Status: this preprint is open for discussion and under review for Climate of the Past (CP).
Nonlinear increase of Greenland Ice Sheet runoff into Disko Bay
Steffen Hetzinger,Jochen Halfar,Takaaki K. Watanabe,and Alexandra Tsay
Abstract. The Greenland Ice Sheet (GrIS) has experienced accelerated mass loss with record peaks in 2012 and 2019. Despite its role as an important tipping element in the climate system, the GrIS's response to recent warming is poorly understood. Here we use Ba/Ca ratios in coralline algae as a proxy for runoff into Disko Bay which is strongly influenced by the input of meltwater from glaciers connected to the GrIS, particularly from Jakobshavn Glacier – the fastest flowing marine-terminating glacier of the GrIS. The 115-year multispecimen master chronology confirms an unprecedented trend change in runoff beginning in the early 2000s. Statistical trend- and time of emergence analysis of the algal proxy record suggests that in 2007 western GrIS runoff has permanently emerged above the 20th century reference period, while temperature observations have not yet exceeded this threshold. This provides independent evidence for a non-linear accelerated response of the largest GrIS glacier, underscoring modelling results that a tipping point in glacial mass balance might soon be reached. Massive GrIS meltwater influx could intensify upper ocean stratification and contribute to global sea level rise.
Received: 05 Dec 2025 – Discussion started: 05 Feb 2026
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.
I think this is an interesting manuscript built around a nice new dataset. The 115-year Ba/Ca record in Disko Bay presented here is a good contribution, and I think this paper should be published on that merit. I get the impression the record was collected carefully and I very much appreciate the extra effort to place it alongside climate and ice-mass products. I gotta say upfront that my background is on the ice-sheet/climate modelling side, not on proxy development or biological climate records. So, I am not even going to try and out-specialist the authors on the proxy part itself, but I think a non-specialist reader like me should still be able to clearly see why the title, abstract, and conclusions are justified, and at the moment I am not there yet. I had a look at some of the references to try to understand the methods better, and most of the points below come from that experience. So here’s the disclaimer (and request): if any of my points/questions/etc. have obvious answers within the field, I would simply ask that those answers are visible in the text for the non-specialist reader. As said, I think the record is worth publishing, but the manuscript currently pushes the interpretation quite a bit harder than the text can support. I’ll go for “major revisions”, but not because the study is weak; I just think the framing is trying to do too much.
General comments
On statistics
I am not 100% there regarding the statistics. The text seems to give two different answers to the same question. For exmaple, the caption in figure 2 says the filtered Ba/Ca–MAR correlation is significant at the 99% level (the number that catches the eye), but at the very end the table A1 reports the same filtered comparison with autocorrelation accounted for and a p value of 0.08. The other filtered comparisons in Table A1 also seem to miss significance thresholds once effective sample size is reduced. I may be off regarding which test the authors want the reader to treat as the relevant one, but as it is written I struggle to reconcile caption and table. If table A1 uses the appropriate treatment for 15-point-smoothed series then is the 99% significance quoted in fig. 2 really the number the main text should be leaning on, or? If it truly is, then I’m a bit confused on the role table A1 is supposed to play here. So, if I’m not that off, please pick the appropriate test and apply it consistently, and then make the abstract, short summary, figure captions, and conclusions in line with that pick. My worry is not really that the relationship evaporates, but that the manuscript might be sounding much more statistically grounded than it actually is.
On nonlinearity
“Nonlinear” makes the paper sound quite strong, but I do not think it’s fully delivering here. From what I can see there is a change point in the smoothed series followed by a steeper increase, plus a plot with the highest years clustered at the end. Interesting, for sure, but none of that truly establishes a nonlinear response in the strong sense implied by the title. I imagine a linear trend with persistence / autocorrelation could produce a pattern like this as well, or? How can I truly distinguish between nonlinear and monotonic increase? I think the manuscript’s got to choose here: either test the claim directly and show why nonlinear is the right description, or tone down to something like “accelerated” or “stepwise” or similar. Same with the way the paper positions its emergence result. Defining emergence relative to 1932-2000 is fine if that is what the study supports, but it is different from emergence above a pre-industrial baseline. Text should be much clearer about that, especially when refeerncing that earlier Greenland work to avoid sounding more general than it really is. By the way, both the emergence date and the change-point date come from a 15-year Gaussian-smoothed series, so it’d be nice to see a bit more caution in how precisely specific dates are presented in-text.
