the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 16 Apr 2025
Trends and seasonality of 2019–2023 global methane emissions inferred from a localized ensemble transform Kalman filter (CHEEREIO v1.3.1) applied to TROPOMI satellite observations
Abstract. We use 2019–2023 TROPOMI satellite observations of atmospheric methane to quantify global emissions at monthly 2°×2.5° resolution with a localized ensemble transform Kalman filter (LETKF) inversion, deriving monthly posterior estimates of emissions and year-to-year evolution. We evaluate the sensitivity of the inversion to the assumed wetland distribution by using two alternative wetland inventories (WetCHARTs and LPJ-wsl) as prior estimates. Our best posterior estimate of global emissions shows a surge from 560 Tg a-1 in 2019 to 587–592 Tg a-1 in 2020–2021 before declining to 572–570 Tg a-1 in 2022–2023. Posterior emissions reproduce the observed 2019–2023 trends in methane concentrations at NOAA surface sites and from TROPOMI with minimal regional bias. Consistent with previous studies, we attribute the 2020–2021 methane surge to a 14 Tg a-1 increase in emissions from sub-Saharan Africa but find that previous attribution of this surge to anthropogenic sources (livestock) reflects errors in the assumed wetland spatial distribution. Correlation with GRACE-FO inundation data suggests wetlands in South Sudan played a major role in the 2020–2021 surge but are poorly represented in wetland models. By contrast, boreal wetland emissions decreased over 2020–2023 consistent with drying measured by GRACE-FO. We find that the global seasonality of methane emissions is influenced by northern tropical wetlands and peaks in September, later than the July wetland model peak and consistent with GRACE-FO. We find no global seasonality in oil/gas emissions, but US fields show elevated cold season emissions that could reflect increased leakage.
- Preprint
(3969 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 31 May 2025)
-
RC1: 'Comment on egusphere-2025-1554', Anonymous Referee #1, 12 May 2025
reply
Review :
Pendergrass et al. Trends and Seasonality of 2019-2023 global methane emissions.
General Comments
This is an important paper that should be published, but after revision. The topic is of major interest, both scientifically and to wider society, since the dramatic rise in the atmospheric methane burden during the 2019-2023 period was very unexpected and poses a major challenge to the hopes of the UN Paris Agreement and to the Global Methane Pledge. The paper is scientifically innovative, using satellite data and impressive new modelling methodology to assess emissions, and the findings, that tropical wetlands played a major role, are consistent with other studies. However I have some concerns about details of the work, and recommend minor revision before final publication.
Wider comments
The manuscript makes no attempt to discern a wider cause of the growth. For example, Lunt et al 2021, Palmer et al 2023 and Nisbet et al 2023a all discussed the Indian Ocean Dipole. These are large events, unusual and important. It would be nice to address the deeper meterological causes.
Overall conclusion
This is a very interesting and potentially important paper that should be published, but it needs some revision.
Specific Points
Line 32- “previous work has attributed the surge to human-caused emissions rather than wetland.” That’s not fully correct – see for example Drinkwater et al., 2023, who “suggest that wetlands have played a significant role in recent growth of atmospheric CH4”. Similar conclusions were reached by Nisbet et al. 2023a,b, and Michel et al. 2024, neither of whom are cited. Similarly, discussing a slightly earlier time frame, there are many papers such as Zhang et al. 2024, and Feng et al. 2023 (already cited elsewhere in the manuscript).
Lines 50-55 – should also cite Michel et al. 2024 who highlight wetlands and Nisbet et al. 2023a who has a detailed discussion of the wetland hypothesis.
Line 78 – solar backscatter: somewhere in the paper or supplemental information there should be a comment on the constant thick cloud cover over the wetter parts of the Congo basin during the rainy seasons, making the boundary layer invisible from remote sensing at times of highest wetness (and presumably productivity). It is very difficult to see the Congo and Amazon wetlands when they are most busy. The Sudd, Okavango and Niger inland delta are exceptions as they are river-fed, far from the rainfall.
Line 90 – August peak of emissions? That depends on latitude zone. While it’s likely true for the NH north of about 10N (but including the Sudd) it is not true for most Southern hemisphere wetlands (Okavango excepted), nor for parts of the equatorial zone (2 wet seasons, 2 dry seasons annually). During the missing data episodes, was scaling to seasonality done by retrieval latitude? Or just one-size-fits-all?
Line 112 – WETCHART - I’m concerned that over the Upper Congo Bangweulu wetland, Shaw et al (2022) found a huge discrepancy between their measured results and model estimates from WETCHART and GCP. Now Bangweulu is papyrus/reed wetland at about 11oSouth and about 1100m altitude, and the Sudd is papyrus swamp about 5-10oN and a warmer 400m altitude, but they are broadly comparable. Shaw et al said “The models underestimated emissions by a factor of 10 on average when compared with fluxes derived from the airborne measurements”. Note that France et al (2022) also got very high emissions. In other words I’m concerned that WETCHART has its problems in modelling tropical wetlands. Shaw et al. suggest “that land surface models currently lack the ability to accurately predict emissions from vegetated tropical wetlands in Africa” Are there any field campaign tests to show LPJ-MERRA performs better?
Table 1 and Line 118-127 (Priors) – part of the problem is that in Africa the emissions from livestock can be indistinguishable from wetland emissions – same place, same seasons, same isotopes. In dry areas wetlands are few and cattle eat tree foliage and often stray far from water (especially now as boreholes are common), but in water-rich regions like the Zambian wetland and South Sudan the wetlands are full of both cows and antelope (which are ruminant) (not to mention pseudo-ruminant hippos also, and camels in some areas). In a large wetland like the Sudd, the only difference between wetland methane and cow methane is that the wetland source moves a bit more slowly. Same location and same d13C.
Line 123 and Line 238 – do you really think scaling South Sudan cattle to 0.1o x 0.1o is valid? I have never been to South Sudan but I’ve flown over the area, and I’ve been all around (Uganda, N. Sudan, Ethiopia) and talked to people with knowledge – I’d be surprised if anyone has valid estimates of South Sudan cattle populations, especially not in unrest areas. FAO’s basically a wild guess. African wetlands are full of cows, but the woodland and scrub around the wetland are also full of browsing cows and goats. Moreover, there are 6 million antelope seasonally migrating (and ruminating) through the Sudd, if you believe the African Parks estimate (https://www.africanparks.org/worlds-largest-land-mammal-migration-confirmed-south-sudan). Maybe that’s an overestimate, maybe not. The manuscript doesn’t make clear how these ruminant population issues are assessed. Populations of kob and reedbuck are likely rapid responders to good nutrition from water-driven growth. So gridding down is a tough call.
Line 125 – geological scaling to Hmiel – yes, agree.
Line 129-134 and also line 148 – OH is a key part of the paper but skimmed over – the discussion is all about sources but very little about sinks. I’m especially concerned about the year 2020. While Qu et al 2022 (cited in ms) attributed only 14% of the surge to OH changes, Peng et al. 2022 state they “attribute the methane growth rate anomaly in 2020 relative to 2019 to lower OH sink (53 ± 10 per cent) and higher natural emissions (47 ± 16 per cent), mostly from wetlands.” Also note Bouarar et al 2021 – something big was going on with OH and needs to be considered. Finally, note Morgenstern et al on 14CO. More generally, yes, OH is a huge and ill-constrained topic in its own right, and maybe it’s wise to bite off one chunk of the problem at a time (e.g. sources), but these 5 lines do gloss over a whole mega-can of OH worms (and soil sinks, as well as Cl which has big fractionation on 13C).
Line 130 – OH lifetime – specify the type of lifetime (steady state vs perturbation)
Line 148 – ‘we do not optimize OH’….Line 149 – ‘prevent constraining OH as a local variable’ ? – explain? This may be a key factor in the 2020 surge – see Peng et al and also Bouarar et al.
Line 249 – attributing between livestock and wetland in the Sudd? Maybe if you think the cattle have been shot in the unrest and ongoing military violence there, but it seems an impossible task to distinguish between [cows+goats+antelope] and [wetland]
Line 296-301 – biogenic emissions, not OH. See also Nisbet et al. 2023.
Line 301 – livestock growth in 2020s – see Nisbet et al 2025.
Line 305 – use of GRACE – good…
Line 320-332 – yes, despite my scepticism about the livestock data, this paragraph sounds plausible. But it might be good to add real data about the huge floods in both South Sudan and Sudan (e.g. 2020).
Line 318 – repeat - I really don’t think you can split livestock and wetland emissions in the Sudd. Intricately intermixed and interwoven, and I suspect d13C is the same.
Line 324 – note that while the water in the Sudd is from the White Nile (and Lake Victoria was pouring out at record highs between April and July 2020), the Blue Nile is also a factor. In 2020 the Blue Nile Flood level reached over 17m at Khartoum. When you fly over the join at Khartoum you can see how the cloudy White Nile’s steady capacitor discharges into the drk Blue Nile’s dramatic peak flow and very small low-water river flow. Khartoum is at 381m altitude, while Malakal is at 385m and Bor is at 407m. so the Blue Nile flood backs up the White for months until the Ethiopian flood subsides in September…Yes, the Blue Nile has been affected by the enormous GERD dam filling in Ethiopia, but nevertheless in 2020 the Blue Nile would have had quite an impact on the lower northern part of the Sudd wetlands.
Line 343 – when it’s wet the trees have lots of leaf for browse and the palatable grasses grow tall – livestock eat when it’s wet, and when its dry they starve, so the co-seasonality of livestock and wetland is to be expected.
Line 343 – not yet much rice in Africa though production is growing fast.
Line 368 – Siberian gas production is (was) seasonal – used to do maintenance in high summer, and then fill up German pipes in the last quarter. See Reshetnikov paper.
Line 393 – assuming no OH change….NO, you can’t assume that. It clearly changed but we don’t know by how much. You can say you left it out of the calculation, and then make some generalised remarks to say if OH went down by say 10% in mid 2020, then the atmospheric methane burden would go up by X Tg. etc etc.
Line 399 – 403 – maybe here you could mention the failure of models to fit the measurements by Shaw et al, and the need to develop better models!
I found figure 8 a bit disconcerting. Maybe some attention also to Russia?
Some References to consider
Bouarar, I., et al. 2021. Ozone anomalies in the free troposphere during the COVID‐19 pandemic. Geophysical Research Letters, 48(16), p.e2021GL094204.
Drinkwater, A., et al. 2023. Atmospheric data support a multi-decadal shift in the global methane budget towards natural tropical emissions, Atmos. Chem. Phys., 23, 8429–8452, https://doi.org/10.5194/acp-23-8429-2023.
France, J.L., et al. 2022. Very large fluxes of methane measured above Bolivian seasonal wetlands. Proc Nat Acad Sci USA, 119(32), p.e2206345119.
Lunt, M.F., et al. 2021. Rain-fed pulses of methane from East Africa during 2018–2019 contributed to atmospheric growth rate. Environ Res Lett, 16, p.024021.
Michel, S.E., et al. 2024. Rapid shift in methane carbon isotopes suggests microbial emissions drove record high atmospheric methane growth in 2020–2022. Proceedings of the National Academy of Sciences, 121(44), p.e2411212121.
Morgenstern, O., Moss, R., Manning, M., Zeng, G., Schaefer, H., Usoskin, I., Turnbull, J., Brailsford, G., Nichol, S. and Bromley, T., 2025. Radiocarbon monoxide indicates increasing atmospheric oxidizing capacity. Nature Communications, 16(1), p.249.
Nisbet, E.G., et al. 2023a. Atmospheric methane: Comparison between methane's record in 2006–2022 and during glacial terminations. Global Biogeochemical Cycles, 37(8), e2023GB007875.
Nisbet, E.G., 2023b. Climate feedback on methane from wetlands. Nature climate change, 13, 421-422.
Nisbet, E.G. 2025. Practical paths towards quantifying and mitigating agricultural methane emissions. Royal Soc Proc A , 481, p. 20240390).
Palmer, P.I., et al. 2023. Drivers and impacts of Eastern African rainfall variability. Nature Reviews Earth & Environment, 4, 254-270.
Peng, S., et al. 2022. Wetland emission and atmospheric sink changes explain methane growth in 2020. Nature, 612, pp.477-482.
Reshetnikov, A.I., Paramonova, N.N. and Shashkov, A.A., 2000. An evaluation of historical methane emissions from the Soviet gas industry. J Geophys Res: Atmos, 105, 3517-3529.
Shaw, J.T., et al. 2022. Large methane emission fluxes observed from tropical wetlands in Zambia. Global Biogeochemical Cycles, 36(6), p.e2021GB007261.
Zhang, Z., et al 2024. Ensemble estimates of global wetland methane emissions over 2000–2020. Biogeosciences, 22, 305–321, https://doi.org/10.5194/bg-22-305-2025.
Citation: https://doi.org/10.5194/egusphere-2025-1554-RC1
Data sets
Replication data for: Trends and seasonality of 2019–2023 global methane emissions inferred from a localized ensemble transform Kalman filter (CHEEREIO v1.3.1) applied to TROPOMI satellite observations Drew C. Pendergrass et al. https://doi.org/10.5281/zenodo.15120760
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 208 | 27 | 8 | 243 | 5 | 7 |
- HTML: 208
- PDF: 27
- XML: 8
- Total: 243
- BibTeX: 5
- EndNote: 7
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1