the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 23 Dec 2025
Reaction-transport modelling of methane cycling beneath the Greenland Ice Sheet
Abstract. Glacial and ice sheet advances have buried large amounts of organic matter (OM), which under anoxic subglacial conditions can be microbially converted into methane (CH4). Although CH4 emissions have been observed at glacier margins, the capacity of subglacial environments to sustain such fluxes remains uncertain. To address this, we developed a reaction–transport model (RTM) to simulate CH4 production, transformation, and transport in sediments beneath warm-based regions of the Greenland Ice Sheet (GrIS) margin. The model explores a wide range of environmental conditions, including sediment thickness, OM quantity and reactivity, O2 availability, and methanotrophic activity.
Model simulations show that subglacial sediments are largely anoxic. Oxygen (O2) penetration into subglacial sediments is generally restricted to the upper few tens of centimetres, with an average penetration depth of 22.8 cm. Microbial OM degradation and aerobic CH4 oxidation (AeOM) represent the main O2 sinks. Their relative contributions vary with CH4 availability. AeOM dominates in methane-rich sediments, whereas OM degradation prevails in methane-poor environments. Modeled depth-integrated methanogenesis rates range from 0.1 to 1600 mmol-CH4 m-2 yr-1 (mean 73 mmol-CH4 m-2 yr-1) and are primarily controlled by OM reactivity, with sediment depth and OM concentration exerting only a small secondary influence. This sensitivity of CH4 production rates to OM reactivity can produce sharp thresholds, where small decreases in reactivity strongly suppress CH4 fluxes. A highly variable fraction of the generated CH4 is consumed by AeOM within the shallow oxygenated zone, and is controlled by OM reactivity and the AeOM rate constant. Resulting net diffusive CH4 fluxes can range between 0–234.7 mmol m-2 yr-1. Results show that even shallow sediments (<1 m) can sustain a significant CH4 release into the subglacial environment when highly reactive OM is available, while oxidation efficiency tends to decline in thick, OM-rich deposits.
Comparison with field measurements of CH4 export data from southwest GrIS catchments suggests that observed fluxes could be already sustained by subglacial sediments that contain as little as 0.6 wt% of relatively unreactive OM assuming a catchment sediment cover of 10 % with sediment depths of 9 m.
- Preprint
(2613 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- CC1: 'Comment on egusphere-2025-5506', Angelos Theodorou, 08 Jan 2026
-
RC1: 'Comment on egusphere-2025-5506', Anonymous Referee #1, 04 Feb 2026
It was my pleasure to read “Reaction-transport modelling of methane cycling beneath the Greenland Ice Sheet”. This paper is an important step in understanding the methane fluxes observed from the Greenland Ice Sheet. As a modelling study, it necessarily makes some simplifying assumptions. Most critically, it assumes that oxygen is the only relevant terminal electron acceptor. The authors recognise this as unlikely, but quantifying the other electron acceptors would be difficult given the current state of science. The paper does give a sense that you can simulate almost any methane flux from the Greenland Ice Sheet by varying the relevant parameters within plausible boundaries. But it does, importantly, give us the tool to match this large range of methane fluxes to the conditions that might produce them. Thus, it offers something to refer back to as variable methane fluxes are observed. Reverse engineering geochemical fluxes under glaciers is tricky and I do think this paper contributes to solving this problem. In general, I think the effort here is solid and the paper should be published after moderate to major revisions.
The most significant issue is that Section 3 needs to be restructured. This section is supposedly methods, but it includes long digressions into background information that offer some sort of justification for the methods selected. This makes it almost impossible to properly read as section 3 as a methods section. All of the material that isn’t methods (which, by word count, is most of it) needs to be pulled into the introduction and ideally shortened. The authors could consider a separate “justification of model parameters” section, but they shouldn’t take this too far. We all know that this kind of modelling is only possible if simplifying assumptions are made.
I also think there needs to be substantial revision to section 4.5, where upscaling calculations are made. Because the authors find their model can produce almost any value depending on assumptions of sediment thickness and organic matter reactivity, plugging these into real glaciers gives fluxes that range over 6 to 7 orders of magnitude. This is not particularly useful as a result. Instead, the authors should present the actually measured values from these glaciers in terms of their model. I.e. explain the range of conditions that might have produced the range of measurements. It would be especially useful if they could analyse seasonal variability through these parameters.
I have some line-by-line comments below:
Line 10: My heart sank when I saw from the first line that the authors were planning to abbreviate organic matter throughout the paper. The use of nonstandard abbreviations of this sort can make the paper significantly harder to read, as the reader is always having to go back and look for the meaning of OM, RTM, AeOM, GrIS, TEAs, etc. I know this a personal pet peeve and other readers might not be as bothered but consider whether this actually necessary.
Line 14: Remove “a wide range of environmental conditions, including”. This is, in fact, the complete list of parameters that you consider in this paper.
Line 33: I might add the word to “potentially” to “burying vast stocks”. The literature cited is fairly speculative, and since glacier growth can be highly erosive, the extent of organic matter burial remains uncertain.
Figure 1: I know this is a conceptual figure, but the image of a meltwater channel equal in thickness to the subglacial sediment package is wrong – especially when you later consider sediment packages up to 15 m thick! Also adjust the text to not overlap with the drawing and change the font to match the font used by the journal.
Lines 70-72: This is not needed here. Equations, etc., should appear in the main text.
Line 95: I thought the top was water, per the conceptual model.
Line 96: There certainly would not be cryoturbation in the subglacial environment! Cryoturbation is caused by seasonal freezing and thawing. The bed of the Greenland Ice Sheet is variably frozen or warm, but it does not vary seasonally. However, there very well could be sediment shear from the driving stress of the glacier overhead. I think it’s okay ignore sediment shear, but the reference to cryoturbation has me worried.
Lines 98-99: You generally should avoid calling out sections in advance.
Line 101: Remove the word “environments”.
Line 110: Salinity wouldn’t exactly be zero, but maybe it’s close enough to zero that it might as well be?
Section 3.3: This is the first of several sections composed more of background information than strictly methods. I would try to take the discussion about electron acceptors, methanogenesis pathways, and previous work into the introduction and have this section just be a description of how the model was implemented in the study (i.e. have the subsection start at line 149). I also think this goes into far too much detail justifying why simplifying assumptions were made. Readers will perfectly well understand why you can’t include every possible electron acceptor or methanogenesis pathway into the model.
Line 117: You can just have “(Table C1)”.
Line 119: I had to go back and look up what TEAs meant, highlighting my earlier point about acronyms.
Line 124: Sulphur is the one electron acceptor that you don’t explore in any further detail here. Of all the electron acceptors, sulphur is by far the most abundant in Greenland Ice Sheet waters.
Line 126: “Statham et al.”; I assume The Cryosphere doesn’t expect three author lists.
Line 137: There’s a missing link here or the logic of the paragraph needs to be reordered. The terms acetoclastic and hydrogenotrophic need context.
Line 138: Does The Cryosphere allow for citation of papers in review?
Line 165: I wonder to what extent the normal aging process might not apply or might not apply in the same way to the subglacial environment.
Line 166: I don’t think it’s necessary to call out upcoming sections in advance.
Lines 168-186: This is also introductory material. The methods section should focus on the methods.
Line 170: Again, you don’t need a preamble before citing a table.
Line 171: Parentheses should only be around the year when the authors names form part of the sentnece.
Line 180: You do not define AOM. (Again, too many acronyms.)
Lines 180-183: To me, these values for sulphur are not particularly low. What are you comparing this to here? As far as I know, there’s never been any evidence for sulphur reduction under the Greenland Ice Sheet. But I don’t think that’s for lack of sulphur; sulphate is one of the major anions found in subglacial waters in Greenland and in glacial environments in general. There’s a missing conceptual piece here. It might be the mechanochemistry angle. Subglacial environments are generally highly oxidative.
Line 185: Again, can you site manuscripts in preparation in The Cryosphere?
Line 194: Not necessary.
Section 3.6.1: It might be worth pointing out the glacial forefield in the study area also consists of a mix of rocky and sediment mantled regions.
Line 241: This is very unlikely. It’s far more plausibly that there are patches where organic matter is highly concentrated due to minimal erosion for whatever happenstance of the subglacial environment alongside vast areas with almost nothing. On the other hand, uniformity is far easier to model (especially in a nonarbitrary way).