On the proxy
The proxy dicsussion moves (in my read) from “Ba/Ca goes up” to “runoff goes up”, then to causality too fast. No issues with the saying that Ba/Ca contains useful runoff information, of course but the manuscript sometimes treats Ba/Ca as a clean one-variable runoff-o-meter, and I am not so sure about that. I’d be OK if the discussion read as “this is an integrated environmental signal, that seems sensitive to runoff, and is consistent with glacier/ocean change patterns”. The current framing promises a clean proxy (and a clean causal interpretation) that is, to my taste, not actually demonstrated. I also think that the manuscript needs to be more careful with external comparisons. MAR SMB, GRACE total mass change, historical/photograph mass-loss estimates, solid ice discharge are surely related but distinct components of the glacier mass-balance system. Here, SMB is the most direct comparison target for a runoff proxy; the others are related, not equivalent quantities. The text flies over those differences and gives me the impression that all of them are agreeing validations of the same runoff interpretation. I do not think that is precise, so please be much more explicit about what those external products are truly supporting.
On light/sea-ice conditions
Related to the the Ba/Ca => runoff link. Here I had to read around, so take this as a question rather than a direct challenge. Checking the references (some of it by authors overlapping with this one) I get that summer light availability seems to have a significant control on the data. From what I saw, Disko Bay has lost a large fraction of its seasonal sea-ice cover across more or less the same emerging interval. So, my question: can changing light/sea-ice conditions at the site modulate either Ba/Ca, the age-model used to build the record, etc., in a significant way post-2000 rise? I don’t see the manuscript addressing this, and as a non-specialist reader I can#t tell from the text whether this possibility is negligible or not (or already ruled out in the specialist literature?). If yes, then I would just ask that the relevant references are part of the discussion here. Else, some short site-specific discussion would help. More generally, I would really like the chronology structure and robustness to be a bit more on the transparent side in the main text. The post-1932 part of the record is doing different work from the earlier single-specimen extension, so I do not think the current “115-year multi-specimen” framing says that transparently enough. A small metadata table and a simple robustness check showing that the post-2000 rise is shared across specimens would be a nice addition, I think.
On AMOC
Sticky point right here. The AMOC paragraph felt like it belonged to a different paper, for sure. Greenland freshwater flux clearly matters to the North Atlantic, and I understand the motivation to add this type of lines, but this manuscript is not presenting so far solid evidence on the AMOC topic. Sounds more like motivation than anything, and not something that follows from the data shown here. I would keep it much, much shorter and toned down.
Summary
So, in short I think the data set itself is a reason for the paper being published. My friction is that the manuscript is trying to make the strongest possible claim at every chance: title, abstract, conclusions, broader implications, etc., when the material in its current form supports a valuable new proxy record consistent with recent Greenland runoff and (broad) glacier change, rather than a cleanly demonstrated nonlinear response with a well-resolved causal interpretation. To be fully clear here: my request is not more analysis for its own sake. I simply expect the strong claims to either be properly supported, or bring the wording back to what the paper is able to support. It would make for a better paper overall, and an easier one to defend (for you) and to read (for us).
—
NB: I expect most of my specific comments covered if the tone and precision is fine-tuned following my comments above, including adding references and a few typos here and there. No point in pointing all that out before a revised version that should catch most of them.
The Greenland Ice Sheet is a climate tipping element; most glaciers have retreated and Arctic warming has accelerated, yet Greenland Ice Sheet response and mass‑loss estimates remain uncertain and likely underestimated. Using Ba/Ca ratios from a >100‑year coralline algal record, we reconstruct seasonal runoff variability into Disko Bay surface waters and find a nonlinear rise in western Greenland Ice Sheet runoff since the early 2000s, now permanently exceeding variability of the 20th century.
The Greenland Ice Sheet is a climate tipping element; most glaciers have retreated and Arctic...
Overall impression
I think this is an interesting manuscript built around a nice new dataset. The 115-year Ba/Ca record in Disko Bay presented here is a good contribution, and I think this paper should be published on that merit. I get the impression the record was collected carefully and I very much appreciate the extra effort to place it alongside climate and ice-mass products. I gotta say upfront that my background is on the ice-sheet/climate modelling side, not on proxy development or biological climate records. So, I am not even going to try and out-specialist the authors on the proxy part itself, but I think a non-specialist reader like me should still be able to clearly see why the title, abstract, and conclusions are justified, and at the moment I am not there yet. I had a look at some of the references to try to understand the methods better, and most of the points below come from that experience. So here’s the disclaimer (and request): if any of my points/questions/etc. have obvious answers within the field, I would simply ask that those answers are visible in the text for the non-specialist reader. As said, I think the record is worth publishing, but the manuscript currently pushes the interpretation quite a bit harder than the text can support. I’ll go for “major revisions”, but not because the study is weak; I just think the framing is trying to do too much.
