the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Freshwater input from glacier melt outside Greenland alters modeled northern high-latitude ocean circulation
Abstract. As anthropogenic climate change depletes Earth's ice reservoirs, large amounts of freshwater are released into the ocean. Since the ocean has a major influence on Earth's climate, understanding how the ocean changes in response to an increased freshwater input is crucial for understanding ongoing shifts in the climate system. Moreover, to comprehend the evolution of ice-ocean interactions, it is important to investigate if and how changes in the ocean might affect marine-terminating glaciers' stability. Though most attention in this context has been on freshwater input from Greenland, the other northern hemisphere glacierized regions are losing ice mass at a combined rate roughly half that of Greenland, and should not be neglected. In order to get a first estimate of how glacier mass loss around the Arctic affects the ocean and how potential changes in the ocean circulation might affect marine-terminating glaciers, we conduct one-way coupled experiments with an ocean general circulation model (NEMO-ANHA4) and a glacier evolution model (Open Global Glacier Model; OGGM) for the years 2010 to 2019. We find an increase in heat content of Baffin Bay and changes in the subpolar gyre's structure. Additionally, we find a decreased heat transport into the Barents Sea due to increased freshwater input from Svalbard and the Russian Arctic. The rerouting of Atlantic water from the Barents Sea Opening through Fram Strait leads to an increased heat transport into the Arctic Ocean and a decrease of sea ice thickness in the Fram Strait area.
- Preprint
(215647 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-1425', Anonymous Referee #1, 15 Jul 2024
Malles et al. study how northern hemisphere glaciers and ice caps outside Greenland impact on the regional (and large-scale) ocean circulation by "one-way" coupling a 1/4° ocean model and a glacier model. They find some changes in ocean heat content, transports and sea ice thickness, as well as to the subpolar gyre circulation, but no impact on the AMOC.
I believe that ocean—glacier coupling is an interesting topic that is very suitable for ESD, and the ocean-related results are interesting and novel, as far as I can tell. (Note that I do not have expertise to review the glaciological aspects of the manuscript.) However, the signal is quite small and the simulation period is only 10 years. This leads to some questions about statistical significance, which should be addressed before publication. Overall, the manuscript would benefit from a more concise presentation, especially regarding the writing of the results section and the figures, to bring out the many interesting results more clearly.
General comments
1. Statistical significance
My concern regarding statistical significance is that – due to the short simulation time – some differences could be deemed significant that only arise due to chaotic/intrinsic ocean variability and may not be a physical effect of the meltwater. For example, the significant red and blue patches in the southeastern part of Fig. 7 may simply be artifacts of sampling different chaotic eddy activity. I would be interested in the authors’ opinion on this and recommend adding a brief discussion to the manuscript. The most elegant solution would probably be to run an ensemble simulation, but I understand that this might not be feasible. However, some results could be at least qualitatively compared to published ensemble simulations (e.g., Carret et al. (2021) for SSH – even though regrettably their data does not seem to be publicly available).
Regardless of this issue, I cannot find a description of significance testing in the methods (the test is only mentioned in some figure captions), where the procedure should be added. Are the individual “samples” used for the test monthly or annual means? Do you assume paired or unpaired observations?
Finally, the numbers quoted in the text should be consistently equipped with p-values, and it should be clarified when you are only using the last 5 years or when you are using the full period (which is the default?).
2. Description of the experimental setup
According to L143, “liquid freshwater and iceberg input around Greenland is derived from Bamber et al. (2018)”. Since the Bamber et al. (2018) dataset also provides runoff for the same set of glaciers and ice caps as modeled by the OGGM, this sentence confused me a lot, but the authors have since clarified that they indeed only use Bamber et al. (2018) for Greenland and set the glacier runoff to zero outside of Greenland in the noOGGM run. This should be clarified in the manuscript. There were some other setup-related aspects that should be clarified, see the specific comments below.
