Satellites and ground stations expose hidden methane gaps across Europe

Methane is the second most important anthropogenic greenhouse gas after carbon dioxide, with a global warming potential roughly 30 times that of CO2 over a century, and industrial-era increases in its atmospheric concentration now contribute 0.54 watts per square metre of effective radiative forcing. Europe has positioned itself as a global leader in cutting this potent gas, anchored by the EU Methane Strategy, commitments under the Global Methane Pledge to reduce global methane emissions 30 percent by 2030 relative to 2020 levels, and the landmark Methane Regulation (EU) 2024/1787, which mandates source-level measurement, reporting and verification, frequent leak detection and repair, and prohibitions on routine venting and flaring. Yet even the most sophisticated national accounting systems can harbour systematic biases, and a new multi-year atmospheric assessment now shows that some of Europe’s most carefully compiled methane inventories may be significantly underestimating the continent’s true emissions.

A research team led by Shuai Sun and Yuzhong Zhang of Westlake University, together with collaborators at Zhejiang University, Nanjing University, Shenzhen University and the Vrije Universiteit Amsterdam, has published a top-down assessment of European methane emissions for 2019 through 2023 in the journal Atmospheric Chemistry and Physics. The study performs a Bayesian inversion that jointly assimilates satellite observations of column-averaged methane from the blended GOSAT and TROPOMI product with hourly ground-based measurements from 28 stations of the Integrated Carbon Observation System, or ICOS, using the GEOS-Chem chemical transport model to simulate atmospheric transport at a horizontal resolution of 0.5 by 0.625 degrees over the European domain. The inversion optimizes annual emissions from 824 spatial clusters across the study region, constructed with a Gaussian Mixture Model, and compares the resulting posterior emissions against both prior inventories and official national reports submitted to the UNFCCC under the Paris Agreement’s enhanced transparency framework.

The technical architecture of the analysis is demanding. Satellite retrievals of column-averaged dry-air methane mole fractions, or XCH4, are drawn from a blended product that employs CO2 proxy GOSAT methane retrievals as a reference to correct retrieval biases in the original TROPOMI data, with roughly eight million satellite observations fed into the inversion each year. Observations poleward of 55 degrees north are excluded because of snow cover, unfavourable illumination and stratospheric model biases. Surface data consist of hourly dry-air methane mole fractions filtered for valid measurements and averaged between 10:00 and 16:00 local time, which generally represent well-mixed boundary layer conditions, while high-altitude mountain stations are excluded due to representativeness issues. The team also quantified and corrected residual systematic biases in the model’s boundary conditions, applying corrections of 12 parts per billion at the western boundary and 30 parts per billion at the northern boundary, an adjustment validated against independent IAGOS-CARIBIC aircraft observations that reduced the mean bias in simulated free-tropospheric methane from 14.1 to 2.9 parts per billion.

One of the study’s most striking findings concerns the information content of the two observing platforms. Using averaging kernel sensitivities and degrees of freedom for signal, or DOFS, the joint inversion achieves 97 independently constrained pieces of information across the entire domain, with the strongest constraints over central and western Europe, particularly Germany and the Benelux states. Sensitivity inversions using each platform alone reveal that surface measurements provide the dominant constraint, with a surface-only DOFS of 73 for the entire domain and 69 for the European countries, while the satellite-only inversion yields just 31. Satellite observations contribute little additional information where the ICOS network is dense, but they provide valuable complementary coverage over the Iberian Peninsula, eastern Europe, western Russia and adjacent regions such as Turkey and North Africa, where ground-based stations are almost entirely absent. The two platforms are also strongly complementary in space: surface sites cluster in Germany, the Netherlands and northern France precisely where satellite coverage is comparatively sparse.

The headline numbers are consequential. The inversion estimates total European methane emissions at 25.7 teragrams per year for 2019 through 2023, including 22.5 teragrams per year from anthropogenic sources, which is 1.0 teragram per year higher than the prior inventory built from the EDGAR v8 database and multiple natural emission datasets. Livestock emerges as the largest anthropogenic source at 11.6 teragrams per year, followed by landfills at 3.9, natural gas at 2.1, wastewater at 1.9, coal at 0.92 and oil at 0.36 teragrams per year. The largest sectoral adjustment is for livestock, increased by about 1.0 teragram per year, or 10 percent, while wetland emissions, the largest natural source, are revised downward from 3.3 to 2.6 teragrams per year. For the EU27 countries alone, posterior anthropogenic emissions of 17.2 teragrams per year are 1.3 teragrams per year, or 8.0 percent, higher than the prior estimate of 15.9, suggesting the underestimation is primarily attributable to EU member states.

The regional discrepancies are where the story turns genuinely uncomfortable for national inventory compilers. Posterior anthropogenic emissions for Germany, the Netherlands and Italy stand at 2.78, 1.15 and 1.87 teragrams per year respectively, 25 to 69 percent higher than prior estimates and 18 to 80 percent higher than the values these countries officially report to the UNFCCC. The inversion attributes most of these upward adjustments to the agriculture sector, a conclusion consistent with a recent satellite inversion over western and central Europe that inferred a 17 to 29 percent increase in livestock emissions, and with a local-scale survey of 51 Dutch dairy farms that reported whole-farm methane emissions averaging 27 percent higher than Dutch National Inventory estimates. Notably, urban emission measurements in Berlin, Groningen, Utrecht, Rotterdam and Milan have not shown significant discrepancies from inventories, pointing away from city sources and toward potential systematic biases in national-level livestock accounting, even when high-tier methodologies incorporating detailed activity data on animal diet, milk production and feed use are employed.

Fossil fuel production regions tell a different tale. Over the North Sea, posterior emissions from offshore oil and gas production are 0.28 teragrams per year, 53.4 percent lower than prior estimates, a downward adjustment informed mainly by surface observations but agreed upon in direction by both satellite-only and surface-only inversions. This contrasts with a ship-based study that found substantial emissions from UK offshore platforms, though those measurements were limited to a short timeframe. Over southern Poland, the inversion refines the spatial pattern of emissions within the Upper Silesian Coal Basin, the production hub accounting for the majority of the EU’s hard coal output, with national emissions nearly unchanged at 1.7 teragrams per year and coal mining emissions revised upward from 0.40 to 0.47 teragrams per year, broadly consistent with previous multi-scale observational studies. Romania and Ukraine show only minor adjustments, but the authors caution this reflects weak observational constraints, with DOFS of just 0.75 for Romania and 1.38 for Ukraine, meaning the inversion cannot conclusively rule out significantly different UNFCCC values for Ukraine’s oil and gas sector.

Urban hotspots also receive moderate upward revisions. Posterior emissions are 2.70 teragrams per year for France and 1.74 for Spain, increases of roughly 7 and 15 percent relative to prior estimates, while the United Kingdom’s posterior anthropogenic total of 1.58 teragrams per year remains nearly unchanged because increases over land are offset by decreases in offshore oil and gas emissions. The spatial pattern reveals concentrated upward adjustments over Paris, Madrid and London, in addition to weaker diffuse adjustments attributed to agriculture. Consistent with this, ground-based and satellite remote sensing studies have identified landfills as a major underrepresented methane source in Madrid, and eddy covariance flux measurements in central London show urban methane fluxes substantially underestimated in inventories, with contributions from both biogenic and fossil fuel sources.

Perhaps the most encouraging result is the trend. Both prior and posterior estimates show a significant decline in European anthropogenic methane emissions over 2019 through 2023, with the inversion yielding a linear trend of minus 0.25 teragrams per year squared, statistically significant at p equals 0.046, driven primarily by reductions in landfills, coal mining and natural gas, while trends for livestock and wastewater are statistically insignificant. Yet the spatial pattern of the decline diverges sharply from bottom-up expectations: increases in livestock and waste emissions across western Europe, including positive trends in the Netherlands, Belgium, western Germany, northern Italy and the southern United Kingdom, contrast with decreases in fossil fuel production regions such as the North Sea and southern Poland. The Netherlands flips from a negative prior trend of minus 0.016 to a slightly positive posterior trend of plus 0.022 teragrams per year squared, while Ukraine exhibits a stronger negative posterior trend of minus 0.09 than the prior minus 0.05. These findings highlight the value of atmospheric observations for independently evaluating methane inventories and informing climate policy, and they suggest that even Europe’s high-quality reporting systems may need atmospheric verification to confirm that the promised reductions are truly materializing.

Subject of Research: Earth and Climate Science

Article Title: Assessment of methane emissions from European countries using satellite and surface observations

Article References: Sun, S., Zhang, Y., Chen, W., Pan, S., Pu, D., Tong, X., Chen, H., & Houweling, S. (2026). Assessment of methane emissions from European countries using satellite and surface observations. Atmospheric Chemistry and Physics, 26(20), 14981-15001. https://doi.org/10.5194/acp-26-14981-2026

Image Credits: AI Generated

DOI: 10.5194/acp-26-14981-2026

Keywords: Assessment, methane, emissions, European, countries, satellite, surface, observations, scientific research