Human activities now fuel two-thirds of global methane emissions

Our methods and data are derived from the Global Carbon Project's most recent global methane budget (Saunois et al 2024). We employ a TD ensemble of 24 inversions based on seven different inverse systems that use data for atmospheric CH₄ concentrations to constrain total emissions and attribute them to primary sources. The TD inversions are either constrained by surface observations for the period 2000–2020 (18 out of 24 simulations), or satellite observations from the Greenhouse Gases Observing Satellite (GOSAT) for the period 2010–2020. Prior fluxes, observations, treatments, and optimization configurations varied modestly across the inversions as described in Saunois et al (2024).

Most of the atmospheric inversions used the same OH field, treated as constant over time, attributing changes in methane concentrations to altered emissions rather than to altered atmospheric oxidative capacity. Although, chemistry-climate models suggest potential OH increases between 2000 and 2010 and decreases after 2010 (e.g. Zhao et al 2020, Skeie et al 2023), OH inferred from hydrofluorocarbon and methyl chloroform observations suggest insignificant trends over the past two decades (Patra et al 2021, Thompson et al 2024). Models and observations also indicate small interannual variability of OH (<2% in Thompson et al (2024) and Patra et al (2021)), implying ∼10−11 Tg yr⁻¹ uncertainty in interannual flux variability if OH variability is ignored in recent decades (Saunois et al 2024). In 2020, OH seems to have decreased sharply by 2% compared to 2019 attributable to reduced NOx emissions that may have explained at least half of the surge in growth rate (Peng et al 2022). In future scenarios, while decarbonization can decrease methane emissions, the associated reduction in NO_x emissions has the potential to increase methane's lifetime in the atmosphere. The overall climate and air quality benefits thus require careful evaluation.

The Global Methane Budget also employs detailed BU accounting that includes global inventories and biogeochemical modeling that provides a more detailed attribution to sources but lacks the constraint of total atmospheric growth rate used in TD approaches. BU trends in methane emissions are available for anthropogenic emissions using four global inventories (EDGARv6 and v7, CEDS, GAINS and EPA2019), five fire products for biomass burning (GFEDv4.1s, QFEDv2.5, GFAS and FINNv1.6 and v2.5) and 13 biogeochemical models for wetland emissions (Saunois et al 2024). However, estimates for other natural sources such as geological, termites, permafrost, rivers, lakes, and reservoirs available in the literature lack sufficient measurements to analyze temporal changes in methane emissions and, as a result, trends cannot currently be calculated for these sources.

Estimated methane emissions from inland freshwaters for the new global methane budget use new spatial products and for the first time attribute some freshwater sources to anthropogenic activities ('direct' via river damming or other human-constructed small lakes and ponds) or influences ('indirect' via eutrophication induced by enhanced nutrient loadings from the surrounding catchments). Notable improvements over the last GMB include the addition of new, spatially explicit estimates of lake and reservoir emissions derived from both observations and models (Johnson et al 2021, 2022, Zhuang et al 2023).

To further interpret estimated methane emissions changes in recent years after the GMB budget period in 2020, we use a TD emission quantification based on methane column measurements from TROPOMI for the years 2018–2023. This inverse system is based on the GEOS-Chem adjoint 4D-Var methane inverse system (Yu et al 2021). Prior fluxes, treatments, and optimization configurations are as described in Yu et al (2021 and 2023). Results here are the average of two inversions: one uses a fixed OH field from chemical transport model simulations, whereas the other optimizes OH concentration magnitude simultaneously with methane emissions. To extend the evaluation period from the GMB model ensemble, the recent-year emission increases are based on the emission estimates of 2019 and 2023 from this model (GEOS-Chem) constrained by a TROPOMI product, rather than a comparison to the GMB year-2019 estimates.

Despite an increasing policy focus on methane as a potent greenhouse gas, methane emissions continue to rise. Global anthropogenic methane emissions reached 370 [range 343–403] and 384 [358–411] Tg CH₄ yr⁻¹ based on BU and TD estimates, respectively, averaged for the three years from 2018 to 2020 (table 1). These emissions are 50–60 Tg CH₄ yr⁻¹ (15%–20%) higher than for the period 2000–2002, nearly two-decades earlier (table 1). Total global methane sources, both natural and anthropogenic, rose by 50–70 Tg CH₄ yr⁻¹.

