the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 17 Mar 2026
Ocean and marine heatwaves responses to multiple net-zero worlds
Abstract. Climate change profoundly modifies the global ocean potentially threatening marine ecosystems. Under the Paris Agreement, the international community aims to limit global warming by achieving net-zero CO2 emissions, through the balance between CO2 emissions and removals. Yet, the response of the ocean to net-zero, and associated impacts, remains poorly understood. Prevailing Global Warming Level (GWL) approaches applied to analyse the impacts of climate change on the ocean overlook committed changes arisen after emissions cessation. Using the CNRM-ESM2-2 Earth system model, we perform 300-year net-zero simulations spanning +1.1 °C to +5 °C above pre-industrial levels, to investigate the ocean responses to CO2 emissions cessation, and how they depend on the warming level at which net-zero emissions is reached. Focusing on sea surface temperature (SST) and marine heatwaves (MHW) – key threats for marine ecosystems – we find substantial departures from transient responses throughout the 300-years of stabilization. At the Paris Agreement targets of +1.5 °C and +2.0 °C above pre-industrial levels, 26 % and 32 % of the global ocean, respectively, exhibit simultaneous increases in SST mean and variance as compared to transient warming simulations. These changes cover particularly the high latitudes and Southern ocean. MHW reorganize rapidly after emissions cease, with a poleward shift in frequency, declines in the tropics, and overall sustained intensity. Altogether, these non-transient responses indicate that standard GWL approaches underestimate MHW reorganization by decades to centuries. Net-zero simulations are therefore critical for robust projections of marine ecosystem risk at stabilized global warming levels.
- Preprint
(10641 KB) - Metadata XML
-
Supplement
(10838 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-993', Jerry Tjiputra, 14 Jul 2026
-
RC2: 'Comment on egusphere-2026-993', Anonymous Referee #2, 14 Jul 2026
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-993/egusphere-2026-993-RC2-supplement.pdf
-
RC3: 'Comment on egusphere-2026-993', Anonymous Referee #3, 21 Jul 2026
First I want to apologize to the authors for taking that long to provide my assessment of their manuscript. Other commitments have kept me exceptionally occupied over the recent weeks and months.
The manuscript of Bossert explores ocean responses to 300 year net-zero CO2 experiments branching based on the TipMIP protocoll. This is a very relevant contribution to the literature and the results on future evolution of ocean heat uptake, SSTs and heatwaves is very interesting.
I’ve made a few specific comments below, but have a few overarching comments that could further help to increase to sharpen the manuscript a bit.
- The authors may want to consider using IPCC ocean regions for their analysis. This can help to tease out additional regional features of ongoing changes that could be informative for assessing climate impact driver changes such as marine heatwaves. MHWs are particularly consequential in tropical regions and a bit more regional analysis beyond latitudinal bands would be useful.
- The authors go back and forth using net-zero CO2 and GSAT stabilisation almost synonymously. But these two things are not equivalent, obviously, and the authors themselves analyse and report ZEC for CNRM-ESM2-2. Being a bit more precise would be helpful, maybe even considering to explore different scalings of regional changes with GSAT changes post net-zero CO2.
Minor comments:
L7: Article 4.1 refes to a balance of sources and sinks of GHGs. Not ‘just’ net-zero CO2.
L 23: There is a profound misunderstanding here. The PA sets upper limits to global warming and does not specify temperature stabilisation. You may want to rewrite this introduction a bit.
L 27: Without being petty, arguably one of the first studies that has applied a GWL approach systematically in the context of the PA is Schleussner et al 2016.
Table 1: Not sure what the utility of this table is, tbh. I find it rather confusing. The model years are stylised and the approach could be clarified from Fig.1. From all I understand it’s ‘early stable’ after 10 years after transient, and then +140 years and +139 years (end of simulation). I think this could also be just be expanded on the illustration of Fig. 1. For readibility, it might also be a lot easier to start model simulations at year 1 and not at 1850 (and just remark in the methods that you start from that year in pre-industrial).
Also, looking at the table, it appears to me that panel c) of Fig. 1 is not quite correct. The ‘early stable’ period is not what is indicated. But simply at t_TWL+10.
L138: I’d consider revising the naming of the “stabilised warming levels” – as they in fact might not be as stable in terms of GSAT (would be a useful diagnostic if they actually are and under what experiment). Maybe revise them to Net Zero Warming states as they are defined in time and not with regards to their GSAT characteristics?
L168, Eq (2): I think this might not be quite correct. GSAT(t_0) also should be a 21yr mean?
Fig. 2: It might be nice to plot versions of panel b and c normalised to the start year of the branches to be able to directly see differences in time trends (they’re not visible from the current plots – consider also some smoothing of the time series).
