the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Feb 2026
The Influence of Atlantic Multidecadal Variability on European Summer Climate: Competing Mechanisms and Implications for Prediction
Abstract. Skilful predictions of European summer climate are increasingly relevant due to an increasing probability of temperature extremes, but prediction skill beyond the forced trend has so far proven limited. Atlantic Multidecadal Variability (AMV), characterised at the surface by North Atlantic sea surface temperatures (SSTs), is both active and predictable during boreal summer, and previous studies have linked it to surface impacts in Europe. Current understanding largely relies on the relatively short observational record of decadal variability and the predictability of impacts and associated mechanisms are poorly studied. In this study, single model large ensemble historical and decadal hindcast simulations using the MPI-ESM-LR model are used to understand the role that AMV plays for North Atlantic-Europe sector climate prediction. It is found that strong AMV-associated SST anomalies in the subpolar gyre region are better represented in the initialised hindcasts than in the uninitialised historical ensemble, and they are highly predictable at lead years 1–7. The observed cyclonic response to positive AMV in the extratropical North Atlantic is not present in historical simulations, but it is found to be predictable in decadal hindcasts, although with underestimated amplitude. The hindcast pressure anomaly nonetheless skilfully predicts observations and highlights a potential role for AMV in the yet-unsolved "signal-to-noise paradox". The upper tropospheric (200 hPa) geopotential height response to AMV is analysed and it is found to differ in reanalyses and models. Further investigation reveals a high frequency component relating to tropical SST anomalies and resembling a Rossby wave train emanating from the Caribbean, and a low frequency component relating to the surface level response, with an imbalance between the two mechanisms in models due to the weak surface response.
- Preprint
(6478 KB) - Metadata XML
-
Supplement
(83 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-6330', Anonymous Referee #1, 25 Mar 2026
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2025-6330/egusphere-2025-6330-RC1-supplement.pdfCitation: https://doi.org/
10.5194/egusphere-2025-6330-RC1 -
RC2: 'Comment on egusphere-2025-6330', Anonymous Referee #2, 31 Mar 2026
Summary
The authors investigate the predictability of the North Atlantic-European summer time circulation focusing on the time scale of 1–7 years. In particular, the atmospheric response to low-frequency SST variability related to the Atlantic multidecadal variability is assessed. The authors then aim to attribute predictability of SLP and jet stream-level geopotential height to two established physical mechanisms: a heating-induced surface low downstream of the largest SPG SST anomalies and a stationary Rossby wave train initiated by convection in the Western Tropical Atlantic. The authors make use of an MPI large ensemble involving decadal ensemble projections initialized each year and compare against observations.
It is found that the hindcast ensemble skillfully predicts SLP over an Eastern North Atlantic domain. Nevertheless, both SLP and geopotential height responses to AMV are underestimated in the hindcasts. The response to diabatic heating over the SPG region is found to be too shallow in model runs while the signal linked to a Rossby wave train dominates, which is found to be at play in reanalysis as well but less dominant compared to the SPG heating mechanism due to stronger vertical coupling in the SPG region. Predictability only emerges when using a large ensemble rather than from individual members, consistent with previous research on this signal-to-noise paradox.
Review
In my view, the questions addressed in this work are highly interesting and relevant. The physical-mechanism approach to predictability is very appreciated and, from my perspective, the analysis of atmosphere-ocean interactions is adequate for this journal. Both datasets and statistical methods are sufficiently sophisticated to tackle the research questions and I consider the results to be valuable for the community. The arguments and interpretations based on the figures are sound and I only have minor comments on the figures themselves.
I do see a few important aspects in which this manuscript needs improvement. The accessibility and coherence of the manuscript would benefit from a clearer description of detailed research objectives that are referred to after the introduction. In some instances, which are detailed in my comments below, the flow can be improved or the structure feels odd. Moreover, the contextualization of the results can be improved. My background is in atmospheric science and I have some expertise in atmosphere-ocean interactions. I am less familiar with literature specific to the AMV and would have appreciated it if the results were more embedded in or compared to similar work to assess the impact of the findings. Limitations, for instance with regards to model dependencies or deficiencies, are generally only sparsely discussed. Lastly, the descriptions of methods and the language used are occasionally imprecise. Overall, while I think the scientific quality of manuscript is good I suggest the authors perform revisions that I would label as major.
Major comments
- The current manuscript contains very accessible summary paragraphs (e.g. L329), some smooth transitions between different analyses (e.g. L206), and clearly formulated research objectives (e.g. L32). Yet, in a few instances the occurrence within the text was unexpected to me. Arguably, there is no right solution to structuring, but the current version of the manuscript has left me disoriented at times – especially during the first iteration of reading.
- The first paragraph of the introduction contains parts of the motivation, a methodological information, and the overarching research objective. It feels a bit like a repetition of the abstract. The overarching research objective alludes to physical mechanisms that are only explained later. In my opinion, the introduction would benefit from a reordering such that research objectives naturally emerge from introduced concepts or gaps. Currently, you repeat differently phrased “aims of the study” in lines L29-34 and L64, which – up to personal structure preferences – could be circumvented by stating one overarching goal and more detailed ones that hint at the structure of the results. Moreover, while I do appreciate the present conciseness, the AMV (and the word horseshoe pattern) could be introduced earlier or with another sentence.
- In Section 4, many paragraphs start with “Figure xy shows …”. I suggest refining the transitions between these paragraphs such that it becomes clearer what is done next and why. The summary from line L257 is very clear, whereas the discussion of Fig. 7 from L228 that starts with introducing the method of year-to-year variability was not very accessible to me.
