the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 26 Jul 2023
Brief Communication: On the mid-summer melt pond fraction–September Arctic sea ice extent relationship in the EC-Earth3 climate model
Abstract. In a recent study, the springtime melt pond fraction has been suggested to be a predictor of subsequent September Arctic sea ice minimum extent anomalies. However, another study based on satellite data did not provide evidence for such a relationship. We explore this association in EC-Earth3, which includes an explicit treatment of melt ponds, for the present-day climate. We find a statistically significant inverse relationship between September sea ice extent and mid-summer (June–July) melt pond fraction on the seasonal scale. Our results support the satellite-based inferences that the mid-summer melt pond fraction highly correlates with the September ice extent.
- Preprint
(854 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2023-1560', Anonymous Referee #1, 24 Aug 2023
General Comments
This manuscript evaluates a simulation produced using a global climate model to explore whether melt pond fraction over the period May – August is a useful predictor of September Arctic Sea ice extent relationship, with a specific aim of resolving existing disagreements between prior studies on this topic. In general, developing understanding of the predictability of Arctic sea ice is a valuable area of research, both in terms of developing new physical insights into the Arctic sea ice system and in terms of improving sea ice forecasting methodologies. Similarly, this manuscript does clearly link to and build upon the wider literature. However, the novelty and impact of this research at present seem limited. This manuscript is concise, and generally well-written. The figures are of a good standard and present the results clearly. The methodology is appropriate but requires further explanation and justification.
Overall, I believe that this paper is within the scope of this journal, but I am not convinced that the paper has sufficient impact and novelty. The manuscript clearly states that its purpose is to address a question prior studies have reached different conclusions to, however it is not clear that this manuscript has made any substantial advances in addressing this question. The methodology applied is also not unique, with there being an existing study using a very similar approach to explore the relationship between melt pond fraction in the earlier melt season and September Arctic Sea ice extent, although this prior study uses a different climatology and with a different research focus.
I believe that the potential clearly exists for this manuscript to extend our understanding of the research question under consideration, but to do so the manuscript needs reframing to more clearly highlight how the methodology used has produced insights that prior methodologies, both observational and modelling, were unable to achieve. There are also prospects for additional simulations and analysis that could add substantial value to this manuscript. Whilst these concerns are substantial, if they can be adequately addressed then there are grounds for the paper to be accepted.
Specific Comments
- General Point: As mentioned above the manuscript, as currently presented, appears to have limited impact and novelty, and requires at the very least reframing, if not further simulations. In particular, if there is a unique aspect to this study, it is the use of a fixed greenhouse gas forcing to capture present day sea ice climatology. Despite this, this point is mentioned briefly in the introduction and methodology sections and not reflected upon in the results and conclusions section when comparing the results of this study to prior studies. It would help if the conclusions were framed in terms of the methodology used here and the insights that it enables e.g. the authors mention early in the manuscript that relationships that emerge using this methodology are more likely to be causal than those using a time-varying greenhouse gas forcing, so this point could be revisited and reflected upon in the discussion / conclusions.
- Abstract: A clear opening statement about the broader relevance of this research would be helpful. I also find the abstract to be written in a confusing manner as the authors initially discuss the springtime meltpond fraction when discussing previous research before talking about mid-summer melt pond fraction for this present study, without making a clear distinction between the two. Whilst this phrasing makes sense in the context of the full article, it could make the abstract confusing in a standalone context.
- P1L22: ‘Melt ponds potentially exert’. There are modelling studies you could reference here that support this point e.g. Roekner et al., (2012); Diamond et al., (2022).
- P1L28-39: Diamond et al. (2022) has also inspected this relationship. Furthermore, this study uses a very similar methodology to the one used here performing climate simulations with fixed forcing, but instead considering both pre-industrial sea ice climatology and sea ice climatology during the last interglacial period.
- P2L31: ‘independent observational data’. It would be useful to give specific details of the observations used in this study given the relevance of the study to this manuscript.
