 | Published Table 1 aggregate; rounded components do not sum exactly |
Which segment accounts for the largest difference?
Production is 7.6 Tg/year in the study versus 3.5 Tg/year in the inventory comparator, a displayed-value difference of 4.1 Tg/year. Gathering is 2.6 versus 2.3, processing 0.72 versus 0.44, and transmission/storage 1.8 versus 1.4. The production comparison therefore identifies the largest absolute discrepancy among these listed segments. It does not establish that the same distribution holds today.
Why do the two remaining rows match?
The local-distribution and oil-refining/transportation study estimates were taken directly from the inventory. Their matching values are not independent evidence that the inventory was accurate in those segments. The paper also excludes losses downstream of customer meters from its distribution estimate and does not claim a complete update of end use. A full-supply-chain label needs to preserve these qualifications.
Why do the component sums differ from the total?
The displayed segment values are rounded. Adding the displayed study components gives 13.194 Tg/year, while the published aggregate is 13 Tg/year. Adding inventory components gives 8.114 Tg/year, while its published total is 8.1. The aggregate is the source’s result, not an error to be “fixed” by replacing it with a sum of rounded entries. Downloads retain both the displayed values and that explanation.
What is a methane leakage rate actually divided by?
A percentage is incomplete without a denominator and system boundary. The 2018 national study normalizes its estimate by gross US natural gas production, with an average methane-content assumption. IEA’s upstream intensity uses fuel supply in energy terms and a 55 MJ/kg methane assumption. These are not interchangeable measures. (Alvarez et al., text following Table 1; accessed, Nov 2026). (IEA 2026, intensity chart notes; accessed, Nov 2026).
View chart data
These published percentages cannot establish a national time trend.
| Evidence | Rate | Scope |
|---|---|---|
| 2018 study | 2.3% | 2015 US gross gas production |
| 2024 study | 2.43% | Six covered US regions |
| IEA intensity | Energy based | Upstream / fuel supply energy |
Values and definitions are also readable in HTML. Source precision and scope apply.
| Evidence | Publication | Geography | Reported rate | Denominator and scope |
|---|---|---|---|---|
| Alvarez et al. | 2018 | National US supply chain, 2015 | 2.3% | Gross US gas production, 90% methane content, oil and gas supply chain |
| Sherwin et al. | 2024 | Six US regions over 15 campaigns | 2.43% (95% CI 2.27% to 2.59%) | Covered natural gas production; sample-weighted, not nationwide |
| Sherwin et al., low covered region | 2024 | Regional campaign sample | 0.71% (95% CI 0.61% to 0.81%) | Covered gas production; not a national minimum |
| Sherwin et al., high covered region | 2024 | Regional campaign sample | 9.77% (95% CI 9.19% to 10.45%) | Covered gas production; not a national maximum |
| IEA upstream intensity | 2026 | Cross-country oil and gas context | Energy-based measure | Upstream methane / fuel supply energy; assumes 55 MJ/kg methane |
Can 2.3% and 2.43% be used as a US trend?
No. Sherwin and colleagues’ 2.43% is a weighted result across six surveyed regions, not an updated nationwide version of the 2015 inventory comparison. Its reported confidence interval is 2.27% to 2.59%. The regions account for 52% of onshore oil production and 29% of gas production in the paper’s coverage statement. Geographic selection, observation campaigns, production weights and detection methods differ. Subtracting the two percentages would create an apparent national change unsupported by a matched design. (Sherwin et al., abstract and supplement; accessed, Nov 2026).
What do the lowest and highest regional rates mean?
The reported regional range is 0.71% to 9.77%, with different confidence intervals and operating contexts. They are outcomes for covered regions in the study, not a verified minimum and maximum across every US basin. A study focused on oil-producing regions can show a high rate relative to the natural-gas denominator because the system produces oil as well as gas. Changing allocation between products changes the interpretation; it does not change the physical mass emitted.
Which unit should be used?
Tg means teragram and Mt means million metric tons. For methane mass, one Tg equals one Mt. A billion cubic meters of natural gas is a volume of a gas mixture, not a million tons of methane. Converting between them requires composition, reference temperature and pressure or an explicitly defined density. The article keeps mass, gas volume, production-normalized rates and energy-normalized intensity in separate fields.
How do measurements, sampling and super-emitters affect an estimate?