L 244: I find this analysis really quite intriguing and wonder if the authors could unpack a bit more the processes through which AMOC changes would induce sustained imbalance. Because there are spikes in imbalances also for multi-decadal periods i.e. in the 1.1C branch, which may be linked to AMOC recovery and variability. Maybe they have another paper in the works on this, but if not a bit more unpacking here (and maybe some statistical analysis how much of the imbalance and ZEC? is explained by AMOC dynamics) would be interesting.
L 283/ Section 3.2: I find these results interesting, but it took me quite some time to understand the visual features of Fig. 6 and 7. I wonder, and I wonder how relevant the changes in variance really are for understanding some of the main changes in the features, which is differences in the mean depending on the GWL. So for low GWLs, SSTs decline almost everywhere. For high GWLs this is reversed and SSTs continue to increase in most parts of the ocean. Once this is established, and understood, IMO the discussion of the variance are the second order effect. Also, because the authors provide little explanation on what’s driving those changes in variance, potentially.
Fig 8: I understand why the authors go for latitude bands, but it might mask some interesting and relevant geographic features. What’s visible here is the ongoing dynamics in relation to AMOC and SMOC. But from an impact perspective, MHWs in particular in subtropical and tropical regions are of interest. This is currently not shown / the latitude focus puts the attention on the high latitudes. Arguably important, but maybe not the full story? Maybe the authors could consider showing some more maps / focus on a bit more on regionally differentiated dynamics (maybe analysing MHWs over IPCC ocean regions?).
L378: I’m not sure I agree. Yes, you only have single members, but you also have 300 years of model experiments. So I think you could be a bit more bold here, too.
L392: I presume the reason is because AMOC collapses in CNRM >4°C and not in ACCESS. Maybe this could be discussed here a bit.
Citation: https://doi.org/10.5194/egusphere-2026-993-RC3
Data sets
Ocean and marine heatwaves responses to multiple net-zero worlds - datasets and scripts for Figures 1 to 10 Isaline Bossert https://doi.org/10.5281/zenodo.18799136
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 733 | 594 | 59 | 1,386 | 829 | 78 | 83 |
- HTML: 733
- PDF: 594
- XML: 59
- Total: 1,386
- Supplement: 829
- BibTeX: 78
- EndNote: 83
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The study by Bossert and co-authors investigates the response of MHW evolution following cessations of CO2 emissions at various global warming levels. Using a single ESM (CNRM-ESM2-2), a set of single ensemble member warming-and-stabilization simulations, they conclude that following a 300-yr of stabilization periods, MHWs depart from their initial transient states and reorganize. Specifically, the frequency increases in the high latitudes of both hemispheres, but declines in the tropics. At very high warming levels, the simulated AMOC strength passes what looks like a threshold where it doesn’t recover over centennial time scales. The study is interesting, suitable for ESD, and add values to better understanding of legacy of anthropogenic climate change once net-zero is achieved at different warming levels and the climate approximately stabilizes. Nevertheless, there are some issues with the current analysis and lack of process-based understanding, which i recommend the authors to address in order to improve the impact of this study. There are also many editorial minor errors, which would benefit from close proof-reading prior to resubmission.
General comments:
The experiments are well thought out and executed. The results with respect to MHWs are clearly presented but there are limited new insights in term of process understanding. For instance, it is not clarified why at the high GWL 4K and above the behaviors is very different in the NH high latitude, whereas in the SH, the frequency and intensity seems to grow almost linearly with GWL. This seems to relate to AMOC. Why do AMOC recovers at low GWL, stabilize at medium GWL and continue to decline at high GWL? Same questions for Fig. 5b: why warming continue at high GWL, even though AMOC decline. Are these two related? Showing Fig. 6 but for 100-500m temperature anomaly at the end of stabilization period could be useful and also potentially disentangle the regional MHW patterns. Are these behaviors unique to CNRM model?
The difference in MHW characteristics across GWLs could be related to the dynamics of heat exchanges in the mixed layer depth. It could be useful to show the air-sea heat fluxes evolution. From Fig. 8, I suggest analyzing heat flux during the winter season of respective hemisphere, linking this to subsurface warming trend (similar to Fig. 5b but for winter NH and SH high latitude).
My other main concern is the use of static baseline during this transient climate experiments. Extreme events in the earlier centuries may not be considered extreme if the baseline has evolved in the latter centuries. This is relevant as the paper is framed around the impact on marine ecosystem, which lifespan is much sorter than centuries. In fact the authors correctly point this out (L205-7). When using the TWL period as the baseline, the MHWs signal may simply represents the long.term subsurface warming (e.g. Fig. 5b). Is this correct?