- P2L34: It would be helpful to briefly state the methodology of Feng et al. (2022) / make clear how it is different to and compliments the results found in Liu et al. (2015).
- P3L64: 95 years seems quite a short period of spin up for a climate simulation, with some studies using much longer periods for the spin-up of a climate simulation (e.g. Menary et al., 2018). What is the justification behind the assumption that the model has achieved stability?
- P3L66: Did you consider repeating this experiment with different greenhouse gas forcing (e.g. 1980, 2020)? It is possible this would be computationally expensive, but doing so would add substantial value to this research both in establishing whether the simulations with 2000 forcing was truly representative of present-day climatology, and in establishing whether the melt pond fraction–September Arctic sea ice extent relationship might be strengthening or weakening. Alternatively, in your discussion and conclusions you could reflect on how the results here compare to the similar study performed in Diamond et al. (2022) using pre-industrial forcing.
- P3L83-84: I think you need to more clearly state that you do not expect your forced system to replicate the observed behaviour since you are using a fixed greenhouse gas forcing - perhaps the idea here is that you want to make sure your simulation nevertheless produces broadly comparable behaviour to the observed system over 1979 - 2020 to demonstrate that it can be taken as representative of present day Arctic sea ice climatology, but this needs to be explained if so. Otherwise, the purpose of the comparison to observations is unclear.
- P3L90-91: ’It may be because of unexplained factors beyond this research’. You are trying to compare a system with fixed forcing to one with time-varying forcing so perhaps differences are to be expected.
- P4L100: ‘the model slightly underestimates’. This seems like an understatement e.g. observed fractions 2-4 x higher in May than the value in the climate model. Can you reliably use the model to assess predictability of September sea ice extent based on May melt pond fraction when the melt pond fraction in the model in May is so small and substantially lower than in observations?
- Figure 2: The value of p appears to be constant within error from June – August. This seems like a notable result and something worth commenting more or within the manuscript.
Technical Comments
- P2L46: ‘topographic scheme’. It may be worth stating this as ‘topographic melt pond scheme’ for clarity.
- P3L73-75: State the algorithm used for the concentration product (NASA Team algorithm version 1?).
- P3L88-89: I assume you mean here, ‘12% less than the observed mean’.
Additional References
Diamond, R., Sime, L. C., Schroeder, D., and Guarino, M.-V.: The contribution of melt ponds to enhanced Arctic sea-ice melt during the Last Interglacial, The Cryosphere, 15, 5099–5114, https://doi.org/10.5194/tc-15-5099-2021, 2021.
Menary, M. B., Kuhlbrodt, T., Ridley, J., Andrews, M. B., Dimdore-Miles, O. B., Deshayes, J., Eade, R., Gray, L., Ineson, S., Mignot, J., Roberts, C. D., Robson, J., Wood, R. A., and Xavier, P.: Preindustrial Control Simulations With HadGEM3- GC3. 1 for CMIP6, J. Adv. Model. Earth Sy., 10, 3049–3075, https://doi.org/10.1029/2018MS001495, 2018.
Roeckner, E., Mauritsen, T., Esch, M., and Brokopf, R.: Impact of melt ponds on Arctic sea ice in past and future climates as simulated by MPI-ESM, J. Adv. Model Earth Sy., 4, https://doi.org/10.1029/2012MS000157, 2012.
Citation: https://doi.org/10.5194/egusphere-2023-1560-RC1 -
RC2: 'Comment on egusphere-2023-1560', Anonymous Referee #2, 04 Sep 2023
Review of 'On the mid-summer melt pond fraction-September Arctic sea ice extent relationship in the EC-Earth3 climate model'.
The authors use EC-Earth3 model, which includes an explicit treatment of melt ponds, to evaluate whether there is a relationship between the summer melt pond fraction and the subsequent September sea ice extent. The paper is succinct, and the figures are clear and the use of panelling keeps them to a minimum whilst displaying enough information. However, the novelty or motivation for the paper is lacking, as is the objectivity of some of the results. The abstract and introduction outline the uncertainty in the relationship between melt pond fraction and sea ice extent, which the current paper aims to clarify, however, the conclusions don't leave the reader feeling less uncertain, and the differences in studies seems to add other variables to the mix. I recommend major corrections to make the novelty clearer and further description of the differences between studies.
