General comments.

This paper explores mean state and jet variability changes in response to increased CO2 and mid-Pliocene boundary conditions through previously published general circulation model experiments using the CESM. The experiments are well-designed, and some interesting, although perhaps not too surprising, results are found, that changes in climate variability can be very different in response to increased CO2 relative to other boundary condition forcings that also create warmer climates. Overall, the structure of the manuscript is logical, and figures are clear. However, I find much of the results section to be very descriptive of the figures, with little interpretation, leaving the reader to try to find their own interpretation - more interpretation of the results (and less description) would greatly improve this manuscript. The discussion section leans heavily on prior literature and could do a much better job of putting the new results in the context of literature and highlighting the novel results in this study and their relevance. Including uncertainty estimates on some of the quantities reported would also improve the paper. 

The research is motivated by the question of whether “the mid-Pliocene climate be used to assess the response of present-day Northern Hemisphere winter atmospheric variability, such as the NAO, NAM and PNA, to increased CO2“; however, I remain a little unclear as to why this is a valuable question to ask? If we trust the model in reproducing climate variability of the mid-Pliocene, why don’t we just trust the models for the future? One argument for using paleo data is that we have proxy observations and so don’t need to rely only on models, but here you are just using a model, so any model biases remain. Exploring and understanding climate variability in different climates is interesting and useful for understanding the underlying climate system and behaviour of internal variability, and your result that variability is very different in two different warm climates is interesting. These insights could be helpful, in a more indirect way, in helping understand projected future changes. I think the research focus is interesting and worthy of publication, but the motivation does not convince me as it is currently presented in the introduction, and I found the structure and order of the introduction rather confusing.

For example, the paragraph starting on line 42: “An issue with investigating the response of climate variability to increasing CO2 in the near-future is that the present-day climate system is not in equilibrium with the mean forcing….” – are you saying that the present day isn’t an analog for the future because we’re not in equilibrium? Or that, for model simulations in which the climate is changing, it is difficult to assess what is trend and what is a change in variability? I would argue that, at least in the near-future, the climate won’t be in equilibrium, so the modelled simulations of future climate seem like a better analog than a past climate that is in equilibrium? Even if you argue that you can only study this in an equilibrium climate, why not use simulations of future changes that have reached equilibrium, e.g. 4xCO2?

Lastly, the title of the paper suggests this is a response that is consistent across many past warm climate conditions, rather than just the mid-Pliocene – given the suggested dependence on orography, and possible dependence on SSTs, this may not be true;  I suggest to be more specific in the title so it's not potentially misleading. 

Specific comments

‘the geological past climate was in equilibrium with forcing’ – if this was always true, the climate couldn’t have changed in the past? You could argue that it was, most of the time, more in equilibrium than we are today.

You mention variability on decadal timescales, but not the PDO (Pacific Decadal Oscillation), which surprised me – is there a reason to not discuss the PDO in modes of variability of the North Pacific? Particularly given, on line 175, you say you are mainly interested in interannual and decadal variability.

Line 173. What re-analysis data do you use? Do you have enough years to use a 50-year window?

Line 175 ‘A window size of 50 years was chosen since we are mainly interested in interannual and decadal winter variability’ what degree did you use for your Lowess smoothing, and aren’t you removing all of the interannual variability and most of the decadal variability by using a 50 year window?

Line 185: how do you determine level of zonality? Just subjectively by looking at the EOFs or some objective method?

Figure 1 caption. I think d) should be NPac-z not NPac-a?

Figure 1. It’s a little confusing whether the contours are the absolute CR20 values, or the differences between E280 and CR20. I think perhaps it is differences for a. and b. and then absolute values for the others, but the caption does not make this clear.

Line 198. MAX and MIN would seem more related than PLUS and MIN

Section 2.3. The description of the re-analysis dataset would be better earlier, before you mention that you’ve used re-analysis data in line 173. The way you have described the re-analysis data is confusing as ‘we use assimilated sea-level pressure data from the NOAA 20C reanalysis….’ – this almost sounds like you’ve done some post-processing to the re-analysis data, or that you’re using the data that was assimilated into the re-analysis. ‘The data runs from 1836 to 2015 and is assimilated using surface pressure observations on a 1.0◦ latitude x 1.0◦ longitude grid’ This sounds like the surface pressure observations are on the 1x1 degree grid, and assimilated the re-analysis, rather than the re-analysis data assimilating the surface pressure observations.  I think it is worth mentioning that a benefit of the CR20 for these longer time periods is that there is more consistency in the data that is assimilated than for, for example, ERA5, which includes satellite data in more recent periods. That said, the amount of data being assimilated certainly does change with tie in CR20.