3. Presentation of results
The introduction, methods and conclusions are generally well-written and easy to follow, but I found the text difficult to follow in the results section and some of the discussion. Especially when discussing physical mechanisms, some of the results could be re-written with a focus on (assumed) causality, reminding the reader from time to time what is the link to the added freshwater, e.g. in L257-266, L274-277, L315-333 and elsewhere. In addition, please check with a critical eye whether the results section could be thinned out a little bit – there are many interesting results in the paper, but I feel that the most interesting parts sometimes get a little bit lost amid the overall volume.
The figures could also be reduced. For example, it would be possible to remove (or at least move to the appendix/supplement) the time series figures (Figs 3, 4, 10) and simply give the difference and p-value in brackets in the text (“ocean heat transport increases by 0.1 TW (p = 0.02)”) without losing much information. Some maps that show the same quantity in slightly different regions could also be merged – for example, no less than five (!) figures show the potential temperature difference over 0-200m.
Specific comments:
L34 and following: Some more recent references such as Swingedouw et al. (2022), Martin and Biastoch (2023), Devilliers et al. (2024), could be added here.
L35: As far as I can see, these studies all cover the historical period. Have there been any studies on the impact of glacier meltwater on ocean circulation in future projections? Or in more idealized setups?
L41: “roughly half that of the GrIS”: in which period? Some values would also be insightful
L42: increased compared to what?
L56: Is there no observational study that measured submarine melt (and could be cited here)?
L92: Since it is referred to several times later, what is the physical or numerical motivation behind increasing the mixing at discharge locations in NEMO?
L102: The causality between lack of data and computational cost of a model is not clear to me.
L105: Reference Fig. A2 here (and consider moving that figure to the main text). Why use exactly this set of glaciers and not, for example, also the Scandinavian glaciers?
L141: “halfsolid” is a rather unintuitive name, it sounds more like a sensitivity run to test different liquid-to-solid ratios. Please consider renaming it.
L146: Remove or clarify the parenthesis here, as “baseline” could be ambiguous (Dai and/or Bamber runoff?) to the reader that doesn’t remember Section 3.1 in detail.
L146 and following: Why not just quote the numbers for Greenland (which are prescribed and should be the same across both runs) and the contribution from OGGM separately? This would also help reduce the confusion mentioned in the general comments.
L149: As a check, it would be good to compare these 4 mSv to observational estimates (e.g. the “glacier and ice cap runoff outside of Greenland” from Bamber et al. 2018)
L150: “approximately half the freshwater amount released to the ocean due to GrIS mass loss”: please provide a reference. Or remove this part, since for the ocean circulation, it does not really matter whether the freshwater is due to mass loss or not.
L214-220: I do not find that the choice of integrating a relatively large volume (area*depth range) is justified well enough here. It is true that we don’t know whether picking a single cell “actually reflects water properties at the glacier front best”, but certainly this would be the simplest choice, and therefore the authors should argue why averaging over a 50km+ radius may reflect the properties at the glacier front more accurately. Similarly, I do not know how the bathymetry of this NEMO configuration looks like exactly, but it might vary between different (near-)coastal grid points, and averaging over different depths may introduce some unintended artifacts. It would be good to check (perhaps for some selected glaciers) how sensitive the melt rates are to these choices. Or is this already covered by one of the parameters in the latin hypercube ensemble?
L230-234: I suggest removing this note about water masses. It sounds overly negative and does not add anything to the analysis.
Results (general comment): I commend the authors for incorporating an interactive iceberg model into their setup, but unfortunately no results from this model component are shown. For example, it would be very interesting to see a map of the meltwater distribution from icebergs released from the glaciers outside of Greenland, which should also help interpreting some ocean changes away from the coasts.
L240: Already say here that the surface is colder than the 0-200m average (as you mention later)
L241: Is mixing only increased at coastal points that receive runoff or everywhere?
L247: For stratification, it would be informative to show a comparison of the potential density (and maybe T-S) profiles averaged over a suitable region.