Major concerns
In the abstract you outline springtime relationships (or lack thereof) in previous studies, but you then study the mid-summer relationship. In the introduction you outline the three studies which vary in analysing May, May to June, May to July and June alone. You then add a fourth study, which investigates 'spring-summer' as it is called in line 38, but earlier is called 'mid-summer', and throughout the paper you seem to vary whether you are looking at May, June-July (line 144), integrated May through July (line 114). This makes it quite confusing to understand the study and how it compares. Additionally, repeating the same analysis for 1 more model/dataset doesn't 'shed light' on the conundrum but rather adds to it. As the model of choice includes the melt pond model, this could add novelty and importance to the study, but with the current phrasing and confusing methods/choices, this aspect is hidden.
Abstract
'Satellite data' is used in the second sentence to describe the second study, which found no evidence of a relationship. Then you use 'satellite-based' in the final sentence to confirm that there is a relationship. This is confusing, so find some different way to separate the studies. E.g by publication author and year?
Introduction
See major concerns.
Methods
Line 66-70: you use the statistically significant relationship between ice extent and melt pond fraction here to justify that the model is stable. But if there was no relationship found, would you have concluded that the model was unstable, or that no relationship was found. What other metrics could you use to justify the spinup period. Whilst it is important to discard spin up, justifying that it is stable by using the hypothesis you are trying to prove doesn't fit well.
Line 76: Is there no updated dataset as the Roesel et al (2012) paper is now 10 years old?
Line 77/79: Is something different meant by 'entire Arctic' to 'pan Arctic'?
Results
Line 126: May through July. Line 127: Mid summer. Mid-summer is only defined on line 144.
I am not convinced that there are enough results here for a publication – albeit a brief communication. If the results were very clear, then fine, but you agree with some of the Schroeder study, some from Liu and the Feng study is only bought in during the conclusion, and not in the discussion. As a reader I feel no further secure in the relationship between melt ponds and sea ice extent than before I read the paper.
Citation: https://doi.org/10.5194/egusphere-2023-1560-RC2
Status: closed
-
RC1: 'Comment on egusphere-2023-1560', Anonymous Referee #1, 24 Aug 2023
General Comments
This manuscript evaluates a simulation produced using a global climate model to explore whether melt pond fraction over the period May – August is a useful predictor of September Arctic Sea ice extent relationship, with a specific aim of resolving existing disagreements between prior studies on this topic. In general, developing understanding of the predictability of Arctic sea ice is a valuable area of research, both in terms of developing new physical insights into the Arctic sea ice system and in terms of improving sea ice forecasting methodologies. Similarly, this manuscript does clearly link to and build upon the wider literature. However, the novelty and impact of this research at present seem limited. This manuscript is concise, and generally well-written. The figures are of a good standard and present the results clearly. The methodology is appropriate but requires further explanation and justification.
Overall, I believe that this paper is within the scope of this journal, but I am not convinced that the paper has sufficient impact and novelty. The manuscript clearly states that its purpose is to address a question prior studies have reached different conclusions to, however it is not clear that this manuscript has made any substantial advances in addressing this question. The methodology applied is also not unique, with there being an existing study using a very similar approach to explore the relationship between melt pond fraction in the earlier melt season and September Arctic Sea ice extent, although this prior study uses a different climatology and with a different research focus.
I believe that the potential clearly exists for this manuscript to extend our understanding of the research question under consideration, but to do so the manuscript needs reframing to more clearly highlight how the methodology used has produced insights that prior methodologies, both observational and modelling, were unable to achieve. There are also prospects for additional simulations and analysis that could add substantial value to this manuscript. Whilst these concerns are substantial, if they can be adequately addressed then there are grounds for the paper to be accepted.