Section 2.3. Why do you use all of the data in the re-analysis, through to 2015, when your simulation is just pre-industrial, as you mention? Do you de-trend the re-analysis data to try to take out any climate change signal in MSLP? Would choosing a shorter time period that is closer to your pre-industrial modelled dataset not be better?

Line 209. Do you mean spatial mean MSLP? Is the model bias in global mean sea level pressure particularly meaningful? Seems reasonable this is relatively small since sea level pressure is essentially a measure of the amount of atmosphere above a point, so the global mean MSLP is broadly a measure of how much atmosphere there is in your model?

Figure 2. It might be interesting to show Eoi400-Eoi280 as an addition panel in fig 2. I realize this isn’t a doubling, and so isn’t equivalent to panel b, but it would be interesting to see if the pattern response to increased CO2 is similar in the Pliocene (or you could scale the responses, e.g. to 1 degree of global warming)? This would also help with seeing your result described on lines 253-254, as it seems this result (that the MSLP difference are predominantly caused by the different surface boundary conditions, not the different CO2 levels) is quite a key one, given that one of your main conclusions is about the change in variability in response to these two forcings. However, from the panels presented in Fig. 2 and Fig. S2, it is not easy to understand what comparison to make to come to this conclusion. I think adding Eoi400-Eoi280 to Fig 2 would help.

Figure 2. Is any of the signal over Greenland in panel b likely to be because of differences in the height of the ice sheet, and how the interpolation to sea level is performed?

Fig. 3. In understanding the sea ice changes, it seems there are significant changes in land-ocean boundaries (as shown in the coastlines in panel c, which is different to that for present day) but this doesn’t seem to be discussed. For example, there looks like there is large sea ice retreat over Hudson’s bay, but this looks to be more related to shifting coastlines? Mentioning this would be useful.

Section 3.1. The results are useful for putting the variability differences in context of mean climate changes, but this isn’t mentioned explicitly in the text, so the reader is left to add this interpretation themselves. Also, the section is very descriptive, with little attempt to provide mechanistic or physical explanations for the differences you see. For example, can you give some suggestions as to why you see such a strong response in the upper level circulation in response to Pliocene surface boundary conditions? Do you think this is related to differences in topography? Or tropical SSTs? Or something else?

Fig. 5. Having the positive contours dashed and the negative dot-dashed is a rather subtle difference (although after looking at Fig. 6 I now realize they are also coloured – stating this in the caption would help). I would find it easier to interpret the figures is positive was solid contours and negative dashed (and zero bold if you wanted to distinguish the zero contour). At least mention whatever convention you use in the caption to make it easy to get the information.

Changes in variance explained – I think you should calculate uncertainties on these values (e.g. using bootstrap resampling), to get a sense of how robust the differences you are reporting are. With 200 months I would hypothesise that changes of 10% are robust, but it depends on low frequency variability in the model. But are differences of 63.9 vs 66.7, or 17.5 vs 20.3 robust, or just noise, I’m unsure.

Section 3.2.3. I am not sure what to conclude from the section – more context is required. I also wonder whether the correlation between the NHem leading mode and the zonal or azonal modes is sensitive to which mode is picked up as the leading mode in Nhem and therefore how meaningful it is in a physical sense? Is the second leading mode of Nhem in Eoi280 more similar to the leading mode in E560 and E280, and therefore would correlated better with NPac-a?

Line 378. Could you provide more details for why you conclude that the more northern jet state is stronger? NPac-z seems to correspond more to a meridional shift of the jet strength (as implied in line 379, and from Linkin and Nigam 2008). You do mention in line 228 that NPac-z is linked to meridional ‘modulation’ of the Asian-Pacific jet, but it’s a little unclear what this means (meridional shifts? Changes in jet strength?) and the reader is left to work out a lot of these details themselves.

Line 381. “the correlation between the two NPac modes” to me implies that the 2 modes are correlated to each other (which they can’t be, being EOFs), rather than (on second reading) one is correlated with the jet intensity and one with jet latitude.

Line 414. I assumed ‘we want to explore’ implied that you were going to analyse some more simulations with different boundary conditions, but it seems more just literature review.

