the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 14 Jul 2025
Global Climate System Response to SOFIA Antarctic Meltwater
Abstract. As the climate continues to warm, the Antarctic Ice Sheet (AIS) and its surrounding floating ice shelves are becoming increasingly susceptible to rapid collapse. Despite the potential impact this poses to the global climate system, the effects of AIS meltwater are not considered by most existing coupled climate models, including those in the most recent Coupled Model Intercomparison Project (CMIP6). As such, there remains much uncertainty over the impact of this additional meltwater on current global climate change projections. Here, we use the coupled atmospheric-ocean general circulation model HadCM3-M2.1 to study the effect of a continuous meltwater discharge from the AIS on the global climate system. This involves carrying out a series of freshwater hosing experiments based on the newly proposed Southern Ocean Freshwater Input from Antarctica (SOFIA) Initiative. Due to the relative computational efficiency of the HadCM3-M2.1 model, we are able to explore longer timescales than is usual. We find that ∼1000 years of continuous meltwater drives global atmospheric cooling, sea ice expansion in both hemispheres and a northward shift of the Intertropical Convergence Zone (ITCZ). The resulting freshening of the global ocean results in the weakening of both the AMOC and AABW. This triggers pervasive ocean warming at depths greater than 5000 m. An additional sensitivity study is also conducted in which the sensitivity of the climate model response to a change in the horizontal distribution of AIS meltwater is tested. As a result, we find that the manner in which the AIS loses mass, whether that be predominately through iceberg mass loss or basal melt, is unlikely to affect the global climate response to AIS meltwater.
- Preprint
(5982 KB) - Metadata XML
-
Supplement
(1962 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-2750', Nicholas Golledge, 06 Aug 2025
Mistray et al: “Global Climate System Response to SOFIA Antarctic Meltwater”
Review by N R Golledge 6th Aug 2025
This paper reports results from a suite of experiments using the climate model HadCM3-M2.1, in which a freshwater flux is added to the ocean around Antarctica to mimic the effect of ice melt over the current and future centuries. The description of the methodology and results is relatively clear and the figures are adequate to illustrate the key points raised in the paper. In terms of novelty, this is, ultimately, just another hosing experiment, of which there are already quite a few, and the results are very much in line with previous studies. But the fact that the perturbation experiments are run for 1150 years means that it is one of the few studies that explore long-term climate responses to SO hosing, which is useful. Treatment of previous literature is rather superficial, often defaulting to review papers or the same few studies repeatedly. There are also relevant papers that are absent from the reference list, and some of the claims in the paper seem to be rather bold as a consequence.
Overall, my feeling is that the manuscript is probably a useful contribution to the literature, but could be improved with some closer attention to the points detailed below.
line 27: you mention that previous studies used ‘idealised freshwater hosing experiments’, but don’t clarify what you mean by ‘idealised’. This is important, because your study uses a low-resolution highly-parameterised GCM and imposes a constant salinity flux as a proxy for meltwater. This is what I would call ‘idealised’, since the meltwater flux is entirely fictional, it does not derive from any ice sheet model. By contrast, the papers you cite in lines 30-31 do a far more robust job with the meltwater - Sadai et al used a coupled ice sheet - climate model, Golledge et al used an offline-coupled ice sheet - climate model, and Bronselaer used meltwater fluxes from a previous ice sheet simulation (DeConto & Pollard, 2016). In all three cases, the meltwater added to the climate model comes from a dynamically-evolving ice sheet model, and, in the studies of Sadai and Golledge, that ice sheet evolution is in turn coupled to the evolution of the simulated climate. I wouldn’t therefore call them ‘idealised’.
line 32: again, you mention ‘interactive ice sheets and shelves’ as being absent from previous work, but they are also absent from your model, so...?
line 46-7: I assume here you mean future projections? Because there are several studies that consider multi-centennial to multi-millennial hosing under palaeo scenarios (Weaver et al., 2003; Menviel et al., 2010; Bakker et al., 2017; for example).
line 63-4: Noble et al and Bentley et al are both review papers, and PD09 do not mention meltwater pulses, only ice growth and decay, so why were these references chosen specifically? There are ice sheet simulations published that explicitly focus on AIS meltwater pulses (Stuhne & Peltier 2015; Golledge et al., 2014), so perhaps the referencing could be broadened to at least acknowledge where original work on this topic has been done?
line 68-70: “The results produced here will provide novel insight into the global climate implications of rapid AIS melt expected within this century and beyond” - I’m not sure how well the study succeeds in providing novel insights - perhaps a summary could be added here or later that identifies what the study shows that is consistent with existing knowledge, and what has been found that is new?