Our best estimates for anthropogenic methane emissions in 2020, the last year for which full data for the GMB are available, are 372 [345–409] and 392 [368–409] Tg CH₄ yr⁻¹ for BU and TD methods, respectively (figure 1, table 1). The largest emissions sources are: wetland and inland freshwaters, agriculture and waste, and fossil fuel production and use (figure 1). Direct anthropogenic emissions from TD estimates now comprise ∼65% of global emissions. When the ∼50 yr⁻¹ or more of 'indirect anthropogenic emissions,' such as those from dams and reservoirs, are included (Saunois et al 2024), the total is more than two-thirds anthropogenic.

Zoom In Zoom Out Reset image size

Almost all major sectors of anthropogenic emissions rose substantially from 2000 to 2020. Emissions from agriculture and waste rose by 33 Tg CH₄ yr⁻¹ or one-sixth overall to 219 and 239 Tg CH₄ yr⁻¹ for BU and TD estimates, respectively (figure 2, table 1). Emissions from cows (and other ruminants) and from landfills (and other waste) both rose by ∼15 Tg CH₄ yr⁻¹ from 2000–2002 to 2018–2020. Fossil fuel emissions rose an estimated 18–27 Tg CH₄ yr⁻¹ (18%–28%) as estimated by TD and BU approaches, respectively (table 1). Methane emissions from fossil fuel extraction and use are now comparable to direct methane emissions from cows and other ruminants globally based on our estimate, but emissions from agriculture and waste, including landfills, remain approximately twice those associated with fossil fuels (table 1). This partitioning may vary among the individual inventories and TD estimates.

Zoom In Zoom Out Reset image size

For natural emissions, inland freshwaters were estimated to emit 112 (49–202) Tg CH₄ yr⁻¹ globally. One major change from previous Global Methane Budgets is the allocation of some freshwater and wetland emissions to anthropogenic actions, such as emissions from human-built reservoirs (Saunois et al 2024). For example, 50% of inland water emissions (56 of 112 Tg yr⁻¹) are now estimated to be influenced by anthropogenic actions, including those from human-built reservoirs (30 Tg yr⁻¹) and through eutrophication, warming, and other anthropogenically driven factors (Saunois et al 2024). Similarly, 30 Tg yr⁻¹ of the ∼160 Tg yr⁻¹ emitted from wetlands globally (table 1) are estimated to be influenced by anthropogenic factors such as climate change and CO₂ fertilization. Given this new partitioning of previously categorized 'natural' wetland and inland freshwater emissions, the contribution of 'anthropogenic methane emissions' is likely to be greater than two-thirds, even when including only the additional 30 Tg yr⁻¹ of emissions from human-built reservoirs.

The indirect (non-reservoir) anthropogenic emissions, however, need follow-up investigations using process-based models similar to those used to perform regional (e.g. Guo et al 2020) or global (e.g. Zhuang et al 2023) assessments of future emissions under future climate scenarios. Such models consistently project a substantial increase in methane emissions by the end of the 21st century (in the range of about 30%–80% depending on the scenario), supporting the notion that—like wetlands—inland freshwaters are also sensitive to warming. This feedback is poorly accounted for in the current decomposition between present-day natural and anthropogenic component fluxes and eutrophication. Such increases, whether induced by direct or indirect (climate change, eutrophication) drivers, have likely already contributed to the observed decadal increase in atmospheric CH₄ concentrations. Their contribution remains relatively unquantified, however, contributing to uncertainties in the assessment of decadal changes in the GMB. Natural wetlands showed no statistically significant increase in methane emissions when comparing the three-year periods 2000–2002 to 2018–2020 (table 1).

Global methane sinks are increasing in response to rising atmospheric methane concentrations. Total global sinks estimated by TD approaches rose by 35 Tg CH₄ yr⁻¹ from an average 538 [529–549] Tg CH₄ yr⁻¹ in 2000–2002 to 573 Tg CH₄ yr⁻¹ for the period 2018–2020 (table 1). Unsurprisingly, most of the increased oxidation of methane (∼30 Tg CH₄ yr⁻¹) came from OH radicals in the troposphere.

