the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 28 Apr 2026
Seasonal icings reveal subsurface water drainage of a Greenland Ice Sheet outlet glacier
Abstract. In many glacial settings, winter outflows of proglacial water create stratified domed ice structures on glacial forefields. These structures, called naledi, provide an opportunity to characterise the solute fluxes of hydrological systems in winter which are poorly constrained compared to summer outflows which are dominated by supraglacial meltwater input. To characterise the different hydrological systems feeding naledi and provide conceptual models of their formation, we sampled 12 overwinter naledi and 4 supercooled summer ice accumulations (accreted ice) from the forefield of Isunnguata Sermia, a western outlet glacier of the Greenland Ice Sheet, during four field campaigns. Major ions and stable water isotopes reveal complex geochemical signatures, where the composition of successive naled layers reveal fluctuations in water source and transport throughout winter. In comparison, core analysis of summer accreted ice shows uniform geochemistry throughout, suggesting freeze-on of water from one single source in quick discrete events. These findings are supported by Electrical Resistivity Tomography (ERT) geophysical surveys, which reveal shallow hydrological pathways (1–5 m below the surface) in the glacial foreland (surveyed 400 m from the glacier terminus). Here, meltwater is transported through a saturated sediment zone feeding proglacial upwellings. Our results inform the first conceptual models of summer vs. wintertime subglacial water routing and naled formation in the proglacial zone of the Greenland Ice Sheet. We show that wintertime naledi form from a combination of top-down (unconfined) and bottom-up (confined) freezing processes which incorporates subglacial discharge, overland flow and precipitation, whereas accreted ice forms from upwellings of high velocity, highly pressurised water which freezes upon release. These seasonal changes drive different geochemical and nutrient outputs, which are critical for evaluating glacier meltwater contributions to proglacial groundwater systems and the impact of glacial discharge on downstream ecosystems.
- Preprint
(3007 KB) - Metadata XML
-
Supplement
(1194 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-2131', Marek Stibal, 01 Jun 2026
-
AC1: 'Reply on RC1', Rebecca McCerery, 07 Sep 2026
We’d like to thank Marek Stibal for taking the time to read and review our submission, and for their kind words, supporting our manuscript. We have provided comments in response to Marek’s suggestions below and think that our manuscript will be strengthened as a result of our discussions here.
- 208-220 Please give your limits of detection and precisions and accuracies for your chemical analyses.
We will update the methods section to include a range of LODs and precision percentages. We can also add further information to our supplementary data, for example, by providing all of the LODs and RSDs for each analyte for every date of analysis (as our samples were run over multiple days).
- Is there a way to quantify the uncertainty of your charge balance estimations of HCO3-?
The uncertainty of HCO3- is inherently tied to the propagated uncertainties of the measured cation and anion concentrations. We would be happy to add a statement noting that HCO3- values carry an uncertainty derived from the combined analytical uncertainties of all measured ions.
- Do you know the concentration of phosphate?
Phosphate is only detectable in 22 individual samples using our instruments and these values can be found in the final database. Given the low-nutrient glacial environment sampled, phosphate concentrations were expected to be at or below detection limits.
- Did you measure TSS (particulates) at all?
TSS was not measured in this study, in many naled ice samples and all of the naled waters we did not find any suspended sediments. Further, our sampling protocol focused on dissolved constituents following filtration through 0.45 µm membranes, meaning particulate material was not retained for analysis. We acknowledge this as a limitation of the current dataset.
- 263-271 Any idea what this yellow water was?
We assume this water has originated from a large upwelling eruption nearby, as its colour is distinctly similar to the sub-naled waters we sampled. This is an important observation and we would be happy to discuss this further in the discussion section.
- Figures 7-9 Would it be possible to integrate the two bars (total ions and ion proportions) into one to make the differences between samples/layers clearer?
We considered integrating the two panels into a single stacked bar chart prior to submission, however, the wide range of TDS values across samples (~1 to 1,300 mg L⁻¹) would mean the bar heights on the low TDS samples would be so small relative to high-TDS samples that the ion proportion fills would become too difficult to see, so we decided against this.
- Figure 13 Is the subsurface water flow constrained by permafrost? I.e., is the depth of it determined by the permafrost table and/or distribution?
It is difficult to say with certainty either way. We’ve included the permafrost layer in our figure because it’s an important feature of this system, but whether or not it contributes to the depth of the subsurface water flow is unknown.
- 555-557 If the accreted ice features were formed quickly during outburst events, could the samples be used as representatives of those outbursts? I would then expect them to contain more sediment than ‘normal’ accreted ice that forms during the melt season around the upwelling. Also, as per Livingstone et al 2026 JGlac I understand the October 2023 outburst was linked to lake drainage (primarily of an ice-marginal lake but potentially also secondarily of subglacial lakes)? Could the sediment in these features then be expected to come from these lakes?
Firstly, we agree that the accreted ice structures, given their inferred rapid formation during high-discharge outburst events, could serve as geochemical snapshots of those outbursts. We will add more information to the manuscript in light of these discussions.