Line 415-416. What about sea surface temperature changes? Surely these could impact the mean state, and thus potentially variability as well? Tropical SSTs in particular can have a large impact on extratropical atmospheric dynamics (e.g. https://journals.ametsoc.org/view/journals/clim/14/4/1520-0442_2001_014_0565_tiotss_2.0.co_2.xml ). Indeed, in Chandan and Peltier (2018) they note that their model does not agree with some paleoproxies of SSTs in some tropical locations. Does your model have the same bias? This is a limitation that should be discussed in this paper.

In most of your discussion I find it hard to understand which is new information and understanding produced by this paper, and which is taken from the literature – it seems like a lot of discussion of the literature, and perhaps some of this belongs in the introduction?

Line 499. I’m not sure you really address the non-additivity here, you just discuss the literature saying that the responses might be non-additive? It seems like perhaps you don’t have all the experiments you would need to address non-additivity: you’d need E400 to compare the differences to CO2 forcing (E280 vs E400), boundary condition forcing (E280 vs EOI280) relative to both together: E280 vs EOI400).  And how are you defining nonlinearity and nonadditivity distinctly?

Review of: Past warm climate conditions show a shift in Northern Hemisphere winter variability towards a dominant North Pacific Oscillation (Oldeman et al., 2023)

Authors: Arthur M. Oldeman, Michiel L.J. Baatsen, Anna S. von der Heydt, Aarnout J. van Delden, & Henk A. Dijkstra

In this study the authors pose the question of whether the mid-Pliocene can act as a suitable analog for future climate change scenarios given the relatively similar CO₂ concentrations and geography to present day. In particular they focus on suitability for studies of NH winter variability. They use a suite of simulations from CESM1.0.5, part of PlioMIP2, to separate the climate response to CO₂ doubling vs. mid-Pliocene boundary conditions. A major finding of this manuscript is that the dominant mode of North Pacific variability differs depending on whether we consider CO₂ doubling or mid-Pliocene boundary conditions: the preindustrial dominant mode (the PNA) strengthens under climate change whereas under mid-Pliocene boundary conditions the NPO becomes the dominant mode. The authors conclude by saying that the mid-Pliocene is not a suitable analog for future climate change.

I thank the authors for an interesting study. While the result that different boundary conditions lead to different climate variability is not necessarily shocking, it’s valuable to explain that the North Pacific variability in particular changes depending on the forcing. I have some questions and suggestions that I hope will be addressed before publication. I hope the authors find the commends below constructive and useful and I look forward to reading their responses.

Overarching Comments

1. General Motivation: Overall I found the general motivation of this paper to be somewhat unclear. After requisite background on NH wintertime variability and why the mid-Pliocene is considered a potential analog for future climate change, the paragraph beginning at L61 seems to lay out the main question of this study well. The authors then proceed to cite studies which say the two periods are not good analogs for each other. My question, then, is where does the current study fit in? Do prior studies not discuss wintertime NH variability between the periods? Does this study start from the hypothesis that the two periods will be not agree well but the authors seek to quantify that hypothesis? As you clarify this, the discussion at L417 (and throughout Sec 4.1) might become more suited for the introduction.
2. Results Section: As it stands, Section 3 reads as if the authors made a list of plots and then simply describe them to the reader. Instead of this, I recommend the authors consider their primary motivation (see above) and then present their results in a way that strengthens their argument and presents a coherent storyline rather than just marching through mean SLP, SST, and U200, then std SLP, SST, and U200, and then figures about the jet. For instance, if the hypothesis is that the two eras won’t agree well, weave in interpretation of what you’re seeing in the difference between the b and c panels of Figs 2-4. Explain why differences in the EOFs lead you to consider the jet, how the differences in surface forcing impact the jet and weather, etc. Interpretation is not inappropriate in the Results section and will lead to a stronger article.
3. Section 4.2: I had a hard time understanding which dynamical interpretations were from past studies and which were from the figures in this paper. I especially was confused at the “three major aspects” section. For instance, atmospheric heat transport was not discussed elsewhere in this paper but makes an appearance here. I recommend the authors clearly outline where these dynamical interpretations come from their own results and where they are pulling from prior studies. It seems like a lot of this section results from prior studies, which makes me wonder how much of it should be in the introduction.
4. Title: I have two thoughts about the title, the second of which relates to Overarching Comment #1. First, I think “warm climate conditions” is too general; the authors only focus on the mid-Pliocene. Second, the title doesn’t really seem to capture the main point of the paper. Although the increased dominance of the NPO over the PNA is an interesting result, it seems to me that the main point of this paper is that the mid-Pliocene is a bad analog for future climate change when studying NH winter variability. The authors say this specifically in L13-14 of the abstract.
5. Zonality & Azonality: At L185 the concepts of zonality and azonality are introduced and for the rest of the paper I proceeded to get confused about which was which. To begin, I don’t think that “azonal” is a common word—for instance, there is no definition in the AMS glossary. I think this a good first check on whether a term needs to be defined. How do the authors define azonal specifically? Presumably they mean something different than “meridional,” correct? How do they decide which mode is zonal and which is azonal? Eyeballing? Second, when I think of the NAO and the NPO, I generally think of them as meridional modes of variability with a center of action in the subtropics and then the other center of action directly above the first in the subpolar region. The authors however say “The NAO is essentially the zonal mode in the NAtl” in L185-186. The PNA, which I think of as a zonal mode with a center of action over the Aleutian Low and one directly eastward over North America, is then identified as the azonal mode. Please explain your reasoning for these descriptions as they seem to be the opposite of my intuition. Last, I found the NPac-z/NPac-a/NAtl-z/NAtl-a terminology confusing, partially because of the confusion around zonal and azonal mentioned before. Why not just say the NPO mode, the PNA mode, the NAO mode, and the EA modes? In the titles of the figures, you say e.g. “NAtl-z (NAO)”, so why not just cut out the strange labels and make it so the readers have four less things to keep track of?
6. Simulation Permutations: It might be informative to show Eoi400-Eoi280 differences. While this isn’t precisely the equivalent for comparison with E560-E280 since it’s not a doubling, this difference can help us understand whether increased CO₂ with mid-Pliocene BCs is similar or different in pattern. This might help with interpretation in the discussion, though it doesn’t seem like you have enough simulations to explore the full space of nonlinearity. I think E400 would be needed for that.

