the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Atmospheric data support a multi-decadal shift in the global methane budget towards natural tropical emissions
Alice Drinkwater
Paul Palmer
Liang Feng
Tim Arnold
Sylvia Michel
Robert Parker
Hartmut Boesch
Abstract. We use the GEOS-Chem global 3-D model and a Maximum A Posteriori inverse method to infer regional methane emissions and the corresponding carbon stable isotope source signatures, 2004–2020, across the globe using in situ and satellite remote sensing data. Over our study period, we find consistent evidence from both atmospheric CH4 datasets of a progressive increase of methane emissions at tropical (30° N to 30° S) latitudes (+3.80 Tg/yr/yr), accompanied by a progressively lighter atmospheric δ13C signature, consistent with increasing natural emissions. The satellite remote sensing data provide evidence of higher spatially resolved hotspots of methane that are consistent with the location and seasonal timing of wetland emissions, limiting the hypothesis about the hydroxyl radical (OH) sink for methane playing a significant role in observed global growth in atmospheric methane. We find that since 2004, the largest growing regional contributions (2004–2020) are from North Africa (+19.9 Tg/yr), China (+21.6 Tg/yr), and Tropical South America (+14.2 Tg/yr). To quantify the influence of our results to 10 changes in OH, we also report regional emission estimates using an alternative scenario of a 0.5 %/yr decrease in OH since 2004, followed by a 5 % drop in 2020 during the first COVID-19 lockdown. We find that our main findings are robust against those year-to-year changes in OH.
Alice Drinkwater et al.
Status: final response (author comments only)
-
RC1: 'Comment on acp-2022-561', Anonymous Referee #1, 10 Oct 2022
Interactive comments on “Atmospheric data support a multi-decadal shift in the global methane budget towards natural tropical emissions”
General comments
This paper investigates the post-2007 global increase of methane in the atmosphere. It presents two Bayesian invers methods to estimate both, regional CH4 emissions and its δ13C signatures. Moreover, it examines the impact of a global OH reduction on the emission estimates.
The applied methods and the outcome of the study is of interest to the scientific community and fits within the scope of the ACP journal. The paper provides interesting findings and is in general well written in terms of language. Unfortunately, the paper is poorly structured and often lacks precision. I suggest to accept the paper for publication with major revision, which is mainly due to the poor introduction and missing explanations of what is being studied.
Specific comments
Abstract
The abstract is short and concise, and contrary to the introduction (see comments below), provides a comprehensive overview of this paper and its results.
1. Page (P) 1 Line (L) 1: “…a Maximum A Posteriori inverse method…”. Please also mention the second inverse method, i.e. the Ensemble Kalman Filter.
Introduction
The introduction needs to be better structured. It lacks precise explanations (especially regarding the application of stable isotopes) and exact numbers. It should be clearly explained what is being studied and why stable isotopes are being used. I suggest to put more afford into this first part (see specific comments below).
2. P1 L16-18: Here the authors only explain, that CH4 is a GHG. I suggest to briefly state why CH4 is an important greenhouse gas (e.g. compared to CO2, secondary/indirect radiative forcing) and include references.
3. P1 L18-19: Be careful with the terms "anthropogenic" and "biogenic". Biogenic CH4 emissions are also anthropogenic (e.g. emissions from livestock, rice cultivation, landfills, etc.). There are also non-biogenic natural emissions (e.g. from volcanic eruptions, wild fires). Better: use "biogenic", "thermogenic" and "pyrogenic" and further distinguish between "natural" and "anthropogenic". This mistake (and the associated misinterpretation of the results) is found throughout the paper. Please check the entire text.
4. P1 L19: Please mention that the oxidation by OH is the main loss of atmospheric CH4. This explains why later in your study you only examine the role of isotopic fractionation by OH oxidation and neglect all other loss processes. Moreover, the adjective “small” in front of “stratospheric loss” is not informative, since you do not give any other orders of magnitude for CH4 loss processes.