Line 247: RCM is undefined.
Line 254-272; Figure 2: I needed to re-read this several times to understand the main point. State the main point – i.e. how all of this affects your model – much more clearly and consider omitting some of the detail. The figure could maybe go to the supplement.
Line 287: Is there a good reason not to have the temperature increase by 0.5°C over the 15 m sediment package? Unlike the other simplifying assumptions, I don’t see how this would make the model any harder to run.
Line 323: The pressure in the subglacial environment is usually considerably below ice overburden pressure, at least during the summer when water is discharging from the proglacial environment. You might want to look into borehole studies to see what is actually more realistic. This fluctuating pressure would also have implications for the stability of clathrates, should they form. I.e. if a clathrate formed in winter, when water pressure was high, it might well melt in summer as pressures drop.
Section 3.6.5: This should at least briefing discuss all the work Jon Telling’s group has been putting out on the mechanochemistry of glaciers, which may be critically important to why they’re so highly oxidative (e.g. Stone et al., 2023, Gill-Olivas et al., 2024, etc.).
Line 385: This goes back to the hard bed / soft bed question. In the areas with thick sediment packages, methanogenesis is potentially extensive. But it remains unclear how prevalent those are – i.e. it could be mostly a hard bed with only patches of thick sediment.
Line 394: I keep coming back to this, but to me GSA is Geological Society of America or perhaps General Services Administration.
Lines 396-400: I find this section very confusing. You first say that saturation is rarely reached and then say that it is often exceeded. Which is it?
Line 399: I would not point the reader back to the methods here. If the reader goes back there, they will also find the suspiciously even number of 50 mM proposed without any citation or justification.
Line 400: That is a huge range.
Line 439: Russell has a much smaller catchment than Isunnguata Sermia or Leverett. Perhaps relevant to Figure 6.
Line 448: Missing “The”.
Line 448: The dominant role of organic matter reactivity is also clear just by scrutinising your figures. Are there not any datasets available to put greater constraints on this? I.e. studies of particulate organic matter in the outflow?
Line 454: This again brings us back to the question of characterising sediment thickness under the ice.
Line 490: You need to say how this estimate is derived. Cowton got to 600 km2 by just assuming the active area started below 1200 m. Did you make a similar assumption here? If so, I find this assumption dubious. Anywhere where meltwater contacts the bed is “active”, cutting the hydrologic catchment off at a certain elevation doesn’t make any sense unless there is no surface melt above that elevation.
Line 494: You must be using the entire catchments not the hydrological active ones to get these values. Otherwise Isunnguata Sermia would only be twice Leverett, not an order of magnitude larger.
Line 495: I know the Hatton paper is still under review, but how do they get a flux ranging from 0.002- 28.9 Mg per year. That’s a 4 order of magnitude range. This must be a measurement range not error bars on a single estimate of flux. If there really is a 4-order-of-magnitude measurement range at Isunnguata Sermia, that’s actually something very meaty that your model could explore. When do you get the low values, when do you get the high values? Over what temporal and spatial scale does this vary? If your model is right, it would imply that the glacier is accessing different parts of the bed that either have thicker sediment or more reactive sediment over these timescales. Or perhaps it’s simply a function of water residence time per figure 6.
Line 498: This needs to be completely rethought. You can’t make any “suggestion” based on a model output that varies over 6 orders of magnitude depending on model input.
Line 501: I’m having trouble following your math here. How do you get from 0.006-0.13 micromolarity multiplied by 3.1-7.9 cubic km of water flux to 0.002-28.9 Mg per year for Isunnguata Sermia? Whether multiplying the top and bottom end members or doing formal error propagation, you cannot get such a large range in the final value. A mistake seems to have been made somewhere.
Line 520: I would remove “or evades”. It seems very unlikely that gas could evade from the subglacial environments of Greenland.
Line 540: Or it may be substantially smaller; you produced values over an almost 7 order of magnitude range. I don’t think this model can be used in this way. I think it can be used to interpret measurements, but not to postulate about fluxes that haven’t been measured.
Lines 544-547: I don’t follow this argument. This distance between moulins in the interior and the subglacial outflow can be 10s of km, I don’t see how potential evasion at the outlet could have anything to do with diffusive flux to the meltwater channel in the interior.
Line 549: There isn’t an unpressurised zone under any flow condition at Isunnguata Sermia, which is highly overdeepened and terminates in a fountain that sprays the subglacial water several meters into the air. There might be under certain conditions at Leverrett, but I find Chandler’s evidence for it (which is based on the behaviour of a tracer gas) less than convincing.
Line 552-557: This is highly, highly unlikely. The routing of subglacial water is entirely driven by the surface slope of the glacier and the shape of the bed. The notion that the large outlets might flow backward against these gradients is incredibly doubtful and isn’t based on anything observed in subglacial hydrology (e.g. in borehole studies). On the contrary, borehole studies and fundamental physics (e.g. Rothlisberger, 1972) suggest that pressure is lower in the large outlets during high flow, so if there were to be any rerouting of methane, it would be to the large outlets, not away from them. But, then again, there is no evidence that the small outlets ever flow up ice overburden pressure gradients either.
I might point out the authors have presented a very good reason why the big outlets have less methane – longer water residence times. This seems fully explanatory as to why the Russell Glacier study and the Christiansen study (which was of a small, minor outlet) both present far higher methane concentrations than Leverett and Russell. It’s unclear why the authors seem to want to run away from this conclusion by coming up with highly dubious explanations of where the missing methane might have gone.
Line 580: This is very repetitive with what you wrote on line 448.
Lines 592-598: I would bring up the water residence time finding here too. In my view this, as much as anything, explains why methane fluxes are sometimes miniscule (especially in large systems like Isunnguata Sermia).
Table C2: The values for the relative proportions of carbon, nitrogen, and phosphorus in subglacial organic matter are from a 1934 study of plankton? Surely there must be some more current and relevant values to go by? Say, studies of arctic soil organic matter and/or plant material?
Citation: https://doi.org/10.5194/egusphere-2025-5506-RC1 -
AC2: 'Revisions in response to comments by Anonymous Referee #1', Philip Pika, 31 Jul 2026
Response to Referee 2
It was my pleasure to read “Reaction-transport modelling of methane cycling beneath the Greenland Ice Sheet”. This paper is an important step in understanding the methane fluxes observed from the Greenland Ice Sheet. As a modelling study, it necessarily makes some simplifying assumptions. Most critically, it assumes that oxygen is the only relevant terminal electron acceptor. The authors recognise this as unlikely, but quantifying the other electron acceptors would be difficult given the current state of science. The paper does give a sense that you can simulate almost any methane flux from the Greenland Ice Sheet by varying the relevant parameters within plausible boundaries. But it does, importantly, give us the tool to match this large range of methane fluxes to the conditions that might produce them. Thus, it offers something to refer back to as variable methane fluxes are observed. Reverse engineering geochemical fluxes under glaciers is tricky and I do think this paper contributes to solving this problem. In general, I think the effort here is solid and the paper should be published after moderate to major revisions.
We appreciate the positive viewpoint of the reviewer and seek to improve clarity and conciseness as detailed below.
The most significant issue is that Section 3 needs to be restructured. This section is supposedly methods, but it includes long digressions into background information that offer some sort of justification for the methods selected. This makes it almost impossible to properly read as section 3 as a methods section. All of the material that isn’t methods (which, by word count, is most of it) needs to be pulled into the introduction and ideally shortened. The authors could consider a separate “justification of model parameters” section, but they shouldn’t take this too far. We all know that this kind of modelling is only possible if simplifying assumptions are made.
We agree that Section 3 previously mixed methodological description with elements of model justification, which made it harder to read as a clear methods section. We have therefore extensively streamlined and restructured the section to improve the separation between the conceptual basis for the reaction network and its mathematical implementation. We also have emphasized the use of the model results as a diagnostic tool to derive inverse constrains on inaccessible environmental information based on meltwater outflow concentrations.
