Recommendation: Major revisions before publication

In this work, the mechanism of sub-decadal climate variability leading to extreme summer temperature events over Europe are investigated by using a Large Ensemble (LE) performed with the MPI-ESM model. The results are relevant in the context of climate prediction, the analyses are robust and well presented, as the quality of the figures is good. However, the language used in this manuscript is not precise and rigorous enough for a scientific article. There are many inaccuracies that make the text difficult to understand. Surprisingly the second part of the article, from section 3.3 is much better written than the first part.

Please, revise your manuscript and improve the precision of the language to make the text clearer. Below some examples and other remarks:

1.Abstract. The abstract is a very important part in the article. It needs to be well written. Please consider to rewrite the abstract to be more precise. Below an example of a rewritten abstract. This precision should be used throughout the text

The internal variability of European summer temperatures has been linked to various mechanisms from sub-seasonal to multi-decadal timescales. However, the mechanisms controlling sub-decadal (< 10 years) variations remain unexplored. We find that sub-decadal time scales (3-5 years) dominate summer temperature variability over large parts of the European continent. We show that extremely warm summers over Europe, occurring in sub-decadal periods, are related by a strengthening of the Atlantic Ocean subtropical gyre, an increase of meridional heat transport, and an accumulation of ocean heat content over the North Atlantic several years prior to the extreme event episode. The ocean warming affects the ocean-atmosphere heat fluxes, leading to a weakening and northward displacement of the jet stream and increased probability of occurrence of atmospheric blockings over Scandinavia. Thus, our findings link the occurrence of extremely warm European summers to the thermal inertia of the North Atlantic Ocean, whose potential to improve the predictability of extremely warm summers several years ahead is of great societal interest, especially in a warming climate.

1. Line 16: “increase in variability” needs to be more precise.
2. Line 21: Remove “as part of the large-scale….”
3. Line 25: “In detail” means what here ???
4. Different ocean-related quantities: Define
5. Line 28: Please, include all the references at the end of the sentence and not in the middle. This is applicable to the rest of the article.
6. Lines 31-33: Please rewrite, not very understandable.
7. In the introduction there should be a reference to DCPP-C (AMV pace makers experiments) and their findings. I think that the findings of this project are relevant and need to be mentioned in this article.
8. The term “Multi-year” is not precise. It is better to use sub-decadal.
9. We use the largest ensemble…are you sure ?? I think that other LE also provide 100 members.
10. What’s a single model initial condition large ensemble ?
11. North Atlantic ocean heat inertia, ocean is missing several times in the text after North Atlantic
12. Remove “doing so”, it’s too familiar
13. Line 41: We confirm that the MPI-GE can represent sub-decadal temperature variability well…, where?
14. Remove” by identifying periods that increase….summers”
15. …are done -> are performed
16. 12-150 km à ??? what does it mean ?
17. ERA5 reanalysis -> include reference
18. MPI-GE can represent the sub-decadal time scales well -> with respect to ???
19. Please rewrite sentence Doing so, ….
20. Remove “around the red noise”
21. Are you using daily data ? how the summer is defined here ?
22. The way the temperature extremes are defined in MPI-GE should be explained in the “Methods” section.
23. I understand that the computations are done separately for each member in MPI-GE ? So what it is presented in the figures ? the ensemble mean of the quantity ?
24. Line 81. This is the conclusion of the paper, this is indeed what you want to prove…
25. Which variable is used as “temperature” to determine the extremes ? surface temperature, 2m temperature ?
26. Line 87. In summary, sub-decadal timescales of 5-10….i do not understand the link to MEAN summer temperature since the focus are here the extremes…
27. Lines 90-91, Figure 1.d. I see that there are significant peaks at periods of 15 and 2-4 years. I would not conclude that the sub-decadal is the dominant frequency here…
28. Line 94. North Atlantic ocean heat content variability
29. Figure1.a and 1.b are not performed at the same spatial resolution, however it is mentioned that ERA5 is regridded on MPI grid, can you clarify ?
30. Line98. Please be more accurate and define “years” throughout the text.
31. These high “positive” anomalies…
32. Line 104. Some relation to other long-term climate variability models over the Pacific…can you specify??
33. Please, rewrite lines 106-110.
34. In addition to ENSO, have you consider the TPI (Tripolar Pacific Index) or IPV ? Anomalies shown in fig.2 look very much to TPI spatial structure.
35. Line 120. Please, define the meridional ocean heat transport better.
36. Line 145. “During the extreme.” Please be more accurate.
37. 149. “temperature anomalies fit…” do you mean spatially fit ??
38. Line 150. The heat transfer from the ocean to the atmosphere.
39. Line 152. Extremely European warm summers.
40. Line 165: The North Atlantic OCEAN heat inertia…
41. Line 168. “together with an above average…” of what??
42. Lines 170-171. Please, specify where the “released heat” goes…to the atmosphere??
43. Line 181. Rewrite: We find that the coupled oscillation in the North Atlantic prescribes extremely warm European summers on sub-decadal timescales.
44. Line 182. Other modes of ocean variability? Could you specify?
45. There are other mechanisms leading to the occurrence of extreme events over Europe-Mediterranean, this should be mentioned in the text. Please see Qasmi et al. 2021 and references therein.
46. The fact that the study was conducted with only one climate model should be discussed in the last section, as the mechanism described in this study could change from one model to another… so there is still uncertainty linked to the model used. We are not sure that a different Large Ensemble performed with a different model would give the same results.