5. P2 L21: The term “unprecedented values” is not precise. Please give exact values.
6. P2 L22: What do you mean by “lighter CH4”? Please use correct terms, such as “isotopically lighter” or “depleted in the heavier stable isotope 13C”. Please also indicate, if you mean the isotopic signature of globally averaged atmospheric CH4. Again, please be more precise here.
Note: The isotopic signature δ13C should also be explained at some point in the paper: why do you use isotopic signatures and what can we infer from them? In addition, please mention to which standard the carbon isotopes are reported to (i.e. relative to the Vienna Pee Dee Belemnite standard). This can also be done in the Section 2. I also suggest noting once, that throughout the text δ13C refers to δ13C in CH4.
7. P2 L22: A change towards more negative δ13C possibly indicates a shift towards biogenic emissions. However, another explanation, could also be a reduction of isotopically heavier pyrogenic emissions. I suggest writing that this shift could indicate a change in the contributions of the emission sources.
8. Moreover, since you also investigate the role of OH oxidation, please explain the influence of isotopic fractionation during the sink processes in the atmosphere. How does the isotopic fractionation change the δ13C signature of atmospheric CH4?
9. P2 L24: Explain why you start to investigate emissions from 2004 onwards. Until which year?
10. P2 L24: What means “short-term” in this case?
11. P2 L27: Again: anthropogenic emissions can also be microbial. Please describe which emissions you mean exactly, I guess natural (tropical wetland) emissions.
12. P2 L27-28: Does this sentence describe the motivation of your study or the conclusion from your results? Rewrite it to fit the context of the introduction or move it to the end of your conclusion or outlook. Moreover, please add a reference, if this shift “from anthropogenic to microbial sources” is not your conclusion.
13. P2 L35. It is not explained, what the authors mean by “From the source signature of a region”. Please use correct and precise terms, i.e. “the isotopic source signature” and check throughout the entire text.
14. P2 L38-39: Why are changes in OH unlikely to play a dominant role? I also suggest mentioning that the role of OH is highly uncertain as atmospheric OH is very difficult to quantify.
15. P2 L41: Please write “isotopic fractionation” instead of “fractionation”.
16. P2 L41-42: Do you explore the impact on the δ13C signature? If so, please add this information to your sentence.
17. P2 L46: How can this be corroborated by δ13C studies? How do you investigate the assumption of reduced OH in 2020, due to lower NOx emissions?
18. P2 L49: Here you only mention OH, although this is just a part of your analysis. Please write exactly what else you analyse and present in Section 3.
19. Up to this point, it is not clear to me what exactly the scientific questions are, and how the authors intend to investigate them. I suggest to restructure the introduction and work out or include:
- what the scientific questions are (e.g. reason of CH4 short-term variations after 2004, the role of OH reduction, CH4 increase in 2020 due to reduce NOx emissions)
- that you are using two different approaches of inverse modelling
- that you perform these inversions for CH4 emission fluxes and δ13C signatures
- that you use in-situ and satellite-derived remote sensing observations for the inversion
- indicate if you only focus on certain regions (i.e. low, mid or high latitudes?) or on all
- the usage of stable isotopes here.
Data and Methods
In general, it would be easier to follow this section if the methods were briefly introduced (only 1-2 sentences, see comments above), and the need for observations to constrain emissions of different regions was pointed out. Moreover, Section 2 lacks a description of the sensitivity simulations performed to investigate the role of OH (although this is done in the last paragraph of Section 3). I suggest to restructure the paper and describe the simulations in Section2, including what is meant by “control calculation”.
20. P3 L52: What do you mean by “total” CH4 emissions? Total emissions in the respective regions you study or total global emissions?
21. P3 L52: Please make clear that you mean δ13C of the CH4 emissions, if you have not yet in the introduction (see also comment 7 on P2 L22).
