Review of: The Southern Ocean as the freight train of the global climate under zero-emission scenarios with ACCESS-ESM1.5

Overall assessment:

The paper documents experiments with ACCESS-ESM1.5 that expand upon the standard ZECMIP A-class experiments to better show the transition from negative ZEC at low cumulative emissions to positive ZEC at high cumulative emissions. Additionally that paper uses analysis of ACCESS-ESM1.5 and a slab ocean model to show that in ACCESS-ESM1.5 ZEC is dominated by Southern Ocean processes. While the paper is interesting, generally scientifically sound, and well written – some revisions are needed before publication.

General Comments:

(1)  The authors appear to be using two different algorithms to compute ZEC values. In the caption for Table 2 the authors indicate that they are using the standard algorithm outlined in MacDougall et al. 2020. That is "Values are the differences between 20-year averages centred at the year of the ZEC branch [...], relative to the 20-year average from the respective 1pctCO2 centred at the branch point." While the caption for Figure 2 says the regional ZEC is being computed as "Differences are with respect to the average of the first 10 years of each experiment, and smoothed with a 5-year filter."

I strongly recommend that the authors compute regional ZEC using the same algorithm as global ZEC, using maps of the 20-years average temperature from the 1pctCO2 experiment centred on the year emissions cease as the cessation temperature reference value. Using a different algorithm to compute regional ZEC risks making the results of this study incomparable with other similar studies.

Additionally, a description of the algorithm used to compute ZEC should be included in the methods section.

(2) Section 4.2 "Multi model Comparison" is the least convincing part of the study. From Figure 11 it is clear the using the slab model tuned to  ACCESS-ESM1.5 does not capture MIROC or UKESM temperature trajectories well. While the match to GFLD is better both ESMs use the same ocean model (MOM) so a better match is to be expected. Additionally recent analysis of regional ZEC (MacDougall et al. 2022) showed that for a least some ESMs AMOC is dominating the ZEC response not the Southern Ocean, with some models having regional ZEC dominated by AMOC collapse (CESM2). Thus I suggest the the existing section 4.2 be deleted and a more qualitative comparison be made to the regional ZEC effects shown in MacDougall et al. 2022.

(3) Throughout the manuscript 4 digit model codes for time are used instead of years. Model years in all figures, tables and in text, should be given in standard Arabic numerals (no leading zero) with appropriate units (years). For figure captions please include a x-axis label of 'Model Years', to be clear that Gregorian calendar years are not being used.

(4) Please use consistent notation for the 1% CO2 experiment.

Specific Comments:

Line 1: "Climate Projection Experiment" is an odd start. "Climate model simulations" would be more consistent with past terminology.

Line 5: Delete 'Multi'

Line 16 to 20: Odd framing since ECS becomes a moving target as CO2 concentration will not stabilize for a very long time. Processes not included in ESMs such as the CO2 weathering feedback will cause a slow drawdown to close to pre-industrial over the next few 100,000 years (e.g. Archer 2005). Statement makes more sense after reading the paper but not best framing for an abstract.

Line 24: Delete "safe".

Line 26: Change"greenhouse gas" to 'forcing agent' (aerosols are not gases), and add 'unaccounted for' before 'climate feedbacks'

Line 93: Seems to indicate the ACCESS-ESM1.5 does not have dynamic vegetation. Should say this in model description. Many of the models that participated in ZECMIP did have dynamic vegetation so could be re-written to say this is the implementation in ACCESS-ESM1.5 but other models have prognostically changing vegetation maps.

Line 96: I never liked the description of this as 'unrealistic'. An asteroid strike, global thermonuclear war, or supervolcano  eruption, would probably do the trick. Add 'baring global cataclysm' before 'a global instantaneous', and change 'unrealistic' too 'unlikely' to fix. 

Line 109: Change 'about linear' to 'approximately linear'. Change 'gradient' to 'trend'. A gradient usually indicates a change in space, not time.

Line 110: Tipping points can be subtle transitions.

Line 117: change 'of average' to 'in average'

Figure 1d: The preindustrial TOA energy balance seems to be negative. Is this plotted correctly? If so, is the model drifting?

Line 141 to 149: Could compare results to MacDougall et al. 2022 here.