Importantly, however, we retain the conceptual justification as a separate sub-section within the method section. It is not background or introductory material in the conventional sense, but an integral part of the model design. In subglacial environments there is no universally agreed set of dominant biogeochemical pathways, and the relative importance of intermediate terminal electron acceptor processes, methane oxidation pathways, and external methane sources remains actively debated and strongly system-dependent. For this reason, the selection and exclusion of specific reactions is itself a key modelling decision guided by our current understanding, guided by aspects of identifiability and predictive skill.
We therefore consider it necessary to explicitly document and justify these choices within the modelling framework. To address this while improving readability, we have consolidated all process justification into a dedicated sub-section (3.3.1 Selection of biogeochemical processes) placed prior to the mathematical formulation. This section clearly outlines the evidence base and modelling rationale for each included and excluded pathway, using a consistent structure, while the remainder of Section 3 now focuses on the reaction network formulation and implementation.
In this way, we aim to preserve methodological clarity while still transparently documenting the assumptions that fundamentally define the model structure.
I also think there needs to be substantial revision to section 4.5, where upscaling calculations are made. Because the authors find their model can produce almost any value depending on assumptions of sediment thickness and organic matter reactivity, plugging these into real glaciers gives fluxes that range over 6 to 7 orders of magnitude. This is not particularly useful as a result. Instead, the authors should present the actually measured values from these glaciers in terms of their model. I.e. explain the range of conditions that might have produced the range of measurements. It would be especially useful if they could analyse seasonal variability through these parameters.
This is indeed one of the main intentions of this section, and it does work backwards from observations, identifying which combinations of OM content, reactivity, sediment cover, and oxidation reproduce measured CH₄ concentrations at both glaciers. But we agree that the framing is a bit obscured from how the section is currently structured.
In the revised manuscript, we restructure section 4.5 to highlight the measured dissolved CH₄ concentrations at Isunnguata Sermia and Leverett Glacier as the primary constraints, and reframe the analysis explicitly as an inversion (i.e. what subglacial conditions are consistent with these observations?). The broad flux range is retained but reframed as motivation rather than result. We also add a brief discussion of seasonal variability, specifically how changes in discharge affect meltwater dilution and residence time, and how this propagates into the inferred subglacial conditions across the melt season.
I have some line-by-line comments below:
Line 10: My heart sank when I saw from the first line that the authors were planning to abbreviate organic matter throughout the paper. The use of nonstandard abbreviations of this sort can make the paper significantly harder to read, as the reader is always having to go back and look for the meaning of OM, RTM, AeOM, GrIS, TEAs, etc. I know this a personal pet peeve and other readers might not be as bothered but consider whether this actually necessary.
Revised as suggested. Some less used abbreviations (RTM, LG, IS, RU, TEA, GSA, TEA) are now spelled out to improve the reading experience. We kept GrIS for the Greenland Ice Sheet as it is a well-established term within the cryosphere community.
Line 14: Remove “a wide range of environmental conditions, including”. This is, in fact, the complete list of parameters that you consider in this paper.
Revised as suggested.
Line 33: I might add the word to “potentially” to “burying vast stocks”. The literature cited is fairly speculative, and since glacier growth can be highly erosive, the extent of organic matter burial remains uncertain.
Revised as suggested.
Figure 1: I know this is a conceptual figure, but the image of a meltwater channel equal in thickness to the subglacial sediment package is wrong – especially when you later consider sediment packages up to 15 m thick! Also adjust the text to not overlap with the drawing and change the font to match the font used by the journal.
Revised as suggested; added a comment about the dimensions being not to scale in the caption.
Lines 70-72: This is not needed here. Equations, etc., should appear in the main text.
Revised as suggested.
Line 95: I thought the top was water, per the conceptual model.
Revised as suggested and we rearranged the figure.
Line 96: There certainly would not be cryoturbation in the subglacial environment! Cryoturbation is caused by seasonal freezing and thawing. The bed of the Greenland Ice Sheet is variably frozen or warm, but it does not vary seasonally. However, there very well could be sediment shear from the driving stress of the glacier overhead. I think it’s okay ignore sediment shear, but the reference to cryoturbation has me worried.
Agreed, and revised as suggested.
Lines 98-99: You generally should avoid calling out sections in advance.
Revised as suggested
Line 101: Remove the word “environments”.
Revised as suggested
Line 110: Salinity wouldn’t exactly be zero, but maybe it’s close enough to zero that it might as well be?
Clarified to point this out.
Section 3.3: This is the first of several sections composed more of background information than strictly methods. I would try to take the discussion about electron acceptors, methanogenesis pathways, and previous work into the introduction and have this section just be a description of how the model was implemented in the study (i.e. have the subsection start at line 149). I also think this goes into far too much detail justifying why simplifying assumptions were made. Readers will perfectly well understand why you can’t include every possible electron acceptor or methanogenesis pathway into the model.
Agreed. We only keep the model description along with the relevant equations, and replace the justification with a table summarizing the ranges used and values available in the literature that are considered in this study. We refer to specific sections in the appendix where the full justification can be found.
As explained above, we consider the justification of the parameters an important part of this study, as it summarises the current state of the art on subglacial sediments from Greenland, Antarctica, and the Arctic in general.
Line 117: You can just have “(Table C1)”.
Revised as suggested
Line 119: I had to go back and look up what TEAs meant, highlighting my earlier point about acronyms.
Revised as suggested. Some less used abbreviations (RTM, LG, IS, RU, TEA) are now spelled out to improve the reading experience.
Line 124: Sulphur is the one electron acceptor that you don’t explore in any further detail here. Of all the electron acceptors, sulphur is by far the most abundant in Greenland Ice Sheet waters.
Agreed (see our reply to comment to Lines 180-183 further below).
Line 126: “Statham et al.”; I assume The Cryosphere doesn’t expect three author lists.
Revised as according to The Cryosphere guidelines.
Line 137: There’s a missing link here or the logic of the paragraph needs to be reordered. The terms acetoclastic and hydrogenotrophic need context.
We removed the section on acetoclastic and hydrogenotrophic methanogenesis to shorten the parameter description, and deemed it unnecessary.
Line 138: Does The Cryosphere allow for citation of papers in review?
The cited study has been published in the meantime. Changed to Hatton et al. 2026.
Line 165: I wonder to what extent the normal aging process might not apply or might not apply in the same way to the subglacial environment.
The accumulation of organic matter process occurs during interglacial periods, when seasonal vegetation grows and accumulates biomass. The organic material is then degraded by microorganisms under both ice-free and ice-covered, warm-based conditions, driving a continuous decrease in bulk organic matter reactivity with degradation time. Frozen bed conditions likely slow down or halt the degradation process and thus slow down the ageing. The reactive-continuum model accounts for this widely observed decrease in reactivity and we assume here that the organic matter that is supporting metabolic processes in the subglacial sediment has been aging since 1.5–4.4 thousand years ago in agreement with recent observations (Hatton et al. 2026).
Line 166: I don’t think it’s necessary to call out upcoming sections in advance.
Revised as suggested
Lines 168-186: This is also introductory material. The methods section should focus on the methods.
We have moved this section to the appendix together with the rest of the parameter justifications that were removed from the method (section 3).
Line 170: Again, you don’t need a preamble before citing a table.
Revised as suggested
Line 171: Parentheses should only be around the year when the authors names form part of the sentnece.
Revised as suggested
Line 180: You do not define AOM. (Again, too many acronyms.)
Revised as suggested
Lines 180-183: To me, these values for sulphur are not particularly low. What are you comparing this to here? As far as I know, there’s never been any evidence for sulphur reduction under the Greenland Ice Sheet. But I don’t think that’s for lack of sulphur; sulphate is one of the major anions found in subglacial waters in Greenland and in glacial environments in general. There’s a missing conceptual piece here. It might be the mechanochemistry angle. Subglacial environments are generally highly oxidative.
We agree that low porewater sulfate is not the same as sulfate limitation of organoclastic sulfate reduction or anaerobic methane oxidation. We have revised the manuscript to justify the assumed dominance of AeOM over AOM due to the fundamentally oxidative character of the subglacial environment, or at least its hydrologically connected part, beneath the GrIS via the continuous delivery of O₂-rich surface meltwater through moulins and crevasses combined with the oxidative reactivity of freshly ground rock surfaces. Reducing conditions certainly exist beneath the ice, with the in situ CH4 produced as evidence; however, we contend that for a significant role of AOM as a CH4 sink steep and stable redox gradients would need to be established (analogous to the sulphate-methane transition zone in the seafloor), which is unlikely to the dynamic nature of the subglacial environment. We have shifted the focus on clumped CH₄ isotope analysis (Adnew et al., 2025) fingerprinting aerobic oxidation as the dominant CH₄ sink (along with off-gassing) in SW Greenland meltwater.