22. P3 L61: Is there a constant offset due to the different heights where measurements were performed?
23. P3 L63: Here you start to describe δ13C measurements, although you have introduced them before (L62). From the text above, it was not clear that you only described the CH4 observations. Please be more precise above.
24. P3 L65: Please specify to which standard carbon isotopes are reported (i.e. VPDB).
25. P3 L78: I am not sure where the uncertainty of 1.2 % comes from.
26. P3 L80: Please give more details about the GLOBALVIEW data. Are they in-situ or remote sensing measurements, or both? In line 81 you mention uncertainties of in-situ measurement. I suggest to move the term “in situ” to the first part of the sentence (i.e. “GLOBALVIEW in-situ CH4 and CO2 data …”) and also indicate if data are collected at surface level?
27. P4 L86: How is this “driven”?
28. P4 L88: Why do you use EDGAR v4.3.2, which provides only data until 2012? The latest EDGAR emission inventories provide emission estimates until 2015 (v5) and 2018 (v6). Do you repeat emissions after 2012 annually?
29. P4 L90: Are agricultural waste burning emissions included in the GFED or EDGAR inventory? Please give information until which year emission estimates of the respective inventories are available.
30. Table 1: This table shows the global methane emissions from Saunois et al. 2016. Why are these emissions shown, although not applied to any inversion? Is there any reference in the main text? Moreover, there is a more recent publication (i.e. Saunois et al 2020), which provides emission estimates until 2017. Would it be possible to add the annual mean a priori emissions of each sector to the table? Maybe the authors could also compare their results with the emission estimates provided by Saunois et al 2020.
31. P4 L95-96: In my opinion, the single example "(e.g. coal mines)" does not reflect the many different isotopic signatures within the individual sectors. Either skip the example here and discuss the different isotopic source signatures later, or give more examples. (e.g. wetland and pyrogenic emissions) and find references.
32. P4 L99: If you mention the isotopologues here, please briefly describe why you need to convert isotopic ratios into isotopologues.
33. P4 L102-103: Is the atomic chlorine 3D field also derived from the full-chemistry simulation? Please also indicate the time period for which these 3D fields are available.
34. P4 L104: What about transport into soils, is this considered?
35. P4 L105: Please also compare the 9.73 years to other CH4 lifetimes reported in the literature.
36. P4 L113: From which year is this restart file? When do you start your simulation? What is the coarser resolution of the previous GEOS-Chem model simulation?
37. P5 L143-144: Please explain the basis on which you selected these uncertainty values.
38. P6 L158-159: Do I understand this correctly: Each source signature of each sector with the respective region is perturbed by 20 ‰? Please rephrase the sentence to make this clear to the reader.
39. P6 L166: Are you performing this a posteriori simulation using the improved CH4 flux estimates and δ13C, for further analyses? If yes, what do you intend to analyze with this a posteriori simulation and for which time period is this simulation performed? I am not sure if I am understanding this correctly.
Results
In this section, the results are well presented. The figures are comprehensive and qualitatively well presented (with the exception of Figure 4, see comment 50). However, an in-depth interpretation is sometimes missing. I suggest discussing the results here in detail and renaming this section to "Results and Discussion".
40. The authors often use different and ambiguous terms to distinguish between the observational data. For example, in:
- Figure 2: the terms “in situ” and “ground-based” are used to describe the same data in only one figure caption.
- Figure 4: the terms “NOAA” and “δ13C” are used for CH4 and isotope measurements, respectively, although δ13C data are also measured at NOAA sites.
Please ensure that you are using consistent terms for the observational data throughout the text and in the figures and figure captions.
41. In the results section, the authors often refer to figures in the supplements. This direct reference makes me search for them in the text. I would suggest moving some plots that are directly described to the main text. And/or take the focus away from the figures, which remain in the supplements, e.g. by using phrases like " (see Supplements for absolute emission values)” on P7 L186 or “(described in more detail in the Supplements)”. Moreover, the order in which figures are numbered does not correspond to the order in the text.