Line 256: May want to add a few sentences to discuss the potential impact of Ice Sheet loss on the Southern Ocean Feedbacks. I don't think ACCESS-ESM1.5 has a dynamic Ice Sheet model, so discussing what impact Ice Sheet loss may have is important.

Line 319: Spell out ESGF the first time you use it.

Line 341 to 342: Need to note that this is 2 of 4, not 2 of 9.

Line 343: Change 'about neutral' to 'approximately zero'

References:

Archer D. Fate of fossil fuel CO2 in geologic time. Journal of geophysical research: Oceans. 2005 Sep;110(C9).

MacDougall AH, Frölicher TL, Jones CD, Rogelj J, Matthews HD, Zickfeld K, Arora VK, Barrett NJ, Brovkin V, Burger FA, Eby M. Is there warming in the pipeline? A multi-model analysis of the Zero Emissions Commitment from CO 2. Biogeosciences. 2020 Jun 15;17(11):2987-3016.

Chamberlain et al. present a set of idealized Earth system modelling experiments exploring the response of the Earth system to phasing out CO2 emissions. The experiments follow the "Zero Emission Commitment MIP" protocol but additional simulations (additional levels of emissions) are provided. The authors construct a simple slab (restoring) model where temperature is restored towards the equilibrium temperature of the model with two different time scales. They calibrate this model to their ESM and also compare results from three other ESMs with this slab model. The main finding that the authors emphasize is that the Southern Ocean continues to warm on centennial time scales after emissions ceased.

The the response of the Earth system to phasing out emissions and the committed climate change due to prior emissions is a highly relevant research area, and there is generally a lack of ESM simulations that explore such scenarios. The model experiments presented here are therefore highly relevant and the model simulations are interesting and well designed. However, the manuscript reads in large parts like a technical summary of simulation results rather than focusing on new insights. Also, it remains unclear what new insights the two-slab model brings, particularly when it comes to the multi model comparison (see below for more details). Finally, the authors make no attempt to place their study in the context of previous literature, neither in the introduction nor in the discussion/conclusions section. Given these concerns, I would suggest substantial revisions of the manuscript before it might be suitable for publication in Biogeosciences. I cannot address all these points in depth in this review, but I will give some suggestions below.

Major points:

1) The abstract and the introduction contain too many technical details. For example, the abstract describes the ZECMIP simulation design (emission levels and the fact that the zero emission simulations are branched from the 1pctCO2 simulation), but is missing a summary of main results. The introduction is missing an account of previous literature (see below). In the results section, zonal mean section of salinity and oxygen are presented, but these results are never used or discussed (there are a few general sentences on biogeochemical changes in the conclusions section). It remains unclear what insights we gain from these figures, and how this relates to the main topic (the committed warming) of the paper. In general, the manuscript does not have a clear direction (at least it was not obvious to me). What do the authors want to address? Is it the fact that the ZEC750 simulation cools while the ZEC2000 simulation warms in their model? Is it a comparison with other models? Why do some models warm while others cool? What is the role of the Southern Ocean in this? It would help to formulate one or two clear research questions, and really set out to answer these.

2) Related to point 1, it also remains unclear what insights we gain from the application of the 2-slab model. I can see that the model is able to reproduces the global average surface temperature of the ZEC branches of the ACCESS model (no warming for ZEC750, increasingly more committed warming for higher emissions), but what does this mean physically? When it comes to the multi-model inter-comparison, the slab-model (as the authors present it) is not able to reproduce the cooling characteristics of the MIROC model, and the fit is not very good for the GFDL and UKESM models either. So again, what can we learn from this model then? As a side note, I believe that if there is something to learn from the slab-model, the model "ocean" time-scale would need to be fitted to the individual models (the authors only adjusted ECS). Would this improve the slab-model results? Would the slab model be able to reproduce the cooling in MIROC? As the results stand now, the slab-model seems to be able to reproduce the ACCESS ZEC-simulations more or less by chance, and it fails to reproduce relevant aspects of the zero emission commitment for the other models.