We have updated Section [3.3.1] to state: ' Furthermore, clumped-isotope analyses of methane exported from southwest Greenland suggest that aerobic oxidation is the dominant methanotrophic pathway (Adnew et al., 2025), which is consistent with the fundamentally oxidative character of the subglacial hydrological system beneath the GrIS driven by the continuous delivery of O₂-rich surface meltwater through moulins and crevasses, combined with the oxidative reactivity of freshly comminuted rock surfaces generated by subglacial grinding (Stone et al., 2023; Gill-Olivas et al., 2024).
Therefore, we focus on aerobic CH₄ oxidation as the dominant biological CH₄ sink, while acknowledging that anaerobic pathways may operate locally where anoxic microenvironments and steep redox gradients develop. In the model, O2 acts as the operative terminal electron acceptor, and our broad parameter sweep of O₂ boundary concentrations is designed to encompass variability in overall oxidative capacity across the range of plausible subglacial conditions. We explicitly note this as a simplifying assumption and a limitation of the model's scope.‘
Line 185: Again, can you site manuscripts in preparation in The Cryosphere?
The cited study was removed.
Line 194: Not necessary.
Revised as suggested
Section 3.6.1: It might be worth pointing out the glacial forefield in the study area also consists of a mix of rocky and sediment mantled regions.
Agreed, and we added a sentence saying this. Section 3.6.1 was shifted to the appendix A.1.1: “However, sparse data on both sediment thickness and spatial distribution hinder an accurate assessment, but it is safe to assume a strong similarity the glacial forefield and the ice-covered area”
Line 241: This is very unlikely. It’s far more plausibly that there are patches where organic matter is highly concentrated due to minimal erosion for whatever happenstance of the subglacial environment alongside vast areas with almost nothing. On the other hand, uniformity is far easier to model (especially in a nonarbitrary way).
We fully agree that uniform OM distribution across the subglacial bed is unlikely and wish to clarify that the chosen ensemble model approach does not reflect a single uniform OM content or reactivity across the subglacial bed. Rather, we run a large model ensemble spanning the full plausible range of OM concentrations, reactivities, and sediment cover fractions, such that each model realisation represents a distinct subglacial scenario. The ensemble results therefore reflect the full envelope of possible CH₄ fluxes across this parameter space, rather than a single spatially averaged prediction.
We have clarified this distinction in the revised manuscript and note that the ensemble upper bound likely reflects localised OM-rich patches rather than a catchment-wide average, while the lower bound captures the near-bare bedrock end-member. The integrated CH₄ flux from a real, heterogeneous catchment would fall somewhere within this envelope, but its exact position cannot be determined without spatially resolved information on subglacial sediment characteristics that are entirely unconstrained by available observations, making such a parameterisation more speculative rather than more realistic.
We agree that uniformity is far easier to model in a nonarbitrary way. We also agree in this assumption being unlikely, however any assumptions on a concentration profile would be even more speculative. We have emphasised this caveat in the model results (see 4.2 CH₄ production in subglacial sediment) and stress that results here are considered high/-er estimates.
Line 247: RCM is undefined.
RCM is defined on line 149, but we take the reviewers point have repeated here is here for ease of reading.
Line 254-272; Figure 2: I needed to re-read this several times to understand the main point. State the main point – i.e. how all of this affects your model – much more clearly and consider omitting some of the detail. The figure could maybe go to the supplement.
Figure moved to the supplementary.
Line 287: Is there a good reason not to have the temperature increase by 0.5°C over the 15 m sediment package? Unlike the other simplifying assumptions, I don’t see how this would make the model any harder to run.
The effect would be negligible considering the model’s parameterisation. Molecular diffusion coefficients are only weakly sensitive to temperature at near-freezing conditions and a 0.5°C increase from -0.5°C to 0°C produces a change in diffusivity well within the uncertainty of other model parameters. The same is true for the temperature effect on biogeochemical rates.
Line 323: The pressure in the subglacial environment is usually considerably below ice overburden pressure, at least during the summer when water is discharging from the proglacial environment. You might want to look into borehole studies to see what is actually more realistic. This fluctuating pressure would also have implications for the stability of clathrates, should they form. I.e. if a clathrate formed in winter, when water pressure was high, it might well melt in summer as pressures drop.
As rightly pointed out, subglacial water pressure does indeed fluctuate substantially in channelized drainage systems during summer, as documented in borehole studies. However, the majority of the subglacial bed that is relevant for CH4production lies away from discrete channels, in the distributed drainage system, where porewater pressure remains close to ice overburden pressure year-round and seasonal pressure fluctuations are minimal. Since soft sediments are most plausibly preserved precisely in these areas away from concentrated channelized flow, ice overburden pressure is the more appropriate reference for the sediments our model describes.
Regardless, the question of clathrate stability is not a concern for our study area. SW GrIS marginal catchments mostly sit outside the gas hydrate stability zone under any realistic combination of local temperature and pressure (Lamarche-Gagnon et al., 2019). Furthermore, even if water pressure were reduced to well below overburden during summer drainage events, dissolved CH₄ concentrations in subglacial porewaters remain well below saturation, such that free gas phase formation and clathrate stability are not reached. Seasonal pressure fluctuations are therefore unlikely to have meaningful implications for CH₄ phase partitioning or transport in our study catchments, and we have added a brief clarification to this effect in the revised manuscript.
Section 3.6.5: This should at least briefing discuss all the work Jon Telling’s group has been putting out on the mechanochemistry of glaciers, which may be critically important to why they’re so highly oxidative (e.g. Stone et al., 2023, Gill-Olivas et al., 2024, etc.).
We agree on the relevance of freshly and reactive energy sources through comminuted rock and have added suggested references to further highlight the oxidative nature of the subglacial environment. However, we’d like to point out that our focus here are soft bed areas (deep subglacial sediments), which are unlikely influenced by rock comminution occurring on hard bed, especially in the deeper layers of the sediment which demonstrably develop reducing conditions (see also our response above).
Line 385: This goes back to the hard bed / soft bed question. In the areas with thick sediment packages, methanogenesis is potentially extensive. But it remains unclear how prevalent those are – i.e. it could be mostly a hard bed with only patches of thick sediment
We fully agree with the reviewer, and this is precisely the uncertainty our ensemble approach is designed to reflect. We do not assume that thick sediment packages are widespread but treat sediment thickness and areal cover as independent variables spanning the full plausible range, including scenarios where soft sediment occupies only a small fraction of the bed (1 % of the hydrologically active area). Most central estimates rest on conservative estimates of sediment cover (i.e., 10-50% of the hydrologically active area). The upper end of our CH₄ flux estimates corresponds to extensive thick sediment cover, which we consider an end-member rather than a central estimate.
On the other hand, next to inverse model results, the observed wide-spread CH₄ export (Hatton et al., 2026) itself confirms that some sediment must be present and that the question is how much and where. Our quantitative framework aims at constraining plausible ranges and shows that the measured concentrations can indeed be reconciled with sparse and patchy sediment cover without invoking large, thick sediment packages. Even scenarios with less than 1–10% sediment cover of the hydrologically active area, combined with modest OM contents and low-to-moderate reactivity, are sufficient to reproduce observations.
Line 394: I keep coming back to this, but to me GSA is Geological Society of America or perhaps General Services Administration.
Some less used abbreviations (RTM, LG, IS, RU, TEA, GSA, OM) will be spelled out to improve the reading experience.
Lines 396-400: I find this section very confusing. You first say that saturation is rarely reached and then say that it is often exceeded. Which is it?
Section 4.5 was heavily edited and hopefully is now more understandable.
Line 399: I would not point the reader back to the methods here. If the reader goes back there, they will also find the suspiciously even number of 50 mM proposed without any citation or justification.
Corrected the indeed suspiciously even number to 59.4 mM and added explanation and reference how this value is calculated using the SUGAR toolbox (Kossel et al., 2013).