Specific Comments
- General Point: As mentioned above the manuscript, as currently presented, appears to have limited impact and novelty, and requires at the very least reframing, if not further simulations. In particular, if there is a unique aspect to this study, it is the use of a fixed greenhouse gas forcing to capture present day sea ice climatology. Despite this, this point is mentioned briefly in the introduction and methodology sections and not reflected upon in the results and conclusions section when comparing the results of this study to prior studies. It would help if the conclusions were framed in terms of the methodology used here and the insights that it enables e.g. the authors mention early in the manuscript that relationships that emerge using this methodology are more likely to be causal than those using a time-varying greenhouse gas forcing, so this point could be revisited and reflected upon in the discussion / conclusions.
- Abstract: A clear opening statement about the broader relevance of this research would be helpful. I also find the abstract to be written in a confusing manner as the authors initially discuss the springtime meltpond fraction when discussing previous research before talking about mid-summer melt pond fraction for this present study, without making a clear distinction between the two. Whilst this phrasing makes sense in the context of the full article, it could make the abstract confusing in a standalone context.
- P1L22: ‘Melt ponds potentially exert’. There are modelling studies you could reference here that support this point e.g. Roekner et al., (2012); Diamond et al., (2022).
- P1L28-39: Diamond et al. (2022) has also inspected this relationship. Furthermore, this study uses a very similar methodology to the one used here performing climate simulations with fixed forcing, but instead considering both pre-industrial sea ice climatology and sea ice climatology during the last interglacial period.
- P2L31: ‘independent observational data’. It would be useful to give specific details of the observations used in this study given the relevance of the study to this manuscript.
- P2L34: It would be helpful to briefly state the methodology of Feng et al. (2022) / make clear how it is different to and compliments the results found in Liu et al. (2015).
- P3L64: 95 years seems quite a short period of spin up for a climate simulation, with some studies using much longer periods for the spin-up of a climate simulation (e.g. Menary et al., 2018). What is the justification behind the assumption that the model has achieved stability?
- P3L66: Did you consider repeating this experiment with different greenhouse gas forcing (e.g. 1980, 2020)? It is possible this would be computationally expensive, but doing so would add substantial value to this research both in establishing whether the simulations with 2000 forcing was truly representative of present-day climatology, and in establishing whether the melt pond fraction–September Arctic sea ice extent relationship might be strengthening or weakening. Alternatively, in your discussion and conclusions you could reflect on how the results here compare to the similar study performed in Diamond et al. (2022) using pre-industrial forcing.
- P3L83-84: I think you need to more clearly state that you do not expect your forced system to replicate the observed behaviour since you are using a fixed greenhouse gas forcing - perhaps the idea here is that you want to make sure your simulation nevertheless produces broadly comparable behaviour to the observed system over 1979 - 2020 to demonstrate that it can be taken as representative of present day Arctic sea ice climatology, but this needs to be explained if so. Otherwise, the purpose of the comparison to observations is unclear.
- P3L90-91: ’It may be because of unexplained factors beyond this research’. You are trying to compare a system with fixed forcing to one with time-varying forcing so perhaps differences are to be expected.
- P4L100: ‘the model slightly underestimates’. This seems like an understatement e.g. observed fractions 2-4 x higher in May than the value in the climate model. Can you reliably use the model to assess predictability of September sea ice extent based on May melt pond fraction when the melt pond fraction in the model in May is so small and substantially lower than in observations?
- Figure 2: The value of p appears to be constant within error from June – August. This seems like a notable result and something worth commenting more or within the manuscript.
Technical Comments
- P2L46: ‘topographic scheme’. It may be worth stating this as ‘topographic melt pond scheme’ for clarity.
- P3L73-75: State the algorithm used for the concentration product (NASA Team algorithm version 1?).
- P3L88-89: I assume you mean here, ‘12% less than the observed mean’.
Additional References
Diamond, R., Sime, L. C., Schroeder, D., and Guarino, M.-V.: The contribution of melt ponds to enhanced Arctic sea-ice melt during the Last Interglacial, The Cryosphere, 15, 5099–5114, https://doi.org/10.5194/tc-15-5099-2021, 2021.
