Methane is a short-lived measurement priority
Methane remains in the atmosphere for far less time than carbon dioxide but has much stronger near-term warming influence per unit mass. The IPCC's AR6 physical-science assessment gives a methane lifetime of about 11.8 years and global-warming potentials of 81.2 over 20 years and 27.9 over 100 years for methane, with stated methodological conventions. The time horizon matters: it is not legitimate to select a factor merely because it makes a claim look larger or smaller. Rapid reductions can affect near-term warming, while sustained carbon-dioxide reduction remains indispensable for limiting long-term temperature.
The IEA estimates that the energy sector, including oil, gas, coal, and bioenergy, produces around 40% of methane emissions from human activity. It also identifies fossil operations as unusually actionable because known equipment and operating changes can address many sources. This does not make agriculture and waste unimportant; it indicates where detection can be connected to valves, tanks, compressors, vents, mines, and abandoned assets with identifiable owners. A measurement programme should name its sector and source boundary rather than compare totals assembled from incompatible inventories.
Inventories and observations answer different questions
Engineering inventories estimate emissions from equipment counts, activity data, and emission factors. They provide continuous accounting coverage but can miss abnormal operating states and rare large releases. Site measurements observe concentrations or fluxes during specific windows, but sampling, wind, access, and detection limits constrain extrapolation. Satellites offer broad repeat coverage for sufficiently large, visible plumes; aircraft can survey facilities at higher resolution; drones and optical gas imaging can locate components; continuous monitors can reveal temporal behavior. None is a universal ground truth.
A tiered architecture assigns each method a role. Inventories establish the population and reporting frame. Broad-area screening finds anomalous sources. Targeted aerial or ground surveys attribute a plume to equipment. Continuous sensing characterizes persistence and recurrence at priority sites. Direct quantification or engineering reconciliation supports the emissions estimate. The system should retain negative observations with their coverage and sensitivity, because 'not detected' only means below a method's threshold under the observed conditions. Reconciliation should explain divergence between bottom-up and top-down estimates instead of averaging them into a number with no physical interpretation.
Detection limits belong beside every result
Remote methane retrieval depends on absorption signatures, surface reflectance, clouds, sun angle, atmospheric state, wind, pixel size, plume geometry, and the algorithm used. NASA's EMIT instrument demonstrates that an imaging spectrometer designed for surface mineral mapping can identify methane and carbon-dioxide plumes from energy, waste, and agriculture sources. Its observations are powerful evidence of localized high emissions, but they are conditional snapshots. A plume length or enhancement image alone is not an annual mass estimate.
Every reported event should include acquisition time, instrument, processing version, quality flags, estimated detection threshold, wind data and source, spatial uncertainty, flux estimate and confidence interval where supported, and the rule used to group repeated plumes into an event. Quantification uncertainty can be asymmetric and dominated by wind. Persistence requires multiple valid opportunities, not the number of calendar days between images. When instruments disagree, retain both observations and diagnose timing, threshold, geometry, or attribution rather than selecting the value closest to an inventory.
Design the alert-to-repair chain as an operating system
UNEP's Methane Alert and Response System uses data from more than 30 satellite instruments with scientific review and analytical models to notify governments and companies of very large emission events. The architecture illustrates a necessary shift from publication to response. An alert record should progress through detection, quality control, candidate source, responsible jurisdiction and operator, notification, acknowledgement, investigation, action, verification, and closure. Each state needs a timestamp, accountable party, evidence attachment, and reason code for delay or non-action.
Performance should be measured as a funnel. Report valid observation opportunities, detections, attributed sources, notifications delivered, acknowledgements, investigations, repairs, verified cessations, recurrences, and estimated emissions avoided. Publish median and tail latency between each stage. Counting alerts rewards surveillance volume rather than mitigation. Closure should require a post-action observation or documented engineering test appropriate to the source; absence in one satellite pass may be insufficient because cloud, wind, or threshold can hide a continuing release. Repeat observation plans should be risk-based and explicit.
Prioritize sources without losing systematic coverage
The IEA estimates that around 70% of fossil-fuel methane, nearly 85 million tonnes annually, can be abated with existing technology. More than 35 million tonnes could be avoided at no net cost under average 2025 energy prices because captured gas has value. These are modeled global potentials, not promises for a particular facility. A site's economics depend on gas composition, pressure, access to gathering or use, labour, safety, regulation, production state, and whether the source is intermittent. Captured value can disappear where there is no market or infrastructure.
Large sources deserve rapid response because abatement per intervention can be high, but a super-emitter-only strategy can miss the aggregate burden of smaller equipment. Use risk-based tiers: continuous or frequent monitoring for high-throughput sites and known failure modes, periodic surveys for the broader population, and event-triggered inspection after maintenance or process changes. Common measures include leak detection and repair, replacing gas-driven equipment, vapour recovery, eliminating routine venting, improving flare performance, and addressing abandoned wells and mines. Programmes should report both large-event mitigation and inventory-wide intensity.
Verify the outcome and preserve uncertainty
A credible facility record links equipment identity, operating state, inventory estimate, observations, raw data or stable references, calibration, meteorology, attribution, work order, replaced component, captured or routed gas, post-repair test, and subsequent monitoring. Quantify emissions avoided over a stated period with an explicit counterfactual and uncertainty. Do not annualize a short observed release unless persistence is supported. If a repair changes production or shifts emissions to another vent, the boundary must capture it. Independent review should focus on high-materiality events and controls around manual edits.
The public scorecard should separate measured, estimated, and modeled values. Useful indicators include measurement coverage, minimum detectable emissions by method, inventory reconciliation, alert response time, repair completion, verified recurrence, methane intensity with numerator and denominator definitions, and avoided emissions with confidence ranges. Policy should specify acceptable methods while allowing new instruments to qualify through blind tests and controlled releases. Measurement has succeeded only when it produces a verified change in emissions or a better bounded inventory. A visually convincing plume is evidence to investigate, not the end of the scientific process.
Scope and limitations
Global sector shares and abatement potentials are modeled estimates that depend on source boundaries, facility data, technology assumptions, energy prices, and the value assigned to captured gas. Satellite coverage and detection thresholds vary by instrument, surface, cloud, illumination, and wind. A non-detection is not proof of zero emissions, and a detected plume may not represent annual performance. Facility conclusions require source-level investigation, calibrated measurements, operating context, and post-repair verification.
References
Source review: 20 August 2026. Quantitative values retain their original definitions, periods, and boundaries.
- 01Global Methane Tracker 2026
International Energy Agency · 2026
www.iea.org ↗ - 02Global Methane Tracker 2026: Key Findings
International Energy Agency · 2026
www.iea.org ↗ - 03Global Methane Tracker 2026: Strategies to Speed Action
International Energy Agency · 2026
www.iea.org ↗ - 04Methane Alert and Response System
United Nations Environment Programme · 2026
www.unep.org ↗ - 05Climate Change 2021: The Physical Science Basis, Chapter 7 Supplementary Material
Intergovernmental Panel on Climate Change · 2021
www.ipcc.ch ↗ - 06Methane Plumes Detected by EMIT Space Mission
National Aeronautics and Space Administration · 2024
svs.gsfc.nasa.gov ↗