The imbalance between global sources and sinks continues to grow and is reflected in the increasing atmospheric growth rates observed for methane in recent years (Lan et al 2024). The average imbalance between global sources and sinks in the early 2000s was ∼6 Tg CH₄ yr⁻¹ (table 1). In contrast, this imbalance grew to 32 [15–38] and 52 Tg CH₄ yr⁻¹ in 2020 based on TD and BU methods, respectively, values that bracket the actual atmospheric growth rate of 15 ppb CH₄ or 42 Tg CH₄ that year (Lan et al 2024).

The estimated global increase in methane emissions from 2000 to 2020 arises largely from four regions or countries (figure 3, table 2): China (BU: 19 [17–23] Tg CH₄ yr⁻¹; TD: 7 [−7–17] Tg CH₄ yr⁻¹), South Asia (BU: 10 [7–14] Tg CH₄ yr⁻¹; TD: 10 [2–17]Tg CH₄ yr⁻¹), SouthEast Asia (BU: 8 [0–14] Tg CH₄ yr⁻¹; TD: 6 [−2–10] Tg CH₄ yr⁻¹) and the Middle East (BU: 8 [4–15] Tg CH₄ yr⁻¹; TD: 8 [2–22] Tg CH₄ yr⁻¹), mostly attributable to anthropogenic emissions. These regions are followed by Equatorial Africa (BU: 7 [3–17] Tg CH₄ yr⁻¹; TD: 4 [−2–8] Tg CH₄ yr⁻¹) and the USA (BU: 3 [−3–18] Tg CH₄ yr⁻¹; TD: 6 [−6–16] Tg CH₄ yr⁻¹). However, large uncertainties exist among each approach, and discrepancies appear between approaches for China, the USA and Equatorial Africa. Two groups of countries show decreasing emissions: Europe (BU: −7 [−10–4] Tg CH₄ yr⁻¹; TD: −7 [−10–2] Tg CH₄ yr⁻¹) as highlighted previously (Jackson et al 2020) and, possibly, Australasia (BU: −1 [−4–2] Tg CH₄ yr⁻¹; TD: 0 [−2–0] Tg CH₄ yr⁻¹).

Zoom In Zoom Out Reset image size

The distribution of emission changes from 2000 to 2020 by latitude emphasizes the tropics, which contribute an estimated ∼60%–70% of the total global change over the last two decades for both approaches (BU: 45 [29–68] Tg CH₄ yr⁻¹; TD: 36 [6–47] Tg CH₄ yr⁻¹) (table 2). Mid-latitudes are responsible for the additional 30%–40% increase in global emissions; in contrast, emissions from higher latitudes (60–90°N) are estimated to be stable or to have decreased slightly, attributable to slightly decreasing anthropogenic emissions (table 2).

Extending emission estimates into the year 2023 based on TD analyses of TROPOMI satellite measurements, estimated global methane emissions increase by an additional 19 Tg over the 4 year interval from 2019 to 2023 (figure 4). Tropical regions contribute the most to recent emission increases (>7 Tg from 2019 to 2023), particularly in the Congo and, to a lesser extent, parts of southeast Asia and southern Brazil (figure 4). Another area of increase from 2019 to 2023 is observed near Beijing, China (figure 4). As a full ensemble of inversions becomes available with more data through 2023, additional estimates will provide more comprehensive coverage of emission increases.

Zoom In Zoom Out Reset image size

Methane is receiving increasing policy attention because of its potential for reducing warming over the next few decades. Recent analyses suggest that methane mitigation may be cheaper than CO₂ mitigation for a comparable climate benefit (UNEP & CCAC 2021). Better quantification and attribution of methane sources are needed to support such mitigation efforts locally, regionally, and globally.

Closer-to-real-time estimates of methane sources, for example, will be aided by new satellites such as MethaneSAT and CarbonMapper, to identify and quantify methane super-emitters (Duren et al 2019). Current BU inventories do a poor job of representing the long tail of such emissions in inventory budgets. More complete (and rapid) incorporation of satellite data will also improve TD methane inversions, regional emission estimates, and national greenhouse gas inventories. As part of upcoming GMB efforts, we expect to produce national methane budgets for some key countries.