L253: Why does the east-west SSH gradient change qualitatively in the same way when adding freshwater at either the eastern or western side? This does not seem intuitive.
L279: Does Fig. 6 show annual means? If the focus is on deep convection, would it be more useful to show/discuss March or winter means (at least for mixed layer depth) instead?
L281: But surely the interannual variability is not independent between the two model runs, since some/most of it is determined by the surface forcing?
L304: Here it could be mentioned that the lack of a signal in the deep ocean (e.g., the AMOC lower limb) could also be due to the short simulation time, especially since the freshwater signal will take some time to transit from the main input regions to the subpolar North Atlantic.
L312: Percentage changes of FW flux are not really useful, since they depend on the reference salinity (e.g., Schauer & Losch 2019), especially if, like here, the baseline FW flux is not too far from zero. If the aim is to decompose FW transport anomalies, it would be more appropriate to apply a decomposition into a velocity anomaly and a salinity anomaly (and a nonlinear residual) term.
L314: “freshwater input leaving through the BSO is salinified”: The two opposing pairs of words make this statement very confusing.
L317 and Fig. 10: How is “positive volume flux” defined here?
L346: Especially for the western Greenland Sea, I would also expect sea ice advection from the Arctic Ocean to play an important role in setting sea ice thickness. Could you check the differences in sea ice volume transport across Fram Strait and the BSO between your two simulations?
Fig. 10a+d: The p-values (not significant) and line styles (significant) are contradicting each other.
L358: I also find it surprising that there is less sea ice in the “halfsolid” run, I would have expected the opposite for two reasons: a) the meltwater should strengthen the halocline, and b) the latent heat flux from melting icebergs should cool the ocean surface locally. Could you comment on this a bit more? Plotting the seasonal cycle of sea ice thickness and a comparison with Marson et al. (2021) might provide some insights.
L359: It is unclear what you mean by “structural changes”
Fig. 14/Table 1: Layperson question: Is there no contribution from surface runoff or basal melting from these glaciers?
L403: It would be good to remind the reader what the Castro de la Guardia et al. paper was about.
L403 and following: Why compare only with Castro de la Guardia et al.? How about other papers from the introduction? Of course, the setup is not identical, but especially for regional features Devilliers et al. (2021, 2024) might provide a useful comparison with a lower-resolution ocean model.
L417: I do not find the comparison with the PIOMAS trends from Labe et al. very convincing, since from my reading their magnitude (over the entire 1979-2015 period) is at least one order of magnitude larger than your values.
L421-434: The discussion around individual figures is probably a bit too detailed here – one would need to flip back and forth between this paragraph and the figures (in the results section) quite a bit.
L473-485: Which of these many possibilities do you find most interesting/relevant? It might be worth focusing this part a bit.
L490: Do slower quantities (temperature, salinity, sea ice) not suffer from a (potentially longer) initialization shock either?
Fig. A17: Are you sure the units of this figure are correct? 150 Sv is on the order of the ACC transport…
L549: “even” Why is this surprising?
Some additional questions on the NEMO setup:
- How is the baseline (Dai et al.) runoff treated over the OGGM model area? Surely some river runoff in this area is sourced from (seasonal) glacier melt in the first place, so how is double-counting avoided?
- Why was this specific modelling period chosen? 2010–2019 is quite short (probably too short for the AMOC, see above) and it is unfortunate that it does not fully overlap with the Bamber et al. coverage.
- Is any salinity restoring used?
Technical corrections:
L84: Put a line break here
L96: Enderlin 2016 say “up to” 50%
L346 simulation -> simulations
L404 units are missing
L541 and following: The use of past and present tense is inconsistent here, please decide on one of the two.