42. P7 L187: As already mentioned above (see comment 40) please be consistent with naming of the observational data. Here only the term “NOAA” is used. What about the measuring site in Siberia (NIES), are data from this site included? I suggest to define appropriate names for both data sets in Section 2 and stick to them throughout the text.
43. P7 L191-192: The authors describe the emissions of the year 2020. But how do emissions evolve over the time period (2004-2020) investigated here? Emissions in Europe in 2020 have increased, too. What could be the reason? What happened in 2019 in Europe?
44. P7 L204: Why do we see those differences between the two results? Is there an explanation for this?
45. P7 L208: Why do we see this increase in China? Other studies suggest increasing emissions from coal mining here over the last years. Please also discuss this with respect to your δ13C a posteriori results.
46. P8 L211-217: Please name the key message from those results, i.e. statistical analyses show improvement of the emission estimates compared to a priori emission estimates.
47. P8 L219ff: In Figure A7 it can be seen, that the inverse modelling reduced the uncertainty of the δ13C signatures in each region. Moreover, the resulting δ13C signatures and their seasonal variability correspond to the sources which are expected in the respective regions. This is nicely explained by the authors. However, it could be described in combination with Figure 3, since here the main outcome of the inverse study is shown, e.g. the changes in δ13C over time. Moreover, the focus would not be on Figure A7, which is only shown in the Supplements (see also comment 41).
48. P8 P235-238: Why do we know that this shift in 2012 towards heavier isotopes is due to anthropogenic activities? Which anthropogenic emissions caused this shift? The author could for example support their conclusions with other studies, such as ethane measurements. Is there another explanation for this? Why couldn’t it be caused by enhanced biomass burning?
49. P8 L238-239: The term “heavy” or “light” trend does not sound correct to me. I suggest using “trend towards lighter isotopes” or “trend towards more negative δ13C” v.v. Why is this trend “light” in 2012? In which regions do we see the “heavy” trend in 2008? Why?
50. Figure 4:
- Please do not assign axis labels by arrows.
- The figure caption does not say, what is shown on the right side of this plot. I suggest to divide the plot into two subplots called a) and b) and explain exactly what is being shown in subplot b).
- here the authors mix up fluxes and emission rates (which is “mass * time-1 * area-1” and “mass * time-1”, respectively). Please also check throughout the text if the terms “flux” and “emission rate” are used correctly.
51. P9 L271ff: This part should be moved to Section 2, where the authors describe the methods and model simulations.
52. P9-10 L284-285: Why is the reduction only seen in high-emitting regions?
Conclusions
The conclusion summarizes the paper comprehensively.
Technical correction
P1 L5-6: “The satellite remote sensing data provide evidence of higher spatially resolved hotspots of methane…”. This sentence does not make sense to me. Do you mean: “The higher spatially resolved satellite remote sensing data provide evidence of methane hotspots …”?
P3 L65: Instead of “The geographical locations of the in-situ data…” I suggest: “The geographical locations of the in-situ measuring sites are shown in Figure 1.” And maybe begin a new sentence: “They present a subset …” However, I am not quite sure what you are trying to say with this sentence.
P3 L72: Either “carbon dioxide” or “CO2”. Not both.
P3 L76: “The analysis shows…” or “Analyses show…”
P4 L88: There is a missing “v”: EDGAR v4.3.2
P4 L94: Please insert “…isotopic source signatures…”
P4 L94: "which provide" or "dataset/database from Sherwood et al., which provides"
P4 L113: Please change “older” to “previous”. I think the “and “ is wrong here.
P5 L121: Please use the term “grid boxes” instead of “grid squares” (please check throughout the entire text)
P5 L121: Please change to “…correspond to the location of the sampling sites…”
P5 L131: Please be more precise and write “CH4 fluxes” and not only “fluxes”.