I recommend the authors to repeat the analysis, but using moving detrended baseline, e.g., 50 years centered around the analyzed periods, and discuss the implications.
There are numerous general statements presented as the results of this study but they are often unclear or not backed up by respective analysis. I have identified some of these below.
Specific comments:
L32: Suggest replacing “Paris Agreement” with “the use of GWL during climate change transition as a valid method to asses impacts”.
L48: ecosystem => ecosystem
L60: consists in => considers
Fig. 1 and similar other Figs: (i) it would be more intuitive too use the same colors for the circle markers; consider using different color (other than yellow-red shadings) for the 02Kpd line.
Fig. 1 caption: ‘which keep running’ => ‘and the model is kept running’
L101: remove ‘and’
L105:’only emissions driven models can capture’ => ‘it allows the model to simulate’
L106 ‘, which is essential to realistically model the’ => ‘in’
L117: ‘ propagated uniformly …. monthly.’ => ‘into monthly and on each model grid cell.’
L126: you have used past tense most of the time => ‘we ended up ….’
L148: An inventory => A list
L156: evaluated => estimated/calculated
L165: Cessation
L166: remove period after t
L208-11: when calculating the MHW events, did you detrend the long-term evolution? If not (which I think you shouyld), does the observed pattern e.g., Fig. 8a simply represents this long-term trend, I.e. faster warming in the summer than winter?
Fig. 2: d-f: W. m2 => W m-2
Fig. 2 caption: Climate climate state => Climate transition
L224: carbon pump led by the land and ocean sinks => carbon uptakes by the land and ocean. Carbon pump has a different meaning and usually applies only for the ocean.
L225: after net-zero => that simulates a net-zero emissions after a period of positive emissions phase.
L233-4: why it is expected? Any references?
L245: negative imbalance for 02Kpd-40, but why the ZEC is positive?
L259: SST declines with ‘lower’ AMOC in the ’n’orthern ‘o’cean
L260: Southern ‘O’cean
L264: ’n’orthern
L265: 3b => 3
Fig. 6,7,S4,S5 caption: the description of the different colors needs to be corrected: e.g. ‘Increase in the mean only are shown in orange’ is incorrect: orange: increase in mean AND decrease in variance.
Fig. 6,S4,S5: decline => depict
L293: consider adding this in Fig. 6, e.g. using stipplings.
L297: ‘joint increases’ mean red bar only, the number 20-67% indicate both red and orange bars.
L303: for the => at the end of; cessation => cessation.
L304-306: can you elaborate more what drive these patterns? Is it related to the way excess heat propagate in the ocean interior with later emergence in the tropics before it upwelled to the surface laters (e.g. Tjiputra et al., 2023; https://www.nature.com/articles/s41598-023-30159-0)? In the Southern Ocean, does the persistent warming condition is due to the similar processes as shown in previous studies after net-zero or negative emissions (e.g. Cassidy et al. 2023; https://iopscience.iop.org/article/10.1088/1748-9326/ad114a; Fremnger et al., 2025; https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2025AV001700)?
Fig. 7: consider replacing this with Fig. S3 or remove Fig. S3c
Fig. 8a. By your definition (L209), shouldn’t the frequency of MHW should be roughly 10% everywhere? Does the shown variability come as a result of using non-detrended time series?
Fig 8,9,S6,S7 captions: line => row; the term absolute’ is not necessary and even confusing due to the negative values. Replace the last word ‘framework’ with period/state/baseline.
Fig. 8c-d: I would expect some more discussions on the interesting seasonal pattern in the high latitudes. Why increased frequency occur first during winter before spreading into summer? Why at NH high latitude at 5KGWL it even reduces during summer in NH?
L325: is ‘to the pre-industrial mean state’ correct? Or relative to the TWL? The term ‘absolute’ can be removed.
L329-331: This statement is not entirely correct. To say both mean and variability increases suggest red colors >40N/S in Figs 6&S5. But both figures also show many of these regions have purple and orange (or even white) colors. Purple: only increase in std, and orange only increase in mean.
L331-5: This requires further discussions on the temporal and spatial trend patterns.
L370-6: Are these simply assumptions based on other studies or have they been observed in your simulations as well? Given that this is a single ESM study, it would be appropriate to show and elaborate how your model is indeed consistent or different from others.
L377-83: there are also limitations of large scale low resolution global model in simulating MHW events that should be mentioned (Pontoppidan et al., 2023; https://link.springer.com/article/10.1007/s00382-023-06758-y) and how do this potentially impact your results?
L395-6: Please show this analysis in the paper.
L400: automn => autumn
L402-3: I don’t understand this. Could you please elaborate?
L407: where crossing => at
L408: elaborate what do you mean by ‘general behavior’?