General Remarks

The study covers a lot of ground, connecting 5-10 year European summer temperature variability to ocean heat content variability, and posits a plausible and interesting relationship between the two in the MPI-GE. There are sections where the writing is clear and sections that could use some work (e.g. with formality, dangling comparisons). My main concern with the study is the lack of context given along the way from a lead-lag relationship between climatic fields to the causality implied in the title "Extremely Warm European Summers driven by Sub-Decadal North Atlantic Heat Inertia" and throughout. The results are based on a single climate model, run with relatively low resolution, in a region that has some notoriously "obstinate" SST biases likely tied to unresolved oceanic processes (e.g. Athanasiadis et al. 2022). In addition, the North Atlantic sector has been identified as a region where the ocean-atmosphere coupling is weaker in models than in observations, creating issues for NAO variability, blocking, and decadal prediction (e.g. Simpson et al. 2018; albeit focused on winter, but that is the season the coupling is stronger to begin with). For the conclusions made in this paper to stand as they are written, there will need to be a convincing argument made for each link in the causal chain that the MPI-GE lack significant biases with respect to observed fields and that X indeed induces Y. For example, the “driver” is concluded to be a build-up of heat in the North Atlantic Current, but that accumulation ultimately can be traced back to atmospheric variability, right?  If you were to impose the ocean heat anomaly in the model, would the atmosphere respond as you describe? I would recommend toning down the conclusions to reflect what is really being explored, the relationship between North Atlantic OHC variability and periods of exceptionally warm summers in the MPI-GE.

Specific Comments

L3:  Re: "... remain unexplored": Some work appears to be done in the realm, including by the co-authors (e.g. Müller et al. 2020).

L4-5: Please revise this sentence for clarity.

L19: Consider highlighting the recent work of Röthlisberger and Papritz (2023).

L29-31: I’m not convinced this is true. I’ve included a few potential references, but I feel a deeper dive into the literature is warranted.

L36-37: Best in what way?

L39: Re: “Including some of the most extreme European summers”: What do you mean by this? Extreme compared to what?

Section 2.1 Model Description: Maybe in this section, you could also note your study domain and the fields you will use for each part on the analysis

L55: Parenthetical should be its own sentence.

Section 2.2: This section is structured in a very atypical way. Please revise and avoid the use of sub-headings.

L63: How precisely do you determine the significance of spectral peaks?

L65: Is this a linear detrending? Is that appropriate for “all of [y]our data”?

L69: Randomly composed arrays of..?

L74/76: I’m a bit lost. What is meant by: “total summer mean variability during extremely warm summers”? Is this interannual variability? Decadal? Assessed only during warm periods? How are “for times when heat extremes occur” defined?

L79-80: “Also, timescales between 10 and 20 years are dominant in only a few more grid points compared to timescales above 20 years.” What do you mean by this?

Figure 1:

• I thought the ERA5 grid was decimated to match the MPI-GE grid?
• There seems to be disagreement on the dominant timescale of SAT variability in your study region between ERA5 and the MPI-GE. Could you comment on that?
• There are 6 extremely warm summers per 5–10-year period in each ensemble member? How can (almost) every summer be extreme?
• It may make the figure too messy, but it would be nice to see the power spectra of each individual member, maybe in a supplement? And isn't the dominant variability cycle at around 15 years?

Section 3.3: What's missing here is validation that the low-resolution ocean model can capture these processes.

L116-117: What initiates this? How is it related to the AMO?

L149-150: How do you know this is "bottom-up" driven and not "top-down" driven?

Figure 4: Are these the ensemble means?

Figure 5: Including the OHC branch of the mechanism in the schematic would be helpful.

Citations to consider

Athanasiadis, P. J., and Coauthors, 2022: Mitigating Climate Biases in the Midlatitude North Atlantic by Increasing Model Resolution: SST Gradients and Their Relation to Blocking and the Jet. J. Climate, 35, 6985–7006, https://doi.org/10.1175/JCLI-D-21-0515.1.

Simpson, I. R., Deser, C., McKinnon, K. A., & Barnes, E. A. (2018). Modeled and Observed Multidecadal Variability in the North Atlantic Jet Stream and Its Connection to Sea Surface Temperatures. Journal of Climate, 31(20), 8313–8338. https://www.jstor.org/stable/26508075

Röthlisberger, M., Papritz, L. Quantifying the physical processes leading to atmospheric hot extremes at a global scale. Nat. Geosci. 16, 210–216 (2023). https://doi.org/10.1038/s41561-023-01126-1

Hall, R.J., Jones, J.M., Hanna, E. et al. Drivers and potential predictability of summer time North Atlantic polar front jet variability. Clim Dyn 48, 3869–3887 (2017). https://doi.org/10.1007/s00382-016-3307-0

Osborne, J. M., M. Collins, J. A. Screen, S. I. Thomson, and N. Dunstone, 2020: The North Atlantic as a Driver of Summer Atmospheric Circulation. J. Climate, 33, 7335–7351, https://doi.org/10.1175/JCLI-D-19-0423.1.

Wu, B., Zhou, T., Li, C. et al. Improved decadal prediction of Northern-Hemisphere summer land temperature. Clim Dyn 53, 1357–1369 (2019). https://doi.org/10.1007/s00382-019-04658-8