line 95-102: there is a long list here of the various parameterisations employed, but I didn’t find a discussion of how these approximations might influence the results? Are there studies that could be referenced here that have shown eg that eddy-resolving models show xx greater or lesser mixing / transport / whatever? I feel like some justification / explanation is needed here.
line 105: “excess salt formed by melting” - I’m sorry if I misunderstand what is meant here, but I thought it was the formation of sea ice (ie by freezing) that rejected salt?
line 111 and throughout: why the use of future tense here? The work has been done, so either use past tense or present.
line 119: Regarding the addition of meltwater at the surface, this is fine and a common approach, but maybe mention the limitation of it. For example, Pauling et al 2016 showed that releasing meltwater at the depth of ice shelf bases can have an influence on its impact: “We found that injecting water at the depth of the front of ice shelf around Antarctica caused the ocean mixed layer to deepen, while adding freshwater at the surface caused the mixed layer to shoal.”
line 126: ‘perturb from’ sounds odd, maybe ‘branch from’?
line 128: “The difference between the piControl and FW experiments taken as modelling responses.” - seems like some words are missing from this sentence.
line 132: “at the level 95%” --> “at the 95% confidence level” perhaps?
line 139-140: “To date, this is the first study to investigate the climate response to AIS meltwater using a fully coupled climate model over such long time scales.” - Given the limitations of the model and methodology, I would suggest tempering this statement a little. Other studies as identified above have done similar things, for example Bakker et al used actual ice sheet meltwater fluxes and ran for 5000 years.
line 148-150: “The area immediately surrounding the AIS experiences a decrease in SSS of up to 0.5 psu (Fig. 3h). Freshening around the AIS is relatively uniform, with no particular region experiencing significantly more freshening (Fig. 3f-h).” - This is the area you apply the meltwater, right? And that is applied as a salinity anomaly? So these aren’t really 'results', this is just what you've put into the model.
line 153: ‘weaker rate’ - ‘lower rate’ maybe?
line 154: ‘This freshening’ - always best to avoid using ‘this’ unless you clarify what it refers to, eg ‘The surface freshening of xx psu described above...’
line 166: ‘ran’ --> ‘run’
line 184-5: NB - Golledge et al 2019 also used freshwater fluxes from the Greenland Ice Sheet, which most likely affects N.Atl SSTs etc.
line 189-194: You refer to ‘this unusual trend’ and ‘this trend’ and an increase in global sea ice thickness of c. 0.1m. But I don't see any trend in any of the data in 6a or 6b. There is a lot of variability, but how have you calculated a thickening trend? And how can the global increase be 0.1m when the plot in 6a shows values of only 0.07 to 0.1? I think the step change between picontrol and the forced run is 0.01m? It is 0.1m in the SO experiment.
line 200-1: Is this because the location of meltwater input in Sadai et al is controlled by locations of mass loss in the ice sheet model? Or is there some other reason?
line 233: Maybe also look at Golledge et al 2019 for how fluxes from AIS and GIS affect AMOC differently.
line 234 and line 236: I don’t think you can make the claim at 236 when you just made the statement at 234!
line 285-6: Can you say anything about the effect of icebergs, eg from simulations that explicitly simulated them (eg Schloesser et al)?
line 294-8: Note that this may be different if GIS meltwater was included in the expts.
line 309: rather than ‘dramatic’ I would just say ‘abrupt’ - this rapid change just reflects your methodology of meltwater addition, correct?