Menary, M. B., Kuhlbrodt, T., Ridley, J., Andrews, M. B., Dimdore-Miles, O. B., Deshayes, J., Eade, R., Gray, L., Ineson, S., Mignot, J., Roberts, C. D., Robson, J., Wood, R. A., and Xavier, P.: Preindustrial Control Simulations With HadGEM3- GC3. 1 for CMIP6, J. Adv. Model. Earth Sy., 10, 3049–3075, https://doi.org/10.1029/2018MS001495, 2018.
Roeckner, E., Mauritsen, T., Esch, M., and Brokopf, R.: Impact of melt ponds on Arctic sea ice in past and future climates as simulated by MPI-ESM, J. Adv. Model Earth Sy., 4, https://doi.org/10.1029/2012MS000157, 2012.
Citation: https://doi.org/10.5194/egusphere-2023-1560-RC1 -
RC2: 'Comment on egusphere-2023-1560', Anonymous Referee #2, 04 Sep 2023
Review of 'On the mid-summer melt pond fraction-September Arctic sea ice extent relationship in the EC-Earth3 climate model'.
The authors use EC-Earth3 model, which includes an explicit treatment of melt ponds, to evaluate whether there is a relationship between the summer melt pond fraction and the subsequent September sea ice extent. The paper is succinct, and the figures are clear and the use of panelling keeps them to a minimum whilst displaying enough information. However, the novelty or motivation for the paper is lacking, as is the objectivity of some of the results. The abstract and introduction outline the uncertainty in the relationship between melt pond fraction and sea ice extent, which the current paper aims to clarify, however, the conclusions don't leave the reader feeling less uncertain, and the differences in studies seems to add other variables to the mix. I recommend major corrections to make the novelty clearer and further description of the differences between studies.
Major concerns
In the abstract you outline springtime relationships (or lack thereof) in previous studies, but you then study the mid-summer relationship. In the introduction you outline the three studies which vary in analysing May, May to June, May to July and June alone. You then add a fourth study, which investigates 'spring-summer' as it is called in line 38, but earlier is called 'mid-summer', and throughout the paper you seem to vary whether you are looking at May, June-July (line 144), integrated May through July (line 114). This makes it quite confusing to understand the study and how it compares. Additionally, repeating the same analysis for 1 more model/dataset doesn't 'shed light' on the conundrum but rather adds to it. As the model of choice includes the melt pond model, this could add novelty and importance to the study, but with the current phrasing and confusing methods/choices, this aspect is hidden.
Abstract
'Satellite data' is used in the second sentence to describe the second study, which found no evidence of a relationship. Then you use 'satellite-based' in the final sentence to confirm that there is a relationship. This is confusing, so find some different way to separate the studies. E.g by publication author and year?
Introduction
See major concerns.
Methods
Line 66-70: you use the statistically significant relationship between ice extent and melt pond fraction here to justify that the model is stable. But if there was no relationship found, would you have concluded that the model was unstable, or that no relationship was found. What other metrics could you use to justify the spinup period. Whilst it is important to discard spin up, justifying that it is stable by using the hypothesis you are trying to prove doesn't fit well.
Line 76: Is there no updated dataset as the Roesel et al (2012) paper is now 10 years old?
Line 77/79: Is something different meant by 'entire Arctic' to 'pan Arctic'?
Results
Line 126: May through July. Line 127: Mid summer. Mid-summer is only defined on line 144.
I am not convinced that there are enough results here for a publication – albeit a brief communication. If the results were very clear, then fine, but you agree with some of the Schroeder study, some from Liu and the Feng study is only bought in during the conclusion, and not in the discussion. As a reader I feel no further secure in the relationship between melt ponds and sea ice extent than before I read the paper.
Citation: https://doi.org/10.5194/egusphere-2023-1560-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 274 | 95 | 25 | 394 | 22 | 21 |
- HTML: 274
- PDF: 95
- XML: 25
- Total: 394
- BibTeX: 22
- EndNote: 21
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1