General comments
On statistics
I am not 100% there regarding the statistics. The text seems to give two different answers to the same question. For exmaple, the caption in figure 2 says the filtered Ba/Ca–MAR correlation is significant at the 99% level (the number that catches the eye), but at the very end the table A1 reports the same filtered comparison with autocorrelation accounted for and a p value of 0.08. The other filtered comparisons in Table A1 also seem to miss significance thresholds once effective sample size is reduced. I may be off regarding which test the authors want the reader to treat as the relevant one, but as it is written I struggle to reconcile caption and table. If table A1 uses the appropriate treatment for 15-point-smoothed series then is the 99% significance quoted in fig. 2 really the number the main text should be leaning on, or? If it truly is, then I’m a bit confused on the role table A1 is supposed to play here. So, if I’m not that off, please pick the appropriate test and apply it consistently, and then make the abstract, short summary, figure captions, and conclusions in line with that pick. My worry is not really that the relationship evaporates, but that the manuscript might be sounding much more statistically grounded than it actually is.
On nonlinearity
“Nonlinear” makes the paper sound quite strong, but I do not think it’s fully delivering here. From what I can see there is a change point in the smoothed series followed by a steeper increase, plus a plot with the highest years clustered at the end. Interesting, for sure, but none of that truly establishes a nonlinear response in the strong sense implied by the title. I imagine a linear trend with persistence / autocorrelation could produce a pattern like this as well, or? How can I truly distinguish between nonlinear and monotonic increase? I think the manuscript’s got to choose here: either test the claim directly and show why nonlinear is the right description, or tone down to something like “accelerated” or “stepwise” or similar. Same with the way the paper positions its emergence result. Defining emergence relative to 1932-2000 is fine if that is what the study supports, but it is different from emergence above a pre-industrial baseline. Text should be much clearer about that, especially when refeerncing that earlier Greenland work to avoid sounding more general than it really is. By the way, both the emergence date and the change-point date come from a 15-year Gaussian-smoothed series, so it’d be nice to see a bit more caution in how precisely specific dates are presented in-text.
On the proxy
The proxy dicsussion moves (in my read) from “Ba/Ca goes up” to “runoff goes up”, then to causality too fast. No issues with the saying that Ba/Ca contains useful runoff information, of course but the manuscript sometimes treats Ba/Ca as a clean one-variable runoff-o-meter, and I am not so sure about that. I’d be OK if the discussion read as “this is an integrated environmental signal, that seems sensitive to runoff, and is consistent with glacier/ocean change patterns”. The current framing promises a clean proxy (and a clean causal interpretation) that is, to my taste, not actually demonstrated. I also think that the manuscript needs to be more careful with external comparisons. MAR SMB, GRACE total mass change, historical/photograph mass-loss estimates, solid ice discharge are surely related but distinct components of the glacier mass-balance system. Here, SMB is the most direct comparison target for a runoff proxy; the others are related, not equivalent quantities. The text flies over those differences and gives me the impression that all of them are agreeing validations of the same runoff interpretation. I do not think that is precise, so please be much more explicit about what those external products are truly supporting.
On light/sea-ice conditions
Related to the the Ba/Ca => runoff link. Here I had to read around, so take this as a question rather than a direct challenge. Checking the references (some of it by authors overlapping with this one) I get that summer light availability seems to have a significant control on the data. From what I saw, Disko Bay has lost a large fraction of its seasonal sea-ice cover across more or less the same emerging interval. So, my question: can changing light/sea-ice conditions at the site modulate either Ba/Ca, the age-model used to build the record, etc., in a significant way post-2000 rise? I don’t see the manuscript addressing this, and as a non-specialist reader I can#t tell from the text whether this possibility is negligible or not (or already ruled out in the specialist literature?). If yes, then I would just ask that the relevant references are part of the discussion here. Else, some short site-specific discussion would help. More generally, I would really like the chronology structure and robustness to be a bit more on the transparent side in the main text. The post-1932 part of the record is doing different work from the earlier single-specimen extension, so I do not think the current “115-year multi-specimen” framing says that transparently enough. A small metadata table and a simple robustness check showing that the post-2000 rise is shared across specimens would be a nice addition, I think.
On AMOC
Sticky point right here. The AMOC paragraph felt like it belonged to a different paper, for sure. Greenland freshwater flux clearly matters to the North Atlantic, and I understand the motivation to add this type of lines, but this manuscript is not presenting so far solid evidence on the AMOC topic. Sounds more like motivation than anything, and not something that follows from the data shown here. I would keep it much, much shorter and toned down.
Summary
So, in short I think the data set itself is a reason for the paper being published. My friction is that the manuscript is trying to make the strongest possible claim at every chance: title, abstract, conclusions, broader implications, etc., when the material in its current form supports a valuable new proxy record consistent with recent Greenland runoff and (broad) glacier change, rather than a cleanly demonstrated nonlinear response with a well-resolved causal interpretation. To be fully clear here: my request is not more analysis for its own sake. I simply expect the strong claims to either be properly supported, or bring the wording back to what the paper is able to support. It would make for a better paper overall, and an easier one to defend (for you) and to read (for us).
—
NB: I expect most of my specific comments covered if the tone and precision is fine-tuned following my comments above, including adding references and a few typos here and there. No point in pointing all that out before a revised version that should catch most of them.