- In the “Discussion and Conclusions” section, there is a short (kind of stand-alone) summary at the very end. While this might be a question of style, I suggest that this is reordered or labelled better. Moreover, in the current version the first sentence of the final section reads L264 “The capability of … at capturing spatial patterns of SSTs … was tested.” while the overarching goal stated in the introduction is to assess the L33 “atmospheric response to AMV”. Of course, the atmospheric response is discussed right after, but at first this to me seemed like a disconnect between the stated goals and the summary of the work. Please improve coherence in that regard throughout the paper. Similarly, predictability is the last word of the title (as if it were just a side aspect) but then is a key aspect of the abstract and rest of the paper. The way I understood the manuscript, “Predictability of European summer climate: influence of competing mechanisms related to the AMV” could also be a suitable title.
- There are a few instances where methodological details could or should be discussed in more detail. In particular:
- Can you provide a motivation or reference for why to use exactly 7 years for averaging?
- Please specify the model resolution.
- L70 onward: I would have appreciated a (basic) description on how the hindcasts are initialized (like which data is used for initialization) instead of having to go to the indicated references. This would also help to better understand possible limitations.
- L89: The anomaly with respect to what precisely?
- I assume that the detrending using the ensemble mean timeseries is applied to both historical and hindcast model runs, is that correct? In L185, you indicate that also the reanalyses data are detrended. Could you specify how this is done? Is this using the trend derived from the model?
- Figure 1d,e: Could the difference between the two panels be explained by increased SST variability in the SPG region from the early historical period towards the hindcast period? Do the historical simulations exhibit a larger AMV signal during the hindcast period only (without the imposed observational constraint present in the hindcast ensemble)? In other words, I was curious to see Fig. 1d repeated for the hindcast period. More generally, how do MPI SST trends in the historical ensemble compare to observations in the SST region? Would the model simulate, for instance, a stronger or weaker SPG cooling than observed if not constrained as in the hindcast ensemble? Are there any systematic model drifts in the hindcast ensemble after initialisation that would not be filtered out by removing the forced trend determined from the historical simulations?
- Figure 2: From a quick computation, the temporal standard deviation of the JJA mean SLP lies between 2 and 2.5 hPa in ERA5 to the west of the UK. The differences between the AMV+ and AMV- years in Fig. 2 seem comparably low with 0.9 hPa in reanalysis and 0.35 hPa in models. Similarly, most readers likely won’t know whether anomalies of 15 m in 200 hPa geopotential height are a lot (on the synoptic scale, for instance, they aren’t). Can you contextualize the seasonal signals that you find with natural variability and other studies? To what extent would you argue that your decadal predictions can provide an added benefit to people from the climate impact community (compared to existing studies)?
- L142: “The strong SPG signal remains, suggesting that it is highly predictable (consistent with previous studies, e.g. Borchert et al. (2021))” Can’t this be easily supported (possibly in the supplement) using spatial correlations as in Fig. 3, for instance, instead of via defining indices and deriving differences of subsets of the time series thereof?
- L295: Could you further discuss possible reasons for the underestimated ocean-atmosphere coupling? Based on your investigations, do you think that this is related to model resolution, mean-state biases in surface fluxes, vertical stability, or something else? In general, the discussion of possible model biases is rather short.
- I encourage the authors to illustrate the way the competing mechanisms act in the reanalysis versus the hindcasts using a figure schematic. This would serve as a concise summary that readers could refer back to while reading.
Minor comments
- L25: Studying decadal prediction of summertime temperatures is motivated by “increased risk of extreme heat” on the scale of an individual summer (Rousi et al. 2023). It would be more appropriate to keep the time scale consistent here and, ideally, not refer to extreme surface heat as long as land surface temperatures are not subject of the paper.
- There are some instances where the level of precision, which is generally good throughout the manuscript, seems to need some improvements. This list may not be complete:
- L11: It reads a bit weird that the “anomaly … predicts observations”.
- L26: You describe “Atlantic Multidecadal Variability” as “associated with low frequency variability of North Atlantic sea surface temperatures” – isn’t it rather “defined by” than “associated with” (at least in the present context where only the atmospheric response to the SSTs is studied)?
- L51: Does “components” refer to “atmospheric responses”?
- L242: Was this actually “demonstrated” in Figs. 5 and 7, or is it rather “consistent with” them? Maybe I am missing something.
- Figure 1 and others: where it makes sense, including the respective time periods in the subfigure titles could help some readers.
- Figure 4 is very neat. The way you explain it in the text, the arrangement of a), c) and b) makes sense. I assume this arrangement looks better than mirroring panel b) to the left? In the caption, the “(a) (top)” etc. is tautologous.
- L210: I suggest adding or repeating the relevant literature references here.
- Figure 6 or L214: It seems like there is a simple derivation for this, but could you provide a reference for white noise yielding a L^(-0.5) curve for the ratios at different leads?
- L231 and the discussion section: Adding figure references to statements would significantly facilitate reading through these paragraphs and connecting the key take-aways with your specific analyses.
Technical comments
- Please revise the formatting of figure references.
- Figure 8: The ERA5 line is largely outside of the figure bounds; please consider introducing an axis break or another solution to show the full range of the data (even if the conclusion that it is negative does not rely on it).
Citation: https://doi.org/10.5194/egusphere-2025-6330-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 162 | 92 | 15 | 269 | 29 | 44 | 43 |
- HTML: 162
- PDF: 92
- XML: 15
- Total: 269
- Supplement: 29
- BibTeX: 44
- EndNote: 43
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1