P5 L134-135: This sentence sounds complicated to me. I suggest to write: “The MAP solution and the associated a posteriori uncertainty is described as:
FORMULA,
using the conventional that lower-case and upper-case variables denote vectors and matrices, respectively, and where …”
P5 L141: I think the “either” needs to me removed
P5 L145: “∂y/∂x” should be “∂y/∂x”
P5 L148: There is a missing dot: “…normal. The individual…”
P6 L160: δ13C is always given in ‰
P6 L164: “The a posteriori CH4 fluxes…”. Please check throughout the text and specify which fluxes (i.e. CH4) you mean. Even though it is clear from the context, I would prefer the authors to indicate this correctly. Likewise, in L165 please write “isotopic source signatures” or “δ13C source signature” and check throughout the text.
P6 L165: Please add “…horizontal grid”
P8 L 228: “…assumes that sources are…” Please specify which sources.
P8 L241: Please add “atmospheric growth rate of CH4”
P10 L293: Here you write “post-2007 increase”. In the introduction it is written “post-2006”. Please be consistent.
Table 1: Is “magnitude” in this context correct? I would prefer anything like “global mean emissions”. Please also add “(‰)” to the table, as well as the uncertainty range for the isotopic source signatures.
Table 2: I suggest to move this table to the supplements, since it does not provide any new information.
Figure 2A: missing space in figure caption in line 1”…GOSAT data (blue line)…”
Figure A3. double “n” in line 2: “..shown…”
Citation: https://doi.org/10.5194/acp-2022-561-RC1 -
RC2: 'Comment on acp-2022-561', Anonymous Referee #2, 19 Oct 2022
The paper by Drinkwater et al. studied changes in regional methane emissions and d13C source signatures over the period 2004-2020, using two inversion frameworks that assimilated in-situ and GOSAT observations respectively. They found a progressively emission increase from tropical regions accompanied by lighter d13C signature, and concluded a multi-decadal shift in global methane budget towards tropical natural emissions (wetland emissions notably). The subject of the paper fits in the long-term research interests in the community regarding the decadal changes in methane budgets and underlying drivers. In general, I find the paper interesting to read, and relevant to the scope of ACP. However, there are a few major concerns that may weaken the robustness of the main conclusions, and hopefully can be addressed in revisions.
One of the biggest issues is that the inversion results presented in this study lack independent evaluation. While the two inversions based on in-situ and GOSAT observations respectively do show some consistency in the overall emission trends at the global scale and large latitudinal bands (Fig. 4 & Fig. A2), there are clear discrepancies between the two inversions for big regions regarding emission increments after inversion (e.g., Boreal North America & China in Fig. 2), magnitudes of posterior emissions and emission trends (e.g., Temperate North America & Tropical Asia in Fig. A1). The good model-data agreement at some selected sites as shown in Fig. A4 is expected, as these stations were assimilated in inversions and most of them are marine boundary layer stations, where observations are normally reproduced by models. In fact, I would expect poor model performance at some difficult sites such as KRS and BKT even though they were assimilated. I suggest the authors examine model performance at all sites assimilated, and if possible, include non-assimilated sites or observations from other platforms like aircraft campaigns or TCCON sites, so as to evaluate the robustness of the inversion results for big regions.
In particular, I notice that the emission trends for China since 2012 are somehow higher than the estimates from several recent papers (Lines 207–210, 0.72 Tg/yr and 1.34 Tg/yr inferred from in-situ and GOSAT data versus 0.36 Tg/yr from Sheng et al. 2021, also check out the papers by Liu et al. 2021 and Zhang et al. 2022 and references therein). The emission trend inferred from GOSAT data (1.34 Tg/yr) seems beyond the upper limit of previous estimates for the similar period, and contradicts with the recent slowdown of emission increase in China (Liu et al. 2021). Do you have any explanation for that?