Line 400: That is a huge range.
See above
Line 439: Russell has a much smaller catchment than Isunnguata Sermia or Leverett. Perhaps relevant to Figure 6.
Dissolved CH₄ concentrations at Russell Glacier range from 2.7 – 83 µM (Dieser et al., 2014), with a total estimated catchment area of 81 km2 (Hawkings et al., 2021). Under these constrained conditions (namely a small catchment area paired with elevated CH₄ concentrations) no ensemble members reach concentrations above the upper threshold required to infer the soft sediment fraction. An explanation could be that some subglacial meltwater, and therefore CH₄, is diverted from Leverett Glacier to Russell Glacier (water piracy, Lindbäck et al., 2015) where it is measured at the margin and associated with Russell Glacier.
Line 448: Missing “The”.
Revised as suggested
Line 448: The dominant role of organic matter reactivity is also clear just by scrutinising your figures. Are there not any datasets available to put greater constraints on this? I.e. studies of particulate organic matter in the outflow?
Unfortunately, not yet. However, we would like to emphasize that our parameter range for OM reactivity is not arbitrary but informed by a compilation of reactivity parameters derived from hundreds of incubation experiments with thawing (subsea) permafrost and high-latitude terrestrial material under both aerobic and anaerobic conditions (Arndt et al., 2026), as well as a compilation of measured methanogenic rates for subglacial material (Wadham et al., 2012, Stibal et al., 2012). We argue that these are the best available constraints. Direct in-situ observational constraints on subglacial sediment OM reactivity beneath the GrIS remain unavailable, but recent data from SW Greenland glacier forefield lake sediment cores confirm that exported subglacial OM is reactive, lending independent support to the reactivity ranges we adopt. We acknowledge that better direct constraints on this parameter would most effectively reduce uncertainty in modelled CH₄ fluxes, and we highlight targeted subglacial sediment core incubations as a key observational priority in the revised manuscript (Section 5).
Line 454: This again brings us back to the question of characterising sediment thickness under the ice.
See above
Line 490: You need to say how this estimate is derived. Cowton et al., 2012 got to 600 km2 by just assuming the active area started below 1200 m. Did you make a similar assumption here? If so, I find this assumption dubious. Anywhere where meltwater contacts the bed is “active”, cutting the hydrologic catchment off at a certain elevation doesn’t make any sense unless there is no surface melt above that elevation.
We agree that delineating the hydrologically active subglacial area by an elevation cutoff is a simplification, and we agree that in principle any area where surface melt reaches the bed and contacts subglacial sediment could contribute to CH₄ export.
However, for the specific purpose of estimating dissolved CH₄ concentrations in proglacial outflow, the relevant area is not the full extent of subglacial meltwater contact but rather the portion of the bed whose drainage is routed toward the main outlet and whose meltwater residence time is short enough for dissolved CH₄ to reach the margin before being oxidised or off-gassed. In this context, the lower-elevation, higher-flux ablation zone dominates the outflow signal both because it generates the majority of meltwater discharge and because it is most likely to be connected to the efficient channelised drainage system that delivers water and dissolved CH₄ to the glacier outlet on timescales relevant to our analysis. Higher-elevation areas contribute proportionally less discharge and have longer, more tortuous drainage pathways, reducing their contribution to the CH4 outflow concentration signal.
We have clarified how our active area estimate was derived in the revised manuscript, following an approach analogous to Cowton et al. (2012), and explicitly acknowledge that this is a simplification while arguing that it is appropriate given our focus on CH₄ export at the glacier outlet rather than total subglacial CH₄ production.
Line 494: You must be using the entire catchments not the hydrological active ones to get these values. Otherwise Isunnguata Sermia would only be twice Leverett, not an order of magnitude larger.
To explore the full range of possible lateral methane export and concentration we also included the entire catchment. However, we stress in the text that the upper estimated values of are most likely impossible to achieve. We also added two tables (Table S2 and S3) to the supplementary information to show how we got these values.
Line 495: I know the Hatton paper is still under review, but how do they get a flux ranging from 0.002- 28.9 Mg per year. That’s a 4 order of magnitude range. This must be a measurement range not error bars on a single estimate of flux. If there really is a 4-order-of-magnitude measurement range at Isunnguata Sermia, that’s actually something very meaty that your model could explore. When do you get the low values, when do you get the high values? Over what temporal and spatial scale does this vary? If your model is right, it would imply that the glacier is accessing different parts of the bed that either have thicker sediment or more reactive sediment over these timescales. Or perhaps it’s simply a function of water residence time per figure 6.
The presented flux range was from an old upscaling method used in an earlier version of Hatton et al., which has now been revised to a smaller range of 0.3 - 14.6 Mg per year. This was corrected throughout the manuscript and we added a table in the appendix highlighting how we derived these and other used flux values, as they might not be self-evident then reading the original source, i.e. Hatton et al., 2026, Lamarche-Gagnon et al., 2019 and Dieser et al., 2014.
Line 498: This needs to be completely rethought. You can’t make any “suggestion” based on a model output that varies over 6 orders of magnitude depending on model input.
See above
Line 501: I’m having trouble following your math here. How do you get from 0.006-0.13 micromolarity multiplied by 3.1-7.9 cubic km of water flux to 0.002-28.9 Mg per year for Isunnguata Sermia? Whether multiplying the top and bottom end members or doing formal error propagation, you cannot get such a large range in the final value. A mistake seems to have been made somewhere.
There was indeed a mistake in the reported value, which stemmed from an old version of Hatton et al. We have re-calculated the values and corrected them in the manuscript and also added tables in the supplementary information to follow how we calculate these values.
Line 520: I would remove “or evades”. It seems very unlikely that gas could evade from the subglacial environments of Greenland.
Methane off-gasses from a turbulent water body as soon as it finds a headspace. Therefore, any air pocket/space in the melt water channel formed around cavities and more generally due to decreasing water pressure towards the end of the melt season, would lead to loss of methane before the melt water reaches the glacier outlet. This methane would therefore not be captured by dissolved methane measurements, as in Lamarche-Gagnon et al., 2019; Hatton et al., (2025); Dieser et al., 2014; Strock et al., 2024.
Line 540: Or it may be substantially smaller; you produced values over an almost 7 order of magnitude range. I don’t think this model can be used in this way. I think it can be used to interpret measurements, but not to postulate about fluxes that haven’t been measured.
See above
Lines 544-547: I don’t follow this argument. This distance between moulins in the interior and the subglacial outflow can be 10s of km, I don’t see how potential evasion at the outlet could have anything to do with diffusive flux to the meltwater channel in the interior.
See answer to comment on Line 520. Additionally, Christiansen and Jørgensen (2018) show that methane evasion upon exiting the enclosed melt water channel can be substantial.
Line 549: There isn’t an unpressurised zone under any flow condition at Isunnguata Sermia, which is highly overdeepened and terminates in a fountain that sprays the subglacial water several meters into the air. There might be under certain conditions at Leverrett, but I find Chandler’s evidence for it (which is based on the behaviour of a tracer gas) less than convincing.
Isunnguata Sermia and Leverett Galcier are indeed very different systems and their melt water behaviour is also in stark contrast. We specified that continuously pressurised systems with overdeepenings like IS are probably unlikely to experience methane off-gassing/evasion within the subglacial melt water channels.
Line 552-557: This is highly, highly unlikely. The routing of subglacial water is entirely driven by the surface slope of the glacier and the shape of the bed. The notion that the large outlets might flow backward against these gradients is incredibly doubtful and isn’t based on anything observed in subglacial hydrology (e.g. in borehole studies). On the contrary, borehole studies and fundamental physics (e.g. Rothlisberger, 1972) suggest that pressure is lower in the large outlets during high flow, so if there were to be any rerouting of methane, it would be to the large outlets, not away from them. But, then again, there is no evidence that the small outlets ever flow up ice overburden pressure gradients either.
The notion of backward flowing outlets is not implied nor stated in the main text. However, to prevent the readers’ potential misunderstanding of the hydrological dynamics and for it being highly unlikely, we removed this section on this speculative pathway for methane loss.