Carret, A., Llovel, W., Penduff, T., & Molines, J.-M. (2021). Atmospherically forced and chaotic interannual variability of regional sea level and its components over 1993–2015. Journal of Geophysical Research: Oceans, 126, e2020JC017123. https://doi.org/10.1029/2020JC017123
Devilliers, M., Yang, S., Drews, A., Schmith, T., & Olsen, S. M. (2024). Ocean response to a century of observation-based freshwater forcing around Greenland in EC-Earth3. Climate Dynamics. https://doi.org/10.1007/s00382-024-07142-0
Martin, T., & Biastoch, A. (2023). On the ocean’s response to enhanced Greenland runoff in model experiments: relevance of mesoscale dynamics and atmospheric coupling. Ocean Science, 19, 141–167. https://doi.org/10.5194/os-19-141-2023
Schauer, U., & Losch, M. (2019). “Freshwater” in the Ocean is Not a Useful Parameter in Climate Research. Journal of Physical Oceanography, 49, 2309–2321. https://doi.org/10.1175/JPO-D-19-0102.1
Swingedouw, D., Houssais, M.-N., Herbaut, C., Blaizot, A.-C., Devilliers, M., & Deshayes, J. (2022). AMOC Recent and Future Trends: A Crucial Role for Oceanic Resolution and Greenland Melting? Frontiers in Climate, 4, 838310. https://doi.org/10.3389/fclim.2022.838310
Citation: https://doi.org/10.5194/egusphere-2024-1425-RC1 - AC1: 'Reply on RC1', Jan-Hendrik Malles, 15 Oct 2024
-
RC2: 'Comment on egusphere-2024-1425', Anonymous Referee #2, 08 Aug 2024
This manuscript presents a detailed look into the influence of additional freshwater from outside of Greenland via a one-way coupling scheme with NEMO and glacier model. They run two experiments including freshwater input from glacial mass loss from the Arctic versus solely including freshwater input from Greenland. The authors find changes in circulation pathways, gyre strength, increases in ocean heat content, particularly in Baffin Bay, and associated changes in sea ice thickness.
I thank the authors for the thorough analysis and intriguing one-way, ocean-glacier model coupling setup. The paper is relevant and worthy of publication. However I do have some clarification questions and comments.
General commentsBoundary currents:
I am curious as to how boundary currents are represented in the simulations. As near surface, shelf currents transport the majority of freshwater input from the coast, there is a certain degree that must be resolved. Is 1/4th degree sufficient? What is the role that shelf currents may play on freshwater pathways, particularly in the Canadian Arctic Archipelago which is made up of a network of straits and basins?Statistical significance:
Given the relatively “short” time series of 2010 - 2019 and the primary difference between the two model runs is the freshwater input, could one use bootstrapping to test for significance? There is very little said on the reasoning behind using the Wilcoxon signed-rank tests. Why that method in particular versus others? Could more be added in the methods on the choice and reasoning behind that statistical significance test?NEMO model description:
The detailed description of the glacier model, one-way coupling scheme, and submarine melt parameterization is great. However it may be worthwhile to have some more detail on the NEMO model and setup itself; for example: choice of salinity restoring rate, integration time, etc. Could NEMO characteristics be elaborated on in the Data and Methods description?Figures:
There are a lot of figures. I am wondering if a few could be merged or show multiple plots in one figure. For example having Figure 11, A13, and A14 side by side or in a gridded way would allow a clearer way to visualize the increase in temperature and salinity in the eastern Arctic ocean (L335 - 340). Due to the large number of figures and quite a few of them being related and linked to one another in the text, it may facilitate the ease of reading when some figures are presented simultaneously together.Model evaluation:
There is very little quantitative model evaluation. There are many references qualitatively attributing the results to other studies but these are primarily model studies i.e. Castro de la Guardia et al. 2015. I would be interested in a brief discussion on how good a job NEMO and/or the glacier model does simulating processes. There is a gridded monthly mean data set based on moorings at Fram Strait (78.5N) from 2003 - 2019 by Karpozoglou et al. 2022 that could potentially be used for some additional comparison. For example, one could compare the volume transport or temperature depth averaged over 200m across the section with the halfsolid model simulation.Synthesis of discussion:
The discussion section jumps around a bit. There’s a lot of great information on future work, potential deficiencies of the model etc. but it is a bit discombobulated. It would be helpful to synthesize the discussion section more. I also think it may be worth splitting the discussion into subsections i.e. differences between the two model runs, fjord/glacier representation, modeling approach, etc.Specific comments
L10: Regarding the abstract, there could be a few more words on the reasoning behind the increase in Baffin Bay’s heat content. Particularly some more specifics could be given, rather than just stating “changes in the subpolar gyre’s structure”.L35-37: Can some magnitudes of Greenland melt be offered? For example, artificial hosing experiments typically consist of .1 Sv (Jackson et al. 2018) while current magnitudes of Greenland FWFs are at .01 Sv (Martin et al. 2022)
L56: Would replace the word intricate with complex
L70: Rather a long sentence. Would recommend shortening it or breaking it into two parts
L85/L89: Does Dai et al. 2009 provide only the continental runoff i.e. river discharge or does it also provide glacial runoff? Bamber et al. 2018 provides freshwater fluxes from Greenland and the Arctic, so do you include both freshwater from continental runoff (Dai et al. 2009) as a baseline and also add runoff from Bamber simultaneously? Can you clarify which freshwater forcing you use and when per data set, as this part is not clear for me
L86-L90: Since both the Dai et al. 2009 and Bamber et al. 2019 freshwater forcings do not cover the entire time period (2010 - 2019), can a sentence be added on what may be assumed using some climatological means? Particularly as 2019 was an anomalous year of Greenland melt and using a mean of 2010 - 2016 will likely smooth out that interannual variability. What are the potential implications of using a 2010 - 2016 mean?
L92: Why does the additional runoff entail an increase in the diffusivity? Can this be elaborated on further? Would this affect local sea ice formation along the shelves?
L121: What is the influence of basal melt and how is it represented in the glacier model? Are basal melt rates typically smaller than submarine melting rates?
Fig A2: It’s a bit unclear to me if the freshwater forcing (liquid and solid) is distributed uniformly around the Canadian Arctic? Bamber et al. 2018 non-uniformly distributes freshwater along the coastlines. Can this be reiterated or stated in section 2.4 OGGM to NEMO?
L149-L150: I am confused about this sentence, the OCGM has 4 mSv of more freshwater than in the halfsolid run, which is about half of the freshwater released from Greenland? Can this be clarified?
L214: Can you clarify why you use a 50km radius upon obtaining the thermal forcing values?
L219: Over which depths do you average? The full depth range near the terminus of the glacier? Is this a relatively shallow depth range?
L229: This is a long sentence. I would recommend rephrasing something along the lines of: “we use the term Atlantic for water (warmer and saltier) moving from the Atlantic towards the Arctic Ocean, and the term Arctic for water (cooler and fresher) moving in the opposite direction.”
L235: Can a reference or two be added confirming most of the oceanic changes occurring in these depth ranges (0 - 200m and 200 - 600m)?
L239: It appears the cooler area along the western coast is not statistically significant, based on the significance testing
L247: Regarding stratification, when looking at difference maps of stratification say averaged over the top 200m, do you see an overall positive difference between the two runs?
L254: Why might there be such a discrepancy in the increase in SSH gradient at the east compared to Castro de la Guardia et al. 2015? Is it due to the freshwater forcing magnitudes?
Fig 3: I’m a bit confused as to the difference between the dashed and solid blue lines. The horizontal lines represent the differences between the two runs which are statistically significant? I would suggest adding to the legend indicating the black, blue (dashed/solid), and red lines. This may help to clarify the figure.
L273: Can you elaborate on the vertical mixing coefficient here as a way to remind the readers, i.e. more mixing resulting in a potential decrease in temperature in the upper 200m layers?