For the optimization of d13C signature, I don’t quite understand the methodology. It’s not clear whether regional methane fluxes and d13C signatures were solved simultaneously or sequentially? According to the description of methodology in Lines 154–163, it seems that the solution of regional d13C signatures relies on the solution of regional emissions. I wonder how much errors in estimates of regional emissions would impact the solution of d13C signatures. Can we trust the results presented in Fig. 3 if the emission trends detected for certain regions are not robust?
The lighter d13C signature in tropical regions doesn’t necessarily imply an increase in natural emissions (wetland emissions in particular). The tropical regions are known for their agricultural practices and related methane emissions, and recent studies suggested emission increase from agricultural sectors in tropical countries (Stavert et al. 2022; Zhang et al. 2022b), which could also lead to lighter d13C signature according to Table 1. Is it possible that agricultural sectors also had substantial contribution to the recent trends in tropical emissions and d13C signature? How much was it compared to the contribution from wetland emissions? In the abstract the authors claimed that “the satellite remote sensing data provide evidence of higher spatially resolved hotspots of methane that are consistent with the location and seasonal timing of wetland emissions” (see Lines 318–320 as well), which is not clearly shown in this paper. The authors also cited a few papers that reported large CH4 anomalies or trends in Eastern Africa or Amazon, which seems to confirm their conclusions. But it’s not clear how much wetland emissions from tropical regions contributed to the signals detected in this paper.
The use of climatological OH fields for the reference runs is fine, given the large uncertainty in the long-term OH trends and variabilities. The authors should be aware of the range of uncertainties among recent studies (see e.g., Turner et al. 2017; Naus et al. 2021; Patra et al. 2021; Zhao et al. 2020 etc. and references therein), and discuss how this could impact methane budgets and variabilities. The paper by Lan et al. 2021 cited in the introduction (Lines 39–41) seems to deny the hypothesis proposed by Turner et al. 2017. Why did you choose decreasing OH by 0.5%/yr for the sensitivity test that followed this hypothesis? The choice of 5% uniform drop in OH for 2020 is also problematic, given the large spatial and temporal variability in OH changes in response to reduction in NOx emissions due to COVID lockdown.
References:
Liu et al., Recent slowdown of anthropogenic methane emissions in China driven by stabilized coal production. Environ. Sci. Technol. Lett. 8, 739–746 (2021).
Naus et al., A three dimensional-model inversion of methyl chloroform to constrain the atmospheric oxidative capacity. Atmos. Chem. Phys., 21, 4809–4824 (2021).
Patra et al., Methyl Chloroform Continues to Constrain the Hydroxyl (OH) Variability in the Troposphere, J. Geophys. Res.-Atmos., 126, e2020JD033862 (2021).
Sheng et al., Sustained methane emissions from China after 2012 despite declining coal production and rice-cultivated area. Environ. Res. Lett. 16, 104018 (2021).
Stavert et al., Regional trends and drivers of the global methane budget. Glob. Change Biol. 28, 182–200 (2022).
Turner et al., Ambiguity in the causes for decadal trends in atmospheric methane and hydroxyl, P. Natl. Acad. Sci. USA. 114, 5367–5372 (2017).
Zhang et al., Observed changes in China’s methane emissions linked to policy drivers. P. Natl. Acad. Sci. USA. 119, e2202742119 (2022a).
Zhang et al., A 130-year global inventory of methane emissions from livestock: Trends, patterns, and drivers. Glob. Change Biol. 28, 5142-5158 (2022b).
Zhao et al., On the role of trend and variability in the hydroxyl radical (OH) in the global methane budget, Atmos. Chem. Phys., 20, 13011–13022 (2020).
Citation: https://doi.org/10.5194/acp-2022-561-RC2 - AC1: 'Comment on acp-2022-561', Alice Drinkwater, 09 Dec 2022
Alice Drinkwater et al.
Alice Drinkwater et al.
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
508 | 199 | 15 | 722 | 6 | 12 |
- HTML: 508
- PDF: 199
- XML: 15
- Total: 722
- BibTeX: 6
- EndNote: 12
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1