General comments

 This article applies the recently developed jet tracking method of Perez et al. to the PAMIP simulations, in order to assess how Arctic sea-ice loss will impact daily and seasonal timescale jet morphology in boreal winter. The basic message of the paper is that there is basically no impact.

The paper is mostly clearly written, well structured and easy to read. It is quite a quick and easy paper to read, owing to the fact that it’s a “null result” paper, which means there just isn’t that much to say. I think the paper adds a useful contribution to the Arctic-ice-loss literature by examining more nuanced jet morphology on daily timescales. The null-result adds more weight to the idea that ice loss won’t notably affect the jet, by showing that important changes aren’t being concealed by large-scale zonal means that become visible in things like jet tilt etc (with the usual signal-to-noise paradox caveat applying).

I have two major comments and then a few minor comments. I hope the authors can address these without undue burden.

Best wishes, Kristian Strommen

Major comments

L180-185: I’m very confused about the two ZWRI indices being compared, and why they’re not the same. Maybe I am being dense, but the authors don’t provide enough information in the text to be completely sure what exactly is being computed. My interpretation was that the “Smith et al version” was to take zonal averages (i.e., average across longitudes 0-60W), then latitudinal averages over the two latitude regions (54-63N and 30-39N), and then taking the difference of the two regions. My interpretation of your “North Atlantic average version” was that you first take latitudinal averages across those two regions, look at the difference at each longitude 0-60W, and then take the average of these differences. However, these two versions give the same number at the end, because everything is linear; the order in which latitudes and longitudes have been averaged has simply been switched. So this can’t be a correct interpretation, as you show that the two ZWRI versions are different. Can you please clarify exactly what is being computed here and why the two versions differ exactly? Please clarify the text accordingly.

L344-351: I feel some extra care is required with this paragraph.



Firstly, your measure of variability is just the standard deviation. Observing no change in the standard deviation does not mean that important things haven’t changed about the jet variability, because there can be compensation between the types of jet variability. For example, there could be changes to the persistence of certain jet configurations. If the jet overall spans the same latitudes as it did before ice loss, but the persistence of certain jet configurations have increased at the expense of others, then one can easily end up observing no change in standard deviation. However, it is clearly of interest to extremes to know if some jet configurations have become more persistent or not. It has in fact been shown that CMIP6 models project a decrease in the persistence of certain jet configurations (https://doi.org/10.1029/2022GL100811), so this point isn’t just academic. Your framework doesn’t easily allow you to assess such changes, because it would require you to identify specific configurations of interest, i.e., “weather regimes” of some sort. I note that the ability to assess changes to persistence and occurrence separately is a nice advantage of a “regime” approach, and can make sense to do even if you don’t believe the specific regimes are intrinsic to the atmosphere. I don’t expect the authors to compute persistence statistics, but this caveat should be explained.

Secondly, the authors assert that extreme weather is associated with a high-amplitude wavy jet stream. This is only true if you have decided up front what extreme weather you care about! A very strong zonal jet pointing right at the UK during winter can often cause flooding, because all the storms dump their rain on the UK. This is not a wavy jet stream, but it is certainly extreme weather! There is a general theme in a lot of literature on Arctic-ice-loss-may-or-may-not-cause-extreme-weather to never actually specify which extreme weather events one cares about and just assume it’s all wavy. This may or may not be fine when restricting to summer (I don’t know), but for winter it seems like an implicitly biased definition of “extreme weather”. The authors should clarify the text to emphasise that you are only talking about a specific subset of extreme European winter weather.

Thirdly, you point out an increase in split jet days in 3/6 models. However, you argue that this change doesn’t matter because (a) the number of split days is small to begin with and (b) the change is too small to affect the overall distribution of jet latitude etc. But if one is interested in extreme weather, then this dismissal doesn’t seem reasonable. Extremes are also only a small percentage of all events, but we still care about them. Changes in frequencies of extremes will also necessarily be small, but again, we care about them. Furthermore, in summer, split jets have been linked to certain kinds of extremes (e.g. https://doi.org/10.1038/s41467-022-31432-y). Have the authors looked at all at what kind of winter extremes could be associated with split jet days? Or does there exist prior literature on this? If not, then it seems there is an important caveat to note here.