L280: Is the MLD averaged over the whole year, or is it during winter months? Please clarify. I would recommend winter months (JFM) or at least March as the MLD is likely deepest then
L299: Why do you pick 47N latitude to look at the AMOC and transport, can this be justified more? There may not be huge differences but if you take say the OSNAP West section to compute the volume transport, do you get a similar result?
L309: This may be the first time introducing the Barents Sea Opening and thus the acronym, please add (BSO)
L320: Can more be clarified, potentially in the methods and model setup description (see general comment), on the NEMO integration time? Or refer the readers to check the model setup discussion?
L338: “Relatively strongly increased salinity” appears to have opposing meanings, is it a strong or weaker increase in salinity? Can this be rephrased?
L344: Can these plots A14 and A15 be placed side by side or consolidated in some way? It may allow the reader to more clearly see the effect that changes in salinity has on SSH in the interior Arctic (see general comment as well)
L346: On the part about sea ice thickness, assuming that much of sea ice is advected from the Arctic and thus can influence the thickness, particularly along East Greenland via Fram Strait. Can you comment more on sea ice advection? Or one could check the sea ice thickness change over say an averaged area (i.e. western Greenland sea) or cross section (Fram Strait) and compare that with the change in sea ice production
L357: Rework the sentence starting with “That there is a net sea ice thickness…”. I would recommend saying instead: “The decrease in net sea ice thickness in the northern hemisphere between the two NEMO experiments is intriguing…”
L403: Can you explain what the major results from Castro de la Guardia et al. 2015 are to remind the reader? Also highlight the differences between yours and their experiments, as you say the model setup and scope is very similar
L404: “lower increase” sound a bit opposing to me, perhaps reword to “smaller increase”
L413: Rather than say the gradients did not change sufficiently, I would refer to the statistical significance of the SSH gradient differences between the two model experiments
L545: I would cite that previous study i.e. Castro de la Guardia et al. 2015 to remind the readers which study you are referring to
L554: “also goes along with” sounds a bit colloquial, I would suggest alternative wording
L561: Given this small increase in submarine melt, can a research outlook or statement be made on the contribution of submarine melt when including additional glacier-sourced freshwater input? For example, should future model sensitivity studies on freshwater input consider including submarine melting and associated diffusivity scheme at depth? What would the authors advise for future modeling studies?
L573: What is meant by decadal snapshot simulations? Decadally averaged or filtered simulations? Can this be elaborated?
References
Jackson, L. C., & Wood, R. A. (2018). Hysteresis and resilience of the AMOC in an eddy-permitting GCM. Geophysical Research Letters, 45, 8547–8556. https://doi.org/10.1029/2018GL078104
Martin, T., Biastoch, A., Lohmann, G., Mikolajewicz, U., & Wang, X. (2022). On timescales and reversibility of the ocean's response to enhanced Greenland Ice Sheet melting in comprehensive climate models. Geophysical Research Letters, 49, e2021GL097114. https://doi.org/10.1029/2021GL097114
Karpouzoglou, T., de Steur, L., Smedsrud, L. H., & Sumata, H. (2022). Observed changes in the Arctic freshwater outflow in Fram Strait. Journal of Geophysical Research: Oceans, 127, e2021JC018122. https://doi.org/10.1029/2021JC018122
Citation: https://doi.org/10.5194/egusphere-2024-1425-RC2 - AC1: 'Reply on RC1', Jan-Hendrik Malles, 15 Oct 2024
Data sets
Model output for: "Freshwater input from glacier melt outside Greenland alters modeled northern high-latitude ocean circulation" Jan-Hendrik Malles https://doi.org/10.5281/zenodo.10468082
Model code and software
jmalles/oggm/submarine_melt Jan-Hendrik Malles https://doi.org/10.5281/zenodo.10468696
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
244 | 107 | 25 | 376 | 16 | 19 |
- HTML: 244
- PDF: 107
- XML: 25
- Total: 376
- BibTeX: 16
- EndNote: 19
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1