Secondly, regarding sediment content, we note that the accreted ice cores are indeed heavily sediment-laden relative to the naled ice, however, we did not see any differences in sediment content between the accreted cores. Lake-sourced sediments could plausibly contribute to the sediment load in accreted ice formed during such events as well. This would make for some interesting further investigations.
- 607 the correct year for this reference is 2026
We will correct this.
Citation: https://doi.org/10.5194/egusphere-2026-2131-AC1
-
AC1: 'Reply on RC1', Rebecca McCerery, 07 Sep 2026
-
RC2: 'Comment on egusphere-2026-2131', Anonymous Referee #2, 19 Aug 2026
There is much to like in this paper. First and foremost, there is not a sufficient literature on the types of physical, chemical and, perhaps, biological information that becomes stored in aufeis or naledi, and how this compares with that stored in icings. Second, there is precious little of this information on Greenlandic naledi/icings. Third, the interpretation of these data is far from straightforward, and it is great to see photogrammetry and water isotopes being used to help identify the processes controlling the accumulation of the ice and the chemicals stored in them. Overall, I am very supportive of the paper, but you know that the gripes always follow statements of reviewer’s support.
- Meltwater solute fluxes……. You simply cannot say you are contributing to studies of meltwater fluxes without quantifying the amount of solute stored in these types of ice relative to that exported in the meltwater over the rest of the melt season. A FLUX has fundamental units of mass per unit time, and often in meltwater solute studies, it becomes equivalent to concentration x discharge, summed over a period that is often a year. This work presents concentrations, period, and does not contribute to solute flux studies unless more quantitative analysis is included.
- Nutrient fluxes…….. The same argument, but much worse. Only nitrate is measured, and although the figures purport to show nitrate concentrations, these are super difficult to make out. I’m comfortable that you say that this work could contribute to such studies, but not given the analysis you have provided. We’ve measured some big concentrations is not enough. Big concentrations in a small puddle of ice, relative to the small fjord that discharges from the ice sheet each year, does not necessarily sum up to a major contribution of solute and nutrient overall. My hunch is that if you can find a contribution of above a percent you will have done very well.
- Overwinter fluxes of solutes and nutrients…… but you suggest that the solute is all retained in your frozen water. How does this get to the oceans?
- Flash freezing in icings shouldn’t give you much of a chemical gradient in the ice overall (or does it?), in terms of vertical stratification in the ice and along the freezing water sheet. In principle, the di-valent carbonates will precipitate out first, then CaSO4 and then most soluble sulphates that contain monovalent ions and the nitrates and chlorides. Have to check the spatial variation of the icing chemistry to check for this patterning that is found in Svalbard? In principle, the waters should become more concentrated and have a very different chemistry as they flow away from their source, so complicating your interpretation of subglacial/supraglacial chemistries.
- This pattern too should be found in your aufeis. You claim top down and bottom up freezing and it is unclear how much lateral water is occurring when this occurs. In principle, the first formed ice should be dilute (and so may not be supraglacial water), think of the purification of water by freezing, and you should get successive chemical layers building up that might be thick enough to measure these differences if there was an initial water column containing sufficient solute and sufficiently thick. Lateral flow of water during freezing can also give rise to an initial carbonate rich chemistry that evolves into a more sulphate rich chemistry. I think you need to acknowledge these types of caveats when you are claiming that the ice chemistry translates simply to subglacial water chemistry. It is very unlikely to unless there was a static water body that completely froze and through which you have sampled all the ice, from the initial dilute ice through to the last concentrated ice, without bias.
I remain a big fan of the paper. Someone has to go out there and get the data, come up with the ideas and form a case for doing more work. I think you’ve done a good job overall.
Citation: https://doi.org/10.5194/egusphere-2026-2131-RC2 -
AC2: 'Reply on RC2', Rebecca McCerery, 07 Sep 2026
We thank our anonymous reviewer for taking the time to read and review our work, and for their support of our manuscript. We have provided comments in response to their suggestions below.
- Meltwater solute fluxes……. You simply cannot say you are contributing to studies of meltwater fluxes without quantifying the amount of solute stored in these types of ice relative to that exported in the meltwater over the rest of the melt season. A FLUX has fundamental units of mass per unit time, and often in meltwater solute studies, it becomes equivalent to concentration x discharge, summed over a period that is often a year. This work presents concentrations, period, and does not contribute to solute flux studies unless more quantitative analysis is included.
- Nutrient fluxes…….. The same argument, but much worse. Only nitrate is measured, and although the figures purport to show nitrate concentrations, these are super difficult to make out. I’m comfortable that you say that this work could contribute to such studies, but not given the analysis you have provided. We’ve measured some big concentrations is not enough. Big concentrations in a small puddle of ice, relative to the small fjord that discharges from the ice sheet each year, does not necessarily sum up to a major contribution of solute and nutrient overall. My hunch is that if you can find a contribution of above a percent you will have done very well.