Minor comments

L98: ERA5 has been regridded to 2.81 degrees. What effect does this have on the diagnosed jet variability compared to ERA5 at 1 degrees? I am unsure how much of the trimodal JLI pdf is maintained when computing JLI using 2.81 degree zonal winds. The troughs in the pdf are order 4-5 degrees across when computed using 1 degree winds, so it’s easy to imagine a lot of this structure vanishing when using 2.81 degree winds. I would guess the impact on the jet morphology method are small (since everything is unimodal there) but it would be good to check and comment, perhaps including a supplementary figure.

Figure 3: The use of the word “objects” is a bit strange. Maybe “gridpoints” would be better?

Figure 4: I strongly recommend giving (a) and (b) the same y-axis. It will make it easier to parse the key qualitative information for the reader.

L302-303: About the IPSL model: is the p-value for change in jet latitude close to 0.05? If yes, that could indicate that the non-significance for jet latitude vs significance for U850 changes is partly just noise as well.

L314: There is a missing space.

L353: The paper https://doi.org/10.5194/wcd-3-951-2022 also argues that ice-ocean-atmosphere coupling may be important to simulate Arctic-midlatitude links and that such coupling may be missing in most models. I hope the authors will not consider it grossly inappropriate of me to suggest citing it, given its relevance here.

This study investigates the effect of future sea ice changes on the low-level jet. They therefore analyse 6 PAMIP models and quantify changes in the daily jet stream morphology using a new 2-dimensional feature-based approach based on Perez et al. (2024). This allows for not only quantifying jet strength and latitude but also jet tilt, split jet, and no jet days. Overall, the response of the jet stream to sea ice changes is small, with very few significant changes, which is mostly in accordance with earlier studies.

The paper is well written, the methods are sound and I enjoyed reading it. It is also a rather quick read, which is related to the fact that there are few significant or unexpected results. However, I still think it is a valuable contribution to further investigating the PAMIP experiments and the small influence of future Arctic sea ice loss on the jet stream within these experiments. I appreciate the use of daily data, but in my opinion, the potential of this approach was not fully realised (see major comment 2). I also have a concern about the methodology which is related to the coarse resolution (see major comment 1). Most of the minor comments are rather trivial. I hope that it is not too much trouble to address the major comments in a reply or the manuscript.

Major comments:

• Did you test how the regridding of all your model data onto the common coarse 2.81° x 2.81° grid affects the results? I fear that you lose quite a bit of information when you regrid to such a coarse resolution, in particular as the detected signals are rather small (e.g., 0.8 ± 0.1°). It would be nice to test this for one of the models with a higher resolution (or for ERA5). It might not play a large role in a climatological mean sense, but it would be good to test this.
• Different studies have shown that timing is important when investigating the effects of AA / sea ice loss on circulation changes (e.g., Siew et al, 2020; Crasemann et al., 2017): Sea ice loss has been connected to a Scandinavian/Ural blocking type circulation anomalies in December / early winter and a NAO- pattern via a stratospheric pathway in February / late winter. Both circulation patterns are related to distinct changes in the North Atlantic jet stream. Although you use daily data your results are only shown as winter (DJF) mean. Did you investigate the changes in jet morphology on a monthly scale and do you see any differences when you do so?

Minor comments:

L16: mean jet shift of -0.8 ± 0.1°, the negative sign is usually associated with a southward shift. However, it might make sense to actually write southward or equatorward” instead.

L30/31:  Not sea ice but AA weakens the meridional temperature gradient. Might make sense to change the order there.

L61-62: I found this a bit confusing as Section 3 starts with the effect of Arctic sea ice loss on winter mean circulation

Fig. 3: The contours are somewhat unclear as they don’t “close”. Maybe it would help to add numbers?

Fig. 3: Maybe this figure could be skipped altogether as it is basically a reproduction of Fig. S1 from Ye et al. (2023), except for the AWI model.

L174: I agree that this is probably due to the smaller ensemble size, but you can’t be 100% sure about this. Maybe you can slightly rephrase this.

Fig. 4: I would suggest using the same scale for a) and b)

Fig. 6,7,8: The text is somewhat small

L237-238: I was confused about where to find the information on the interannual variability, as it’s the only number not given within the figure. Nevertheless, it partly receives more attention than the other metrics. Hence, would it make sense to add this information to the figures?

L246-250: Isn’t IPSL also an exception? At least the daily variation is given in bold font.

Fig. 8: It might make sense to scale the ensemble mean and standard deviation of the jet mass by 1e14 as you do on the x-axis. At the moment the values are not helpful as one cannot identify differences.