Substantive Comments

1. L19-21: The statement that future climate projections fail to give consistent responses to greenhouse forcing among NH winter variability modes merits a citation.
2. L77-87: This paragraph seems out of place. You don’t really touch on proxy reconstructions in this paper, just modeled results.
3. 1.3: I’m confused about what data was previously available and which simulations the authors have run for this study. Some clarification is needed. Have you run these simulations of E560 yourselves? L125 suggests you’re using this model specifically because other sensitivity experiments are available.
4. L181 and throughout: NHem is not the universally accepted abbreviation for Northern Hemisphere. I very strongly recommend changing it to the common “NH.”
5. L209: “The mean MSLP difference is very small…” —> are you talking about some area mean value?
6. Fig 1, 5, 6: Dashed vs. dash-dotted lines are difficult to discern. I recommend solid for positive, dashed for negative, and thicker line for 0.
7. L218: You already defined your acronyms in L181-182. You then redefine them again in L295.
8. Fig 2, 3, 4, 6 Captions: Write “minus” or use a minus sign rather than “min” as it currently stands. It’s only two more letters and you use “min” at other points to mean minimum.
9. L263: An arctic amplification citation would be useful here.
10. L283: “a lot stronger” — this seems arbitrary. I recommend quantifying how much stronger. Similarly, you use “a lot stronger” again in L308 and “a lot weaker” in L513.
11. L284: “significantly” -- do you mean statistically significantly or just substantially weaker?
12. Fig 8: It seems like we’re missing some panels here. Unless I’m misreading this figure, you don’t show the correlation of jet intensity with NPac-z or jet latitude with NPac-a. Why is that? I recommend adding those panels. Also, it would be worth mentioning somewhere that the Eoi280 scatter in Fig 8d is nonlinear so the linear correlation might not be the best metric.
13. L403-405: Citation for split jet and wave breaking?
14. L505-506: Different simulation permutations may help with disentangling the effects of BCs vs. CO₂. See Overarching Comment #6.
15. The capitalizations throughout the citations seem to be somewhat arbitrary and using different citation styles. I suggest standardizing to one citation style.

Minor Typos

-L1: “…we address the question OF whether…”

-L16-17: “…there is a need to make…” -- avoid the passive voice as able

-L106: “Next to that…” -- doesn’t make sense in this sentence

-L114: The community seems to refer to it as “CESM” not “The CESM” when using the acronym

-L117: Why “therefore”? This doesn’t make sense to me.

-L201: “It is not be a one-on-one comparison” -- incorrect wording

-L368: “as well between both azonal modes” -- strange wording

-L381: UK vs. US English spelling is interspersed throughout, e.g., “behavior” here, “behaviour” in L172, 364, 464. I recommend sticking to one or the other.

-L440: Is STJ subtropical jet? This acronym was not introduced before this line.

-L525: “we posed the question OF whether”

-L527: Typo-- “analogous”

-L529: “WHO state”