We thank our anonymous reviewer for highlighting the two points above regarding fluxes, and we will reword the manuscript as necessary to ensure we only refer to concentrations and solutes, rather than flux(es). In our work, we detected orders of magnitude differences in solute concentrations between summer and winter icings. This is not currently considered when making geochemical inferences on weathering cycles and their downstream impacts on ecosystems. In light of your comments and these discussions, we plan to make this more clear in our manuscript (in the discussion and conclusion sections).
- Overwinter fluxes of solutes and nutrients…… but you suggest that the solute is all retained in your frozen water. How does this get to the oceans?
The reviewer is correct, we suggest that overwinter, all of this solute-rich water is stored in the naled icings and/or moving very slowly through the subsurface hydrological system. However, at the onset of the melt season this stored water will be mobilised, resulting in a pulse of solute-rich waters before the dilute summer melt enters the system. We can clarify this in the discussion section of the manuscript.
- Flash freezing in icings shouldn’t give you much of a chemical gradient in the ice overall (or does it?), in terms of vertical stratification in the ice and along the freezing water sheet. In principle, the di-valent carbonates will precipitate out first, then CaSO4 and then most soluble sulphates that contain monovalent ions and the nitrates and chlorides. Have to check the spatial variation of the icing chemistry to check for this patterning that is found in Svalbard? In principle, the waters should become more concentrated and have a very different chemistry as they flow away from their source, so complicating your interpretation of subglacial/supraglacial chemistries.
As you suggest, the flash freezing of summer icings shows no chemical gradient at all - we found consistent solute concentrations and isotopic signatures throughout all the cores we collected and analysed. In our naled samples we don’t hypothesise that flash freezing occured, but we do note the presence of ‘water packages’ where there are some chemical and isotopic shifts within the cores - however, we base this on our isotopic analysis rather than different geochemical signatures.
- This pattern too should be found in your aufeis. You claim top down and bottom up freezing and it is unclear how much lateral water is occurring when this occurs. In principle, the first formed ice should be dilute (and so may not be supraglacial water), think of the purification of water by freezing, and you should get successive chemical layers building up that might be thick enough to measure these differences if there was an initial water column containing sufficient solute and sufficiently thick. Lateral flow of water during freezing can also give rise to an initial carbonate rich chemistry that evolves into a more sulphate rich chemistry. I think you need to acknowledge these types of caveats when you are claiming that the ice chemistry translates simply to subglacial water chemistry. It is very unlikely to unless there was a static water body that completely froze and through which you have sampled all the ice, from the initial dilute ice through to the last concentrated ice, without bias.
In our models we present the incorporation of ‘other’ sources of water to ice structures, whether that be from supraglacial, englacial or other overland flow sources. We also saw that in a couple of our naled samples, the top layers did not fit the typical freezing package model. We would be happy to highlight this more in the discussion section.
- I remain a big fan of the paper. Someone has to go out there and get the data, come up with the ideas and form a case for doing more work. I think you’ve done a good job overall.
Thank you for your kind words and encouragement. We have found your comments helpful, and feel that the discussions we’ve had here will allow us to refine our manuscript, and make it more accessible.
Citation: https://doi.org/10.5194/egusphere-2026-2131-AC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 327 | 139 | 37 | 503 | 75 | 41 | 49 |
- HTML: 327
- PDF: 139
- XML: 37
- Total: 503
- Supplement: 75
- BibTeX: 41
- EndNote: 49
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
(Disclaimer: I am a biogeochemist and not an expert on glacial hydrology and geophysical surveys.)
In this paper McCerery et many al sampled proglacial icings and accreted ice in the forefield of Isunnguata Sermia in SW Greenland to get an insight into their formations and the dynamics of the glacier’s hydrological system. Using geophysical surveys they show water flow saturated sediment in the subsurface, feeding proglacial upwellings. Geochemical analysis revealed differences between the icings and accreted ice, suggesting different origins. Based on their results the authors contend that the icings are formed from winter discharge through both top-down and bottom-up processes , while the accreted ice features are the result of summertime upwelling of pressurised water.
Given the importance of this site as a representative large catchment of the W GrIS these data and their interpretation are extremely interesting and useful for understanding the hydrological system of the ice sheet and can (and will) inform future research in the area and on the GrIS in general.
Comments and questions:
208-220 Please give your limits of detection and precisions and accuracies for your chemical analyses. Is there a way to quantify the uncertainty of your charge balance estimations of HCO3-? Do you know the concentration of phosphate? Did you measure TSS (particulates) at all?
263-271 Any idea what this yellow water was?
Figures 7-9 Would it be possible to integrate the two bars (total ions and ion proportions) into one to make the differences between samples/layers clearer?
Figure 13 Is the subsurface water flow constrained by permafrost? I.e., is the depth of it determined by the permafrost table and/or distribution?
555-557 If the accreted ice features were formed quickly during outburst events, could the samples be used as representatives of those outbursts? I would then expect them to contain more sediment than ‘normal’ accreted ice that forms during the melt season around the upwelling. Also, as per Livingstone et al 2026 JGlac I understand the October 2023 outburst was linked to lake drainage (primarily of an ice-marginal lake but potentially also secondarily of subglacial lakes)? Could the sediment in these features then be expected to come from these lakes?
607 the correct year for this reference is 2026