Summary

The premise of this paper is obviously useful: Extreme synoptic-scale cyclones in the Arctic seem to develop differently than extreme synoptic-scale cyclones in the mid-latitudes, but that theory is based on only several case studies, not a comprehensive review of many storms. The implications the authors present in the paper (that prediction of, projection of, and impacts of these storms all depend on how they develop) make this worthwhile research. Honestly, I’m surprised it’s taken so long for somebody to undertake a composite analysis like this. I am therefore very happy to see that the attempt here has a thorough and logical approach. The figures are clear and well-made. The methods used are ones that align well with the research questions. I have some comments, but everything can be addressed at the text level.  

 A. Content Comments

1. The authors focus on summer Arctic cyclones and how they are distinct from winter Arctic cyclones and winter North Atlantic cyclones. A logical question is: Why winter North Atlantic cyclones rather than summer North Atlantic cyclones? At first blush, the latter seems more logical to control for region.

I think I can infer the reason: The standard cyclone models we use are primarily based on the winter season, and the Norwegian Model is primarily developed from winter North Atlantic cyclones. Therefore, using the North Atlantic winter (rather than summer) is closest to giving a “standard model”. However, I think it would be better to make this explicit in the text – probably around line 65-80 – but I’m not sold on that being the best place.

As an aside, if the authors delved more into “why” extreme Arctic summer cyclones are so different, I think it would be essential to include summer North Atlantic storms in order to control for both season and region. But the authors are writing a paper that outlines “what”, not “why”, so I think the format they chose is fine. Again, I think the logic just needs to be more explicit since it’s counter intuitive.

2. Lines 46-47, Figure 2, Lines 371-372: Simmonds and Rudeva (2014) is a good reference for a few things: The linkage with tropopause polar vortices for extreme cyclones, the location difference between winter/summer extreme storms, and the long lifetimes for extreme Arctic storms. Incorporating that paper for one/some of these reasons would be good.

3. Line 93-95: Vessey et al. (2020) showed that using relative vorticity (rather than mean SLP) leads to greater cyclone frequency using the Hodges algorithm, but they didn’t prove that it is true for other tracking algorithms in use. Therefore, it’s best to specify “with this algorithm”, or something similar.

4. Line 167 – 169: I don’t think using a common isobar is a good comparison of size. The 1000 hPa is not the edge of a cyclone, so it doesn’t have much meaning – especially when storms from different seasons/areas have different central pressures. It would be better to use a) the area with cyclonic relative vorticity or b) the area within the last-closed isobar (given some isobar interval). One of these might be able to be incorporated into the left-hand panels (although the current scale might be too large).

Alternatively, just with text changes, the convex 1008-hPa isobar in summer being entirely within the 2200 km by 2200 km box versus the (seemingly) convex 1016 hPa isobars in the winter cases extending beyond that window seems more compelling. The limit of the 10 m/s wind field would also be reasonable if focused on storm impacts.

5. Figure 3: Thank you for the very obvious direction arrows. That is helpful.

6. Lines 193-196: Clancy et al. (2022) also used different input data (ERA-Interim for one algorithm, NCP-NCAR reanalysis for another – the ERA5 data they use appears to be for non-cyclone tracking purposes), so the authors can add that to the list of differences

7. Lines 195-196: Using the top quartile v. the top 100 is difficult to compare with different periods and overall frequencies from the different algorithms. Reporting what percentage of storms fall into the “top 100” would be an improvement on the imprecise statement “many more”.

8. Figure 9: The “Summer Arctic Cyclones” left of the y-axis label is superfluous and likely better off being replaced with the distinguishing characteristic: north-south v. east-west.

9. Line 316-320: I disagree with the authors on the statement here that Aizawa and Tanaka (2016) did not