I might point out the authors have presented a very good reason why the big outlets have less methane – longer water residence times. This seems fully explanatory as to why the Russell Glacier study and the Christiansen study (which was of a small, minor outlet) both present far higher methane concentrations than Leverett and Russell. It’s unclear why the authors seem to want to run away from this conclusion by coming up with highly dubious explanations of where the missing methane might have gone.
Water residence time is indeed a major driver in explaining the difference in methane export, we have now further stressed this point in section 4.5
Dissolved methane readily escapes into the atmosphere especially when in turbulent water conditions. Chandler’s evidence for a gas to escape the subglacial system can be seen as a proxy for methane gas.
Line 580: This is very repetitive with what you wrote on line 448.
Agreed and revised
Lines 592-598: I would bring up the water residence time finding here too. In my view this, as much as anything, explains why methane fluxes are sometimes miniscule (especially in large systems like Isunnguata Sermia).
Agreed and added
Table C2: The values for the relative proportions of carbon, nitrogen, and phosphorus in subglacial organic matter are from a 1934 study of plankton? Surely there must be some more current and relevant values to go by? Say, studies of arctic soil organic matter and/or plant material?
We agree that there are indeed better values for the C:N:P ratios (for example in soil organic matter; Cleveland & Liptzin 2007, Tipping et al. 2016); however, here our focus is the simple reaction network of organic matter degradation and methanogenesis and methane oxidation, and thus the proportions of nitrogen and phosphorus are not really relevant to this model and the interpretation of the results.
References
Christiansen, J. R. and Jørgensen, C. J.: First observation of direct methane emission to the atmosphere from the subglacial domain of the Greenland Ice Sheet, Sci Rep, 8, 16623, https://doi.org/10.1038/s41598-018-35054-7, 2018.
Cleveland, C. C. and Liptzin, D.: C:N:P stoichiometry in soil: is there a “Redfield ratio” for the microbial biomass?, Biogeochemistry, 85, 235–252, https://doi.org/10.1007/s10533-007-9132-0, 2007.
Lindbäck, K., Pettersson, R., Hubbard, A. L., Doyle, S. H., As, D., Mikkelsen, A. B., and Fitzpatrick, A. A.: Subglacial water drainage, storage, and piracy beneath the Greenland ice sheet, Geophys. Res. Lett., 42, 7606–7614, https://doi.org/10.1002/2015GL065393, 2015.
Tipping, E., Somerville, C. J., and Luster, J.: The C:N:P:S stoichiometry of soil organic matter, Biogeochemistry, 130, 117–131, https://doi.org/10.1007/s10533-016-0247-z, 2016.
Citation: https://doi.org/10.5194/egusphere-2025-5506-AC2
-
AC2: 'Revisions in response to comments by Anonymous Referee #1', Philip Pika, 31 Jul 2026
-
RC2: 'Comment on egusphere-2025-5506', Anonymous Referee #2, 14 Apr 2026
Overall, this manuscript addresses an important and timely topic. Methane cycling beneath ice sheets remains poorly constrained because the subglacial environment is extremely difficult to access and observe directly. In this context, the use of a reaction–transport model to explore methane production, oxidation, and export beneath the Greenland Ice Sheet is meaningful and potentially valuable. I therefore think the study is relevant and could make a useful contribution to the field. At the same time, the manuscript in its current form still has several important issues that need to be addressed before it can be considered further.
- abstract: The authors should review how papers are written in The Cryosphere, as the current abstract is divided into three paragraphs, which I do not consider appropriate. In particular, the sentence in lines 28–30 is too long and difficult to read.
- Regarding Figure 1, I do not think the current layout is clear. Either it should be integrated into a single figure, or the two panels should be presented at the same size. As it stands, the right panel appears to be the overview figure and the left panel a zoomed-in detail, but the current arrangement does not clearly convey this information. In addition, several symbols appear in lines 71–72 of the caption that have not been defined earlier in the text, which makes it difficult to understand what is being expressed. Also, the authors do not cite Figure 1 in the main text.
- Line 74 What does “BRNS” mean? What is its full name? The full term should be given when it first appears.
- Line 87, what is 𝜙?
- The authors devote 11 pages to a detailed description of the methods they used. Although this information is important, I do not think this is an appropriate way to present a scientific paper. A broader issue is that the Methods section does not clearly separate model formulation from background justification. In particular, large parts of Section 3.3 read more like an extended discussion of why certain pathways were excluded (e.g. nitrate, iron, sulfate-related processes) rather than a concise description of what was actually implemented. I suggest moving much of this contextual material to the Introduction or to a short model assumptions and scope subsection, so that the Methods focus more directly on the equations, boundary conditions, parameterization, and workflow used in this study. At present, this very long Methods section mixes parameter sources and applications with figure and table explanations, which, in my view, does not make for a logically clear or well-structured Methods section, and it is quite tiring to read. And Line 168, there are two verbs in this sentence. Line 319 there are two range of.
- Please check the caption of Figure 5 carefully. The figure does not contain the left and right panels mentioned in the caption. If the authors provided the comparison details in lines 391-395, why they did not show the exact observed data in Fig 5.
- The manuscript relies on a very strong structural assumption that O2 is effectively the only terminal electron acceptor relevant for constraining methane consumption in the model domain, while other redox pathways are excluded. I think this simplification may be acceptable, but it needs to be framed much more explicitly as a limitation on model scope rather than as a sufficiently justified representation of the full subglacial system. At present, the manuscript sometimes reads as if this simplified network is broadly representative, whereas the authors themselves later acknowledge that iron, nitrate, and sulfate related processes may be important.
- I am also concerned that several key arguments rely heavily on unpublished, in-preparation sources. This is especially important where these sources are used to support major modelling assumptions (e.g. the negligible role of AOM) or to constrain/compare the case-study calculations. The authors should either replace these with published evidence where possible, or more clearly state that these assumptions and comparisons remain provisional.
- The sensitivity analysis is useful, but it mainly explores parametric uncertainty within a fixed model structure. It did not address several structural assumptions that may be equally important, including steady-state behavior, diffusion-dominated transport, negligible advection across boundaries, and the exclusion of deeper geological CH4 inputs. I think the manuscript should make this distinction much clearer, because at present the global sensitivity analysis may give the impression that the main uncertainty has been comprehensively explored, whereas several conceptual uncertainties remain outside the tested ensemble.
- Section 4.5 currently feels under constrained. The case study combined modelled sediment fluxes with assumptions about hydrologically active area, soft-sediment cover, residence time, and in-transit oxidation/off-gassing, many of which are themselves poorly known. As a result, the inferred concentration and flux ranges become extremely broad, and the analysis reads more like scenario matching than a strongly constrained inference of subglacial conditions. I think the authors should either tighten this section substantially or present it more explicitly as an exploratory exercise rather than an inference of likely catchment properties. I also find the interpretation too strong where the authors state that cases in which simulated CH4 falls within the observed range imply little or no oxidation/off-gassing during transport. Matching the observed outlet concentration does not uniquely support that conclusion; it could equally arise from larger subglacial production combined with stronger in-transit removal. This part should be rephrased more cautiously.
- The manuscript does not clearly distinguish between generic model behavior and site-constrained interpretation. Before Section 4.5, the Results mainly describe model outcomes across a broad parameter space rather than a case-specific application. However, these generic simulations are already discussed in close connection with field observations, which blurs the distinction between exploratory model behavior and site-based inference. I suggest that the authors make this transition much clearer, and reserve the interpretation against real catchment observations for the dedicated case-study section.
- Another general issue is the very heavy use of abbreviations throughout the manuscript. While I understand that some abbreviations are unavoidable in a modelling paper, in the current version there are so many that they interrupt the reading flow and make the manuscript unnecessarily difficult to follow. I frequently had to go back and check what a term meant
- Some statements in the Conclusion appear to go beyond what the current model can firmly support. In particular, the recommendation that drilling strategies should prioritize areas with deep sediment seems somewhat premature, given that the model does not include several potentially important processes, such as additional redox pathways, transient hydrology, erosion, and seasonal changes in O2 supply. I suggest either softening these applied implications or linking them more explicitly to the stated limitations of the model.
- The Conclusion section would also benefit from substantial tightening. At present, it reads partly as an extension of the Discussion rather than as a concise synthesis of the main findings. Several points appear repetitive, and the section would be stronger if it focused more clearly on the key model insights, what can and cannot be inferred from the current framework, and the main priorities for future work.