L314: missing space between “object” and “when”

Discussion and Conclusion: I find the discussion and conclusion somewhat repetitive.  Maybe merging them could help to reduce some of the repetition.

L330-L339: I found these lines somewhat confusing. What is the main message of this paragraph?

Tables S1-S4: It would be nice to also have the information on significance, as given by bold font in the figures of the main text.

References:

Crasemann, B., Handorf, D., Jaiser, R., Dethloff, K., Nakamura, T., Ukita, J., & Yamazaki, K. (2017). Can preferred atmospheric circulation patterns over the North-Atlantic-Eurasian region be associated with arctic sea ice loss?. Polar Science, 14, 9-20. https://doi.org/10.1016/j.polar.2017.09.002

Perez, J., Maycock, A. C., Griffiths, S. D., Hardiman, S. C., and McKenna, C. M. (2024). A new characterisation of the North Atlantic eddy-driven jet using two-dimensional moment analysis, Weather Clim. Dynam., 5, 1061–1078, https://doi.org/10.5194/wcd-5-1061-2024

Siew, P. Y. F., Li, C., Sobolowski, S. P., & King, M. P. (2020). Intermittency of Arctic–mid-latitude teleconnections: stratospheric pathway between autumn sea ice and the winter North Atlantic Oscillation. Weather and Climate Dynamics, 1(1), 261-275. https://doi.org/10.5194/wcd-1-261-2020

Ye, K., Woollings, T., and Screen, J. A. (2023). European Winter Climate Response to Projected Arctic Sea-Ice Loss Strongly Shaped by Change in the North Atlantic Jet, Geophys. Res. Lett., 50, e2022GL102005, https://doi.org/10.1029/2022GL102005

This study investigates how Arctic sea ice loss impacts the North Atlantic jet stream using simulations from PAMIP. While previous studies have investigated changes in jet stream characteristics, this is typically done with zonal average wind speed and seasonal averages. This study examines the changes in jet morphology from daily data using a 2-dimensional feature-based method. The authors find that the jet shifts poleward in response to sea ice loss, but the jet speed, jet tilt, jet mass and daily variability of the jet features show little change. The authors conclude that future sea ice loss is unlikely to cause a significant weakening or increase in variability of the North Atlantic jet.

Overall, I thought this study was well done, and it could be an important contribution to the literature. The research questions are reasonable and well-motivated.  The results are presented clearly (for the most part), and the conclusions are convincing. I appreciate the fact that the authors have chosen to publish and highlight the ‘negative’ results instead of searching to try to find ‘positive’ results to highlight (or not publish at all). I recommend publication after these minor comments are addressed:

- One caveat that should be mentioned somewhere is that sea ice loss is just one potential driver of the midlatitude circulation and associated impacts. Just because there is no change in response to sea ice loss does not mean that there will be no change in response to global warming/increased CO2. It is likely that the authors understand this, but confusing the response to sea ice loss with the response to global warming is a common mistake/misunderstanding I see, so it might be good to be explicit about this.

-L16: If I am understanding correctly (based on the description of this later at L207-209), the value for the multi-model mean shift here is the value for only the three models that show a statistically significant shift. This seems a bit misleading and somewhat cherry-picking. Isn’t the more relevant value the multimodel mean over all models?

-L45:  Blackport and Screen (2020) could be cited here if the authors think it is useful. We showed little change in waviness in observed trends, trends in historical simulations, and in response to sea ice loss/Arctic amplification in targeted model experiments.

-L47: Blackport et al. (2019) is not the most appropriate reference here. This study was primarily questioning the causality of the statistical relationship. Blackport and Screen (2021) was mostly about this as well, but we also did show some weakening of the relationship as well, so it is more appropriate.

-L50-59: A study that is missing in this discussion is Ye et al. (2024) who looked at daily variability of the North Atlantic jet in response to sea ice loss. They were still using the JLI index from Woollings et al. (2010), and they were only using one model, so the results presented here are still novel, but the study should be mentioned.

-L85: A caveat that should be mentioned is that 100 years may not be enough to separate the signal from internal variability in these experiments (e.g. Peings et al. 2021; Ye et al. 2024).

-L110-115: Could these thresholds cause any selection bias because you may not identify certain jet configuration that are less zonal/wavier? You do analyze the number of days where you can’t identify the jet, but there could potentially be interesting things going on in these days that are missed.