Citation: https://doi.org/10.5194/egusphere-2025-2750-RC1 -
RC2: 'Comment on egusphere-2025-2750', Neil Swart, 09 Sep 2025
This paper describes the results of implementing the standardized SOFIA Antarctic meltwater forcing protocol into the HadCM3-M2.1 model. A uniqueness is running the simulations for an extended period (1000 years), relative to most previous studies. The paper is a straightforward description of how the meltwater forcing influences key ocean and atmospheric variables, including ocean salinity, temperature, surface temperature and sea-ice. The results are largely consistent with previous work. There is value in the incremental documentation of the climate impacts in a new model.I do not have any major issues. Below I offer some suggestions to help clarify and perhaps increase the impact of the paper on the community.Suggestions by line:Title: Perhaps including "in HadCM3-M2.1" would help to clarify the scope for readers?3: >"the effects of AIS meltwater are not considered by most existing coupled climate models": you do have to be careful here, because ice sheet dynamics are not considered by these models, but most if not all of them do permit surface melt which does add additional freshwater under warming (indeed, this can be an infinite source of freshwater in many models).18-25: Between para's 1 and 2 there is an implicit link between sea level rise and meltwater input to the ocean. Just noting that these are not strictly the same thing (because melt of floating iceshelves is a freshwater forcing, that does not influence SLR). Not a huge deal, more of a note.40: It is also important to use a consistent design across the community - which you are contributing to (and is a valuable part of the paper) - I note this is mentioned near ln 7054: I wonder if there is a slightly more constructive way to frame this? Such as "it's unclear whether results from simplified EMICs would hold in fully coupled AOGCMs".108-109 / 119: It would be useful to provide more detail on the converting of the FW flux to a virtual salt flux, since this is presumably the same mechanisms used to implement the SOFIA experiments (not mentioned on 119 but should be).121: >"although run over a longer time scale (Table S1)" - the referenced experimental design specifies a length of >=100 years. It is good that this work exceeds the minimum, but its seems entirely consistent with the proposed design.125: Please state what the magnitude of this climatological flux is. In most similar models, this flux is roughly specified to balance P-E over Antarctica. If that is true, it is not a small flux.143: As written, this does not seem to quite make sense to me. A positive salinity trend cannot be due to (positive down) freshwater flux. A positive salinity trend might indicate that the piControl freshwater flux is smaller than P-E over Antarctica, and hence, P-E-R flowing over the ocean is less than 0 (i.e. effectively a negative freshwater flux). That is assuming everything else in the model is conserving of freshwater, which is not clear.Fig 3a/b and timeseries plots in general: The annual scale of variability is not really discussed in this manuscript, and it is very noisy. I think the results would be clearer is the annual mean lines were de-emphasized, and the 10 yr running averages were made more prominent.170 and surrounding. >"in line with observations from the same time period shown here". I find using the term "observations" confusing (with actual real world data). Also, the magnitude of freshening is compared with other studies, but it is not mentioned how the rates of forcing match or do not. If the freshening is similar, but the forcing is very different, this does not indicate a consistency. Please clarify.179: Is that the Labrador Sea? The map is small, but it looks more like Baffin Bay or Davis Straight to me.193: Global sea-ice thickness is not really such a useful metric, given the massive climatogical thickness differences across the hemispheres. Perhaps reporting as a change per NH/SH, and perhaps even a relative (%) change would be clearer?Figure 6 c-h: I think it would be more informative to saturate the high end of the colorbar, so that the broader pattern of thickness increase is visible. As it stands, we only see what happens near the edges. This is subjective, but I feel like we are missing the broader picture.235: What is the climatological AMOC rate in this model (in piControl)?230-260: It might be worth testing whether the AMOC/ACC timeseries are statistically different. As you say, there is a lot of variability.Figure 12: The changes in mid latitude precip are lost in the large changes near the equator. If you present the precip change as relative to the piControl (% as opposed to mm/day), it will take care of the climatological differences across latitudes, and perhaps be more meaningful (again, subjective).290: this explanation does not make physical sense to me, or I'm missing something. If injecting water around the coast makes it "able to spread horizontally across the surface of the SO before diffusing into the ocean depths", then injecting it even south of 60S would surely enable even more spread. Might a possible explanation might be that injection directly around Antarctica in antwater influences SO deep convection more directly than antwater60S, leading to more pronounced salinification at depth. I know Fig 18 shows similar AABW responses, but deep convection changes can be quite different to AABW (e.g. Chen et al. 2023)295: Could this just variability, even in a 50 year mean?314: This "significant impact" statement seems to contradict the earlier results that AMOC changes are very small, and even AABW changes are not huge (but we do not know the relative changes). I would contend that the statification mechanism that you mention next is likely more important.319: Sea-ice changes vs SAT changes are a bit chicken and egg. How much surface cooling could be explained by the ice-albedo feedback alone?359: DATA availability: The authors are describing the results of running a standardizes experimental protocol in their model. The SOFIA project has a data archive and active working group, but it would seem this data has not been contributed to that open data archive. It would be added value to contribute that data:https://sofiamip.github.io/data-access.htmlCitation: https://doi.org/
10.5194/egusphere-2025-2750-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 683 | 44 | 11 | 738 | 17 | 44 | 45 |
- HTML: 683
- PDF: 44
- XML: 11
- Total: 738
- Supplement: 17
- BibTeX: 44
- EndNote: 45
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1