Citation: https://doi.org/10.5194/egusphere-2025-5506-RC2 -
AC1: 'Revisions in response to comments by Anonymous Referee #2', Philip Pika, 31 Jul 2026
Response to Referee 2
Overall, this manuscript addresses an important and timely topic. Methane cycling beneath ice sheets remains poorly constrained because the subglacial environment is extremely difficult to access and observe directly. In this context, the use of a reaction–transport model to explore methane production, oxidation, and export beneath the Greenland Ice Sheet is meaningful and potentially valuable. I therefore think the study is relevant and could make a useful contribution to the field. At the same time, the manuscript in its current form still has several important issues that need to be addressed before it can be considered further.
- abstract: The authors should review how papers are written in The Cryosphere, as the current abstract is divided into three paragraphs, which I do not consider appropriate. In particular, the sentence in lines 28–30 is too long and difficult to read.
Revised as suggested
- Regarding Figure 1, I do not think the current layout is clear. Either it should be integrated into a single figure, or the two panels should be presented at the same size. As it stands, the right panel appears to be the overview figure and the left panel a zoomed-in detail, but the current arrangement does not clearly convey this information. In addition, several symbols appear in lines 71–72 of the caption that have not been defined earlier in the text, which makes it difficult to understand what is being expressed. Also, the authors do not cite Figure 1 in the main text.
Revised as suggested
- Line 74 What does “BRNS” mean? What is its full name? The full term should be given when it first appears.
Revised as suggested
- Line 87, what is 𝜙?
Definition added and revised as suggested
- The authors devote 11 pages to a detailed description of the methods they used. Although this information is important, I do not think this is an appropriate way to present a scientific paper. A broader issue is that the Methods section does not clearly separate model formulation from background justification. In particular, largeparts of Section 3.3 read more like an extended discussion of why certain pathways were excluded (e.g. nitrate, iron, sulfate-related processes) rather than a concise description of what was actually implemented. I suggest moving much of this contextual material to the Introduction or to a short model assumptions and scope subsection, so that the Methods focus more directly on the equations, boundary conditions, parameterization, and workflow used in this study. At present, this very long Methods section mixes parameter sources and applications with figure and table explanations, which, in my view, does not make for a logically clear or well-structured Methods section, and it is quite tiring to read. And Line 168, there are two verbs in this sentence. Line 319 there are two range of.
Here is our response to Reviewer 1, which addresses the reviewer 2 correct concerns:
We agree that Section 3 previously mixed methodological description with elements of model justification, which made it harder to read as a clear methods section. We have therefore extensively streamlined and restructured the section to improve the separation between the conceptual basis for the reaction network and its mathematical implementation. We also have emphasized the use of the model results as a diagnostic tool to derive inverse constrains on inaccessible environmental information based on meltwater outflow concentrations.
Importantly, however, we retain the conceptual justification as a separate sub-section within the method section. It is not background or introductory material in the conventional sense, but an integral part of the model design. In subglacial environments there is no universally agreed set of dominant biogeochemical pathways, and the relative importance of intermediate terminal electron acceptor processes, methane oxidation pathways, and external methane sources remains actively debated and strongly system dependent. For this reason, the selection and exclusion of specific reactions is itself a key modelling decision guided by our current understanding, but aspects of identifiability and predictive skill.
We therefore consider it necessary to explicitly document and justify these choices within the modelling framework. To address this while improving readability, we have consolidated all process justification into a dedicated sub-section (3.3.1 Selection of biogeochemical processes) placed prior to the mathematical formulation. This section clearly outlines the evidence base and modelling rationale for each included and excluded pathway, using a consistent structure, while the remainder of Section 3 now focuses on the reaction network formulation and implementation.
Additionally, we addressed the comments on lines 169 and 319.
- Please check the caption of Figure 5 carefully. The figure does not contain the left and right panels mentioned in the caption. If the authors provided the comparison details in lines 391-395, why they did not show the exact observed data in Fig 5.
The caption has now been revised. Adding the exact observed data is a valid suggestion; however, it would lead to more data points on an already overcrowded panel, adding only limited information, as the focus is the comparison to the observed range. Furthermore, comparing to the exact observed data would give the wrong impression that the model is trying to replicate these observations, which is not the aim.
- The manuscript relies on a very strong structural assumption that O2 is effectively the only terminal electron acceptor relevant for constraining methane consumption in the model domain, while other redox pathways are excluded. I think this simplification may be acceptable, but it needs to be framed much more explicitly as a limitation on model scope rather than as a sufficiently justified representation of the full subglacial system. At present, the manuscript sometimes reads as if this simplified network is broadly representative, whereas the authors themselves later acknowledge that iron, nitrate, and sulfate related processes may be important.
We agree with this point and have revised the relevant section to make this distinction more explicit. Specifically, we now state directly that the exclusion of intermediate TEA pathways (denitrification, iron reduction, sulfate reduction) reflects a deliberate simplification of model scope, adopted because these pathways cannot currently be parameterised as spatially explicit reaction terms at the catchment scale given existing constraints on oxidant availability, reactivity, and transport and not because we consider them absent or negligible in the subglacial system in general.
In addition, we have also revised this section to justify the assumed dominance of AeOM over AOM in the fundamentally oxidative character of the subglacial environment beneath the GrIS.
That said, we would like to highlight that field and experimental evidence consistently shows that aerobic methanotrophy is the primary biological CH₄ sink at glacier margins (e.g. Dieser et al., 2014; Adnew et al., 2025, Lamarche-Gagnon et al., 2019), and that where O₂ is depleted, CH₄ tends to accumulate rather than being consumed by alternative pathways.
While molecular evidence from GrIS-adjacent sediments confirms the presence of ANME-affiliated organisms capable of anaerobic CH₄ oxidation, aerobic methanotrophs emerges as the numerically and functionally dominant CH₄-oxidising community (Ardoin et al. 2026).
- I am also concerned that several key arguments rely heavily on unpublished, in-preparation sources. This is especially important where these sources are used to support major modelling assumptions (e.g. the negligible role of AOM) or to constrain/compare the case-study calculations. The authors should either replace these with published evidence where possible, or more clearly state that these assumptions and comparisons remain
Two of the cited studies have been published in the meantime: Arndt, in review, 2025 is now Arndt et al. 2026 Global Biogeochemical Cycles, while Hatton et al. in review 2025 is now Hatton et al. 2026 Nature Geoscience. The negligible role of AOM, at least at our current understanding, is based on Dieser et al., 2014 and Adnew et al., 2025. Both studies highlight the dominance of aerobic oxidation of methane.
- The sensitivity analysis is useful, but it mainly explores parametric uncertainty within a fixed model structure. It did not address several structural assumptions that may be equally important, including steady-state behaviour, diffusion-dominated transport, negligible advection across boundaries, and the exclusion of deeper geological CH4 inputs. I think the manuscript should make this distinction much clearer, because at present the global sensitivity analysis may give the impression that the main uncertainty has been comprehensively explored, whereas several conceptual uncertainties remain outside the tested ensemble.
We have clarified these points in the revised manuscript:
Diffusive vs advective transport: We note that methane production itself, as quantified by the model, is not mechanistically dependent on the mode of porewater transport (diffusive vs. advective): production is controlled by organic matter availability and degradation kinetics within the anoxic zone, independent of how the resulting methane is subsequently transported. The choice of transport regime instead affects the fate of that methane after production, specifically, the extent to which it is oxidized in the shallow oxic surface layer. We clarified this point in section 3.2 Transport Model
The primary effect of advection would be to allow a fraction of produced methane to bypass the oxic surface layer. Under this scenario, our diffusion-only representation would tend to overestimate the degree of methane oxidation and correspondingly underestimate net methane export. We consider this a conservative bias with respect to our reported export estimates.
Steady State: We acknowledge that subglacial water O₂ concentration may vary across melt-season and seasonal timescales. However, resolving this transient forcing explicitly would require time-resolved boundary condition data (e.g., continuous subglacial O₂ or meltwater flux records) that are not currently available or easily deducible at both local and catchment scale. A steady-state treatment is the standard approach adopted in comparable modelling studies under this same data limitation. This does not mean that our study assumes that subglacial O2 concentrations are constant. It explores how subglacial sediment CH4 cycling responds across the plausible envelope of oxygen concentrations (i.e. from anoxic to fully oxygenated subglacial waters). The applied ensemble approach is useful for bounding the plausible range of subglacial sediment methane fluxes, in the hope that future field work studies will shed light on subglacial oxygen concentrations.