A regional survey and a facility inventory observe different aspects of the system. Alvarez et al. combine facility-scale work with comparisons to regional observations. Sherwin et al. integrate approximately one million aerial site measurements from 15 campaigns, combining detected sources with an approach for smaller emissions. The existence of many measurements does not by itself remove geographic or detection bias. (Alvarez et al., methods discussion; accessed, Nov 2026). (Sherwin et al., abstract and supplementary methods; accessed, Nov 2026).
View chart data
Selected campaigns from supplementary Table S3; unique wells within each campaign.
| Campaign | Period | Wells |
|---|---|---|
| CM Permian 2019 | Sep 22 to Nov 3, 2019 | 84570 |
| CM Permian 2020 | Jul 13 to 24, 2020 | 17120 |
| CM Permian summer 2021 | Jul 26 to Aug 10, 2021 | 19040 |
| CM Permian fall 2021 | Oct 3 to 17, 2021 | 19670 |
Values and definitions are also readable in HTML. Source precision and scope apply.
Sherwin et al., supplement Table S3 Original primary evidence
What does campaign coverage tell a reader?
| Campaign | Observation period | Active wells covered (%) | Unique wells covered |
|---|---|---|---|
| Kairos NM Permian | Oct 2018 to Jan 2020 | 76.1 | 21637 |
| CM Permian 2019 | Sep 22 to Nov 3, 2019 | 62.8 | 84570 |
| CM Permian 2020 | Jul 13 to 24, 2020 | 10.9 | 17120 |
| CM Permian summer 2021 | Jul 26 to Aug 10, 2021 | 12.2 | 19040 |
| CM Permian fall 2021 | Oct 3 to 17, 2021 | 12.7 | 19670 |
| CM San Joaquin 2016 | Sep to Oct 2016 | 94.7 | 41547 |
| CM San Joaquin 2017 | Jun to Oct 2017 | 95.3 | 40365 |
| CM San Joaquin summer 2020 | Jul 8 to Sep 1, 2020 | 85.0 | 31919 |
| CM San Joaquin fall 2020 | Nov 9 to 23, 2020 | 81.9 | 33090 |
| CM San Joaquin fall 2021 | Nov 5 to 13, 2021 | 85.1 | 33418 |
| CM Denver-Julesburg summer 2021 | Jul 15 to 23, 2021 | 59.8 | 11671 |
| CM Denver-Julesburg fall 2021 | Sep 19 to 29, 2021 | 59.9 | 11688 |
| CM Pennsylvania 2021 | May 12 to 26, 2021 | 17.7 | 16105 |
| CM Uinta 2020 | Jul 26 to Aug 7, 2020 | 85.0 | 9185 |
| Kairos Fort Worth | Jul to Nov 2021 | 51.6 | 7399 |
The coverage table is transcribed from supplementary Table S3. Its counts refer to unique wells in each campaign, before conversion to well sites. Repeat campaigns in the Permian, San Joaquin and Denver-Julesburg overlap geographically. The authors count only the largest campaign in each region when calculating coverage totals, to avoid double counting. Summing every row in this table would violate that design.
Why do very large sources matter?
Sherwin et al. report that 0.04% to 1.66% of well sites contribute a majority, 50% to 77%, of well-site emissions in seven of the fifteen surveys. Those ranges describe specific surveys, not one universal national distribution. They explain why an inventory based on typical operating conditions can miss the effect of infrequent high-emission events. A survey also needs to consider emissions below its detection threshold and intermittent sources between flights.
Can satellite detections rank countries by performance?
Only with a compatible method, detection threshold, observation coverage and production denominator. IEA’s source notes describe unequal measured-data availability between countries and limits on scaling detected events. A country with more observed plumes may have more production, better coverage, different atmospheric conditions or more emissions. Detection counts alone do not choose among those explanations.
How should uncertainty and climate comparisons be reported?
The 2015 study’s Table 1 gives a national central estimate of 13 Tg/year with an upper increment of 2.1 and a lower decrement of 1.7 at 95% confidence. The displayed interval is therefore 11.3 to 15.1 Tg/year. Its inventory comparator is 8.1 with an upper increment of 2.1 and lower decrement of 1.4, giving 6.7 to 10.2. These are arithmetic translations of that table, not a re-estimation of statistical uncertainty. (Alvarez et al., Table 1; accessed, Nov 2026).
View chart data
2015 US estimates, Tg/year; 95% intervals calculated from published Table 1 increments.
| Estimate | Central | Interval |
|---|---|---|
| Study | 13 | 11.3 to 15.1 |
| Inventory | 8.1 | 6.7 to 10.2 |
| AMR displayed central gap | 4.9 | No new gap interval inferred |
Values and definitions are also readable in HTML. Source precision and scope apply.