3) This study is not the first to investigate the response of the Earth system to phasing out emissions, but neither the introduction nor the discussion/conclusions sections place the present manuscript in the context of previous literature. The first (to my knowledge) ZEC study with a full ESM was by Gillett et al. (2011), who also emphasize changes in the Southern Ocean. The study by Frölicher et al. (2014) finds a pronounced multi-centennial warming in one model, while a second model shows a cooling trend. Recently, Schwinger et al. (2022) have conducted a study, which also was based on the ZECMIP protocol, and they find a dominant role of AMOC decline and recovery for ZEC in their model. These studies come to my mind immediately, but there are probably more.

Other points:

Not sure if it is because I am not a native speaker, but I find the title of the manuscript not very intuitive to understand. I would encourage the authors to think about an alternative title.

The use of year 101 as start year for the simulations is confusing. I would suggest to set the nominal start year at year 1 in all tables and figures.

The abbreviations of the simulations is too similar to the abbreviation of ZEC values. For example, the authors use ZEC_200  for the temperature change after 200 years into the ZEC-simulations, and ZEC750 to denote the ZEC simulation with 750 PgC emissions. I would suggest to use a different abbreviation for the simulations.

Either in Section 3.2 or 3.3.1, it would be interesting to read something about the role of AMOC changes, which has been identified to play a dominant role in models with strong AMOC decline (Schwinger et al. 2022). Including AMOC strength in Fig 1 could be useful. In Fig 3b it looks like a signal of AMOC decline would be visible in the North Atlantic in the ZEC750 simulation?

Section 3.3.1: What is shown in the figures is the global meridional streamfunction not "the overturning". Overturning strength can be visualized through and calculated from the streamfunction. Please correct throughout the manuscript.

Section 4.1: At least the main idea of the slab-model should be described in the main text, such that the reader can understand what the model is intended to do. Might be even easier to move equation A1 into the main text.

line 22: I suggest to delete "of the global climate"

line 24: "...the potential budget of carbon emissions permitable without exceeding any agreed thresholds of “safe” warming" is very complicated. "remaining carbon budget" has become an established term for this and could be used here.

line 26: "carbon emission budget" -> "carbon budget"

line 31: "This conclusion..." this is not a conclusion, it is an assumption.

line 37: The acronym 1pctCO2 has not been introduced

line 35-41: Much of this paragraph could and should be moved to the methods section.

line 70-75: The main issue with trends and biases for the kind of simulations presented here, is the switch from concentration to emission driven configuration. This could be made clear instead of the generic statement in the last sentence of this paragraph.

line 77-86: The fact that the simulations presented here were run on different computer hardware is a technical detail that is not relevant for the results. This can be a footnote in Table 1 explaining why numerical values are slightly different from previously published results.

lines 88-98: This general description of the ZECMIP experiments/protocoll should come earlier.

lines 100-106: Maybe a personal preference, but I dont think it is necesarry to provide a summary of subsections at the start of a new section. Good descriptive titles of subsections are enough.

line 109: "...gradient increase evenly" -> "... rate of surface air temperature change" or similar.

line 110: The numerical values presented here seem to contradict the values in Table 1? Please clarify.

line 114-115: This sentence doesn't make sense, please consider rephrasing it.

line 130: Unclear what does "the extension of this experiment refer to"? Please clarify.

line 227: TCR is defined at year 70 (not 50) of the 1pctCO2 simulation (at doubling of atmospheric CO2).

line 238-239: "... more than adequate" I don't think this statement is adequate. The simple model can reproduce certain aspects of the result.

line 256-264: It remains unclear to me what the authors intend to say with this paragraph on tipping points. Please restructure/reword/clarify.

line 283-286: There is no "contrast" here this just the different timescale (as the authors note). Please reword these sentences.

References:

Frölicher, T., Winton, M. & Sarmiento, J. Continued global warming after CO2 emissions stoppage. Nature Clim Change 4, 40–44 (2014). https://doi.org/10.1038/nclimate2060

Gillett, N., Arora, V., Zickfeld, K. et al. Ongoing climate change following a complete cessation of carbon dioxide emissions. Nature Geosci 4, 83–87 (2011). https://doi.org/10.1038/ngeo1047

Schwinger, J., Asaadi, A., Goris, N. et al. Possibility for strong northern hemisphere high-latitude cooling under negative emissions. Nat Commun 13, 1095 (2022). https://doi.org/10.1038/s41467-022-28573-5