In addition, our sensitivity analysis shows that CH₄ production is only weakly sensitive to O₂ concentration itself, meaning that uncertainty introduced by unresolved temporal variability in O₂ supply is unlikely to propagate strongly into the model's primary output. We therefore consider the steady-state treatment a pragmatic simplification necessitated by current data constraints, rather than a claim that O₂ dynamics are genuinely static in the subglacial system, and we have revised the manuscript to state this explicitly. We clarified this point in section 3.6 Environmental Ensemble for global sensitivity and diagnostic framework.
Exclusion of deeper geological CH₄ inputs: As discussed in the manuscript, stable isotope signatures and microbial indicators consistently indicate that methane exported from southwest Greenland catchments is predominantly biogenic, and methane hydrates are unlikely to constitute a significant or sustained source beneath ice-sheet margins (Wadham et al., 2012; Lamarche-Gagnon et al., 2019; Adnew et al., 2025). We have further highlighted throughout the manuscript that this study focuses on biogenic methane production and export.
- Section 4.5 currently feels under constrained. The case study combined modelled sediment fluxes with assumptions about hydrologically active area, soft-sediment cover, residence time, and in-transit oxidation/off-gassing, many of which are themselves poorly known. As a result, the inferred concentration and flux ranges become extremely broad, and the analysis reads more like scenario matching than a strongly constrained inference of subglacial conditions. I think the authors should either tighten this section substantially or present it more explicitly as an exploratory exercise rather than an inference of likely catchment properties. I also find the interpretation too strong where the authors state that cases in which simulated CH4 falls within the observed range imply little or no oxidation/off-gassing during transport. Matching the observed outlet concentration does not uniquely support that conclusion; it could equally arise from larger subglacial production combined with stronger in-transit removal. This part should be rephrased more cautiously.
This is indeed one of the main intentions of this section, and it does work backwards from observations, identifying which combinations of OM content, reactivity, sediment cover, and oxidation reproduce measured CH₄ concentrations at both glaciers. But we agree that the framing is a bit obscured from how the section is currently structured.
To clarify our intent, we use the model to simulate a large environmental ensemble with two aims: 1) to quantify potential CH₄ production rates and fluxes in subglacial sediments and identify their main environmental controls, and 2) to use the ensemble as a diagnostic tool, combined with observed meltwater CH₄ concentrations, to infer plausible ranges of environmental conditions and assess the possible extent of in-transit CH₄ loss (i.e. in an exploratory way).
Section 4.1-4.4 address the first aim and results are discussed in the context of available observational data. Section 4.5 addresses aim 2), and we agree this was not made explicit enough, contributing to the section reading as scenario matching. As detailed in our response to Reviewer 1, we have restructured this section, as well as the abstract and methods section to state this more clearly upfront and present the analysis explicitly as an inversion that helps identify which environmental conditions are consistent with the available observations at Isunnguata Sermia and Leverett Glacier rather than a uniquely constrained inference. Broad flux range are retained as motivation and seasonal variability is discussed explicitly.The two main conclusions of this diagnostic framework are that 1) we do not need to infer exceptional sediment cover or OM content to explain observed methane fluxes and that 2) across a wide range of plausible subglacial conditions, predicted CH₄ concentrations are consistently higher than those measured at the glacier outlet, pointing to oxidation, off-gassing, or both as near-ubiquitous features of lateral CH₄ transport. Matching the observed outlet concentration does not necessarily mean that there is no removal, it could indeed equally arise from larger subglacial production combined with stronger in-transit removal. We clarified this in the revised version.
We also explain this in the response to Reviewer 1, who raised a closely related concern about this section.
- The manuscript does not clearly distinguish between generic model behaviour and site-constrained interpretation. Before Section 4.5, the Results mainly describe model outcomes across a broad parameter space rather than a case-specific application. However, these generic simulations are already discussed in close connection with field observations, which blurs the distinction between exploratory model behavior and site-based inference. I suggest that the authors make this transition much clearer, and reserve the interpretation against real catchment observations for the dedicated case-study section.
We agree that the distinction between exploratory ensemble behaviour and site-constrained interpretation could be made clearer, and that Sections 4.1–4.4 currently blur this distinction by discussing generic simulation results in close connection with field observations. As detailed in our response above (and to Reviewer 1), the ensemble serves two distinct purposes and we have now revised the abstract, methods and result sections to make this distinction clearer.
We do, however, consider it important to discuss simulated rates and fluxes in the context of the available literature within Sections 4.1–4.4, even though this is not a site-constrained inference in the sense of Section 4.5. Comparing simulated values against the broader range of published subglacial and proglacial CH₄ observations provides a plausibility check on whether the ensemble produces geochemically realistic rates and fluxes and helps identify which parts of the explored parameter space correspond to conditions already known to occur in nature, versus regions of parameter space that, while mathematically permitted by the model, may be less environmentally realistic. We see this as distinct from the site-specific inversion performed in Section 4.5, which uses two individual, well-characterised observational records to constrain plausible local conditions. We have carefully revised sections 4.1-4.4 in this spirit.
- Another general issue is the very heavy use of abbreviations throughout the manuscript. While I understand that some abbreviations are unavoidable in a modelling paper, in the current version there are so many that they interrupt the reading flow and make the manuscript unnecessarily difficult to follow. I frequently had to go back and check what a term meant
Some less common abbreviations (RTM, LG, IS, RU, TEA, GSA, kAeOM, OM,) are now spelled out to improve the reading experience. While others are kept as they are common in biogeochemical modelling studies (O2, CH4, GrIS, SW). AeOM is kept abbreviated, but its meaning is now often spelled out to improve reading flow.
- Some statements in the Conclusion appear to go beyond what the current model can firmly support. In particular, the recommendation that drilling strategies should prioritize areas with deep sediment seems somewhat premature, given that the model does not include several potentially important processes, such as additional redox pathways, transient hydrology, erosion, and seasonal changes in O2 supply. I suggest either softening these applied implications or linking them more explicitly to the stated limitations of the model.
We have toned down our implications and highlighted the model’s limitations to a certain extent but would like to point out that despite the simplicity of the used model the reviewer’s suggested potentially important processes (additional redox pathways, transient hydrology, erosion, and seasonal changes in O2 supply) would not affect the stated recommendation for deep drilling sites. Biogenic methane production requires stable anoxic conditions, which are found in deeper sediments, and even deeper when additional redox pathways and varying O2 supply are considered.
- The Conclusion section would also benefit from substantial tightening. At present, it reads partly as an extension of the Discussion rather than as a concise synthesis of the main findings. Several points appear repetitive, and the section would be stronger if it focused more clearly on the key model insights, what can and cannot be inferred from the current framework, and the main priorities for future work.
We agree on the benefit of substantial tightening the conclusion and revised as suggested.
References
Ardoin, L., Larose, C., Tison, J.-L., Keuschnig, C., Gkinis, V., El Amri, S., Blard, P.-H., Bierman, P., Blunier, T., Dahl-Jensen, D., Maistriau, C., Steffensen, J.-P., Röckmann, T., and Fripiat, F.: Methane and carbon dioxide dynamics beneath the Greenland Ice Sheet: Insights from ice core basal materials, https://doi.org/10.5194/egusphere-2025-6204, 13 January 2026.
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 1,534 | 549 | 146 | 2,229 | 229 | 207 |
- HTML: 1,534
- PDF: 549
- XML: 146
- Total: 2,229
- BibTeX: 229
- EndNote: 207
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Interesting read. I do not recall coming across such type of research. Considering methane's high warming potential, it has been a hot topic, specifically for permafrost degradation in peatlands and in other periglacial systems.
Could your model be applied to cold-based ice-sheets/glaciers, or do you think that it does not make any sense? We have seen from peatlands that there are some fluxes even during the winter season when grounds are frozen (in subarctic Finland and northern Sweden at least), so maybe there could be some fluxes even below cold-based glaciers?
Angelos