Why specify Table 1 rather than use every interval in the paper?
The adjacent prose gives a slightly different lower uncertainty increment for the study total. AMR binds this comparison to Table 1, reproduces its values and identifies the source location in the download. Mixing the prose lower bound with table central values without explanation would create a new hybrid result. Source fidelity includes preserving small discrepancies rather than silently harmonizing them.
Can the gap receive a simple confidence interval?
Not by subtracting independent-looking endpoints. Error structures may be asymmetric, and the estimates can share inputs. The study itself discusses the statistical comparison. AMR reports the displayed central-value gap and the original intervals separately; it does not manufacture uncertainty for the difference or represent an arithmetic percentage as a new statistical test.
Is methane the same as carbon dioxide equivalent?
No. Methane mass is a physical emissions measure; carbon dioxide equivalent applies a chosen climate metric and time horizon. The 2018 paper explicitly discusses a 20-year radiative-forcing comparison. That comparison cannot be moved into a 100-year accounting frame without changing the calculation. For a lifecycle claim about LNG or coal, include methane, combustion emissions, supply-chain boundaries, efficiency and the selected metric. A single leakage percentage alone cannot answer the entire lifecycle question.
Is flaring the same as venting?
Venting releases gas without combustion. Flaring burns gas but can leave methane unburned, while also producing combustion emissions. A volume reported as vented and flared therefore is not itself methane mass or a methane destruction-efficiency estimate. Use the gas flaring reference for its own source definition and keep conversions explicit.
How much methane could be reduced, and at what cost?
IEA estimates that around 70% of fossil-fuel methane, nearly 85 Mt/year, could be abated using existing technology. It separately estimates that more than 35 Mt/year could be avoided at no net cost using average 2025 energy prices. Technical potential, economic potential and achieved reductions are different measures. (IEA 2026, abatement findings; accessed, Nov 2026).
View chart data
IEA 2026, fossil methane technical and economic assessments.
| Measure | Meaning |
|---|---|
| Around 70% | Technical abatement potential |
| More than 35 Mt/year | No-net-cost potential at 2025 prices |
| Nearly 15 bcm | Targeted near-term supply potential; not additive to long-run totals |
Values and definitions are also readable in HTML. Source precision and scope apply.
| Measure | Value | Unit | Qualification |
|---|---|---|---|
| Existing-technology fossil methane abatement potential | 70 | percent, approximately | Technical estimate, not achieved reduction |
| Existing-technology abatement potential | 85 | Mt/year, nearly | Approximate IEA estimate |
| Avoidable at no net cost | 35 | Mt/year, more than | Based on average 2025 energy prices, not universal zero upfront cost |
| Oil/gas methane capture supply potential | 100 | bcm/year, nearly | Potential supply, deployment and infrastructure needed |
| Non-emergency flaring supply potential | 100 | bcm/year, approximately | Separate source category, not methane mass |
| Near-term targeted potential | 15 | bcm/year, nearly | Subset requiring spare export capacity and importer measures, not additive to long-run totals |
What does no net cost mean?
The source compares capital and operating costs with the value of gas captured and sold or used. It does not mean equipment has no purchase price, that finance is always available or that every operator can monetize captured gas. Infrastructure, access to a market, operating conditions and energy prices affect implementation. A price-dependent estimate should retain its price year.
Can the gas-supply potentials be added together?
IEA describes nearly 100 bcm/year from cutting oil and gas methane and a further approximately 100 bcm/year from eliminating non-emergency flaring. It also identifies nearly 15 bcm of targeted near-term potential. The latter is not a third independent long-run volume; it describes an opportunity under specific export-capacity and importer conditions. Adding 15 to 100 plus 100 would confuse a near-term subset with separate supply categories.
Which operational measures does the source identify?
IEA discusses equipment upgrades, dry rather than wet compressor seals, vapor-recovery units, and changes to detection and repair practices. Coal measures can involve capture and use, oxidation or destruction. Those options address different sources. A global technical potential is not a verified reduction at any specific AMR reader’s asset, and a national commitment does not establish that the resulting emission reduction has occurred.
The reference supports an accurate mitigation question: which physical source is included, what reduction is technically possible, what economic conditions are assumed, and what result has actually been observed? Keeping those four parts together is more useful than labeling every potential volume as gas already available to the market.
How should this research be quoted?
Use the date and denominator beside the number. Attribute original measurements to their producers and AMR’s calculations or comparisons to American Mineral Registry. The specific finding controls below preserve those qualifications and link the exact evidence. A source-derived estimate is not a transaction price, a current operational reading or a forecast guarantee.
Frequently asked questions
What is the current global oil and gas estimate in this reference?
The 2025 IEA oil and gas categories sum to 81 Mt/year, from 45 Mt oil plus 36 Mt natural gas. The broader 124 Mt fossil total includes coal. (IEA 2026; accessed, Nov 2026).
What year is the 2.3% US rate from?
It is the 2015 national supply-chain estimate published by Alvarez and colleagues in 2018, not a 2026 measurement.
Is 2.43% an updated US national rate?
No. It is the six-region weighted result in the 2024 Sherwin paper, with its own coverage and campaigns.
Do Tg and Mt mean the same mass?
One teragram equals one million metric tons. Both units measure methane mass here, not CO2 equivalent.
Why not compare the percentages as a trend?
Their geographies, system boundaries, observation periods and production denominators are not matched.
Did the study independently verify local distribution?
The historical Table 1 uses inventory values for local distribution and oil refining/transportation. Equal rows are not independent validation.
Can I quote the 4.9 Tg gap?
Yes, as AMR’s difference between displayed central values in the 2015 comparison, with that period and source stated.
Are all aerial campaign measurements public?
The Nature paper provides public source data and aggregated summaries but states that some underlying Kairos and Enverus microdata are confidential. AMR reproduces published tables, not a new microdata analysis.
Does zero net cost mean zero investment?
No. It is an economic balance between abatement cost and recovered gas value under specified assumptions.
Are the tables and citations available without registering?
Yes. Read the tables in HTML, download CSV files or copy a finding with its exact source and period.
Methodology and Sources
AMR compiles three distinct evidence sets: IEA’s 2026 tracker estimates for 2025, Alvarez et al.’s Table 1 for the 2015 US oil and gas supply chain, and Sherwin et al.’s 2024 abstract and freely linked supplementary Table S3. These are not pooled into one national rate or time series. The IEA fuel shares divide each displayed fuel value by the published 124 Mt total. The oil-and-gas subtotal is 45 plus 36. The US segment differences subtract the displayed study and inventory central values; the aggregate difference uses the published totals, not a sum of rounded segments. Uncertainty translations consistently use Table 1, including its 1.7 Tg lower decrement. The Nature coverage table preserves campaign-specific well counts and dates without summing repeat geography. (IEA 2026; accessed, Nov 2026). (Alvarez et al., Table 1; accessed, Nov 2026). (Sherwin et al., supplement Table S3; accessed, Nov 2026).
The Nature full article has subscription content, while its abstract, data-availability statement and linked supplement are publicly readable. AMR retrieved its linked PDF and source spreadsheets and inspected the decisive supplement table; it does not claim access to confidential microdata or reproduction of the original Monte Carlo model. The calculation bundle reproduces AMR’s published arithmetic and source comparisons. Primary measurements, uncertainty models, technical abatement estimates and potential gas recovery remain credited to their original authors. Source-cell notes identify every displayed input and distinguish source values from AMR derivations.
Glossary
- Methane mass. The amount of CH4 emitted, here in Tg or Mt.
- Inventory. An accounting estimate based on defined activity and emissions methods.
- Bottom-up estimate. Aggregation from equipment, facility or other source-level evidence.
- Top-down estimate. An estimate inferred from atmospheric observations over an area.
- Leakage rate. Emissions normalized to a specified production or throughput denominator.
- Intensity. Emissions relative to a defined activity or energy unit.
- Confidence interval. An uncertainty interval under the producer’s statistical method.
- Super-emitter. A large source relative to a study’s detection and classification scheme.
- Venting. Release of gas without combustion.
- Flaring. Combustion of gas, with effectiveness dependent on conditions.
- Technical potential. Reduction possible under a specified technology assessment.
- No-net-cost potential. An economic estimate that credits recovered product value against costs.
Sources
- IEA: Global Methane Tracker 2026, key findings. 2025 global estimates, intensity definitions and abatement assumptions.
- Alvarez et al.: US oil and gas supply-chain emissions, Science 2018. 2015 segment estimates and the matched inventory comparison.
- Sherwin et al.: Nearly one million aerial site measurements, Nature 2024. Regional findings, abstract and public data-availability statement.
- Sherwin et al.: Supplementary information. Campaign Table S3, observation dates and coverage definitions.