-L177-186:  This analysis of ZWRI is a bit confusing. It should be clarified that the ZWRI from Smith et al. (2022) is calculated from the zonal mean over all longitudes. I am also not really sure what the point of this analysis is. Is it only to point out that the zonal wind response is stronger over the North Atlantic than in the zonal mean over all longitudes? The fact that the responses are stronger over the Atlantic (and Pacific) region (where the strongest jets are) has been shown in many studies, including Smith et al. (2022).

-L209: Why only include the mean shift for the models that have a statistically significant jet shift? What is that value if you include all models?

-L214 (and also L302, L375): Although it is not statistically significant, it does show a shift that is close to statistically significant (p=0.06), so it does not appear to be entirely inconsistent.

-L231-234 (also L320, L380): It is not clear to me that this contrasts with these previous studies. Ye et al. (2023) concluded that the jet speed response was weak, that models disagreed on the sign, and that there are only a few models that had a statistically significant response (and these did not agree on the sign). Overall, these seem consistent with the results found here.

 Smith et al. (2022) is interesting because they highlight the weakening of the midlatitude westerly winds (including in the abstract), but they never calculate any jet speed index. The figures themselves do not show clear evidence of the weakening of the jet because the weakening occurs only at higher latitudes and there is strengthening at lower latitudes (more indicative of an equatorward shift in the jet than a change in speed). The results of Smith et al. (2022) seem consistent with the results presented here even if some of the conclusions may not completely agree.

-L233-235: I do not understand this point. A positive ZWRI response as shown in Fig 4 is more indicative of an equatorward shift in the latitude which is seen when looking at the daily latitudes. It is not clear to me what the connection is between ZWRI and jet speed.

-L288-290: What is the percentage of days ways with no identifiable jet in the present-day simulations?

-L299: Is this value for all models or only the three that have statistically significant response?

-L349: Is it necessarily the case that a weaker jet is wavier? This is part of the Francis and Vavrus (2012) hypothesis, but I have not seen any theory or evidence that a weaker jet caused by Arctic warming would necessarily become wavier. A very recent study (Batelaan et al. 2024) finds a weaker jet in response to Arctic amplification in aquaplanet simulations, but a decrease in jet waviness.

-L357: I don’t think it has been conclusively proven that the signal-to-noise problem affects the response to sea ice loss and that this means the circulation response to sea ice loss is underestimated, although it is certainly plausible. This should be changed to ‘which may affect…”. This conclusion from Smith et al. (2022) seems to be somewhat challenged by the conclusions of Saffin et al. (2024).

-There is a lot of repetition between the Discussion and Conclusion sections. These repetitions should be minimized and there should be a clearer distinction between Discussion and Conclusions sections. Another possibility is to combine them into one section.

References:

Batelaan, T. J., C. Weijenborg, G. J. Steeneveld, C. C. van Heerwaarden, and V. A. Sinclair, 2024: The Influence of Large-Scale Spatial Warming on Jet Stream Extreme Waviness on an Aquaplanet. Geophysical Research Letters, 51, e2024GL108470, https://doi.org/10.1029/2024GL108470.

Blackport, R., and J. A. Screen, 2020: Insignificant effect of Arctic amplification on the amplitude of midlatitude atmospheric waves. Science Advances, 6, eaay2880, https://doi.org/10.1126/sciadv.aay2880.

Peings, Y., Z. M. Labe, and G. Magnusdottir, 2021: Are 100 Ensemble Members Enough to Capture the Remote Atmospheric Response to +2°C Arctic Sea Ice Loss? Journal of Climate, 34, 3751–3769, https://doi.org/10.1175/JCLI-D-20-0613.1.

Saffin, L., C. M. McKenna, R. Bonnet, and A. C. Maycock, 2024: Large Uncertainties When Diagnosing the “Eddy Feedback Parameter” and Its Role in the Signal-To-Noise Paradox. Geophysical Research Letters, 51, e2024GL108861, https://doi.org/10.1029/2024GL108861.

Ye, K., T. Woollings, S. N. Sparrow, P. A. G. Watson, and J. A. Screen, 2024: Response of winter climate and extreme weather to projected Arctic sea-ice loss in very large-ensemble climate model simulations. npj Clim Atmos Sci, 7, 1–16, https://doi.org/10.1038/s41612-023-00562-5.