Uncertainties in methane emissions remain even greater for natural sources, such as wetlands, freshwater systems, and natural geologic sources. Process-based model assessments of inland freshwater CH₄ emissions are needed that rely on common system delineations and simulation protocols. This effort could follow a format similar to that of the WETland CH4 Inter-comparison of Models Project (WETCHIMP) launched a decade or more ago (Melton et al 2013). Process-based models are needed to constrain uncertainties, to strengthen confidence in the spatial upscaling of CH₄ fluxes, and to resolve temporal variability (e.g. responses to climate extremes, inter-annual variations, and decadal trends). Similar to past wetland studies, simulations should be performed within the broader framework of land-surface models (LSMs) and would benefit from recent advances in LSM-based simulations of inland water CO₂ and N₂O fluxes (e.g. Yao et al 2020, Zhang et al 2022, Tian et al 2023).

TD inversions will benefit from incorporating additional tracers, including ethane and ¹³CH₄—wherever global data are available—to enhance CH₄ abundance data from flask and satellite sampling. Ethane, for instance, and methane are co-emitted from fossil fuel exploration, but not from biological sources (e.g. Lan et al 2019, Barkley et al 2021). Additionally, methane produced from biological processes, such as in wetlands, landfills, or livestock, typically has a lighter ¹³CH₄:¹²CH₄ ratio compared to methane from fossil fuel extraction or industrial processes. Different sinks also fractionate against the slightly heavier ¹³CH₄ consuming it at a slightly slower rate than for ¹²CH₄. Incorporating these additional tracer measurements can thus help researchers identify and quantify the contributions of methane sources and sinks more accurately. Furthermore, TD quantification would benefit from the recent advances in synthesizing ¹³CH₄ source signatures into grid-level estimates for fossil and non-fossil sources (Sherwood et al 2021).

Estimates of methane sinks also require new approaches and research tools. The oxidative capacity of the atmosphere is difficult to measure directly, particularly for short-lived OH and Cl radicals that oxidize most emitted methane. Methyl chloroform (CH₃CCl₃) has provided the most useful constraint on the atmosphere's oxidative capacity (Patra et al 2021). However, its abundance has been declining for decades and is now below 1 pptv (Rigby et al 2017). We need new tracers and tools to assess the abundance of OH radicals, in particular, which have a lifetime of only seconds. Opportunities to constraint the oxidative capacity of the troposphere include other halogens compounds such as HFC-134a, HFC-32 and HCFC-141b (Thompson et al 2024) and isotopic information from¹⁴CO (Brenninkmeijer 1993) and ¹³CO (Mak and Brenninkmeijer 1998).

Methane concentrations have risen faster over the past five-year period than in any period since record-keeping began. Understanding where and why this is happening is a central goal of the Global Methane Budget. At least two-thirds of global methane emissions are now attributable to anthropogenic sources, an outcome that cannot continue if we are to maintain a habitable climate.

The authors acknowledge the many scientists who contribute to the Global Methane Budget released by the Global Carbon Project (globalcarbonproject.org). Our research was supported by the Gordon and Betty Moore Foundation through Grants GBMF5439 'Advancing Understanding of the Global Methane Cycle' and GBMF11519 'Advancing the understanding of methane emissions from tropical wetlands' to Stanford University and the Agence National de la Recherche through the project Advanced Methane Budget through Multi-constraints and Multi-data streams Modelling (AMB-M³)—(ANR-21-CE01-0030). The authors acknowledge additional funding through the framework of UNEP's International Methane Emissions Observatory (IMEO) (#DTIE21-EN3143 to Stanford University, the Laboratoire des Sciences du Climat et de l'Environnement (LSCE-IPSL), and CSIRO Environment, Canberra, Australia) and from Australia's National Environmental Science Program - Climate Systems Hub (JGC) and from Future Earth. PRE acknowledges funding from the FRS-FRNS PDR project T.0191.23 CH4-lakes.

The data that support the findings of this study are openly available at the following URL/DOI: https://www.icos-cp.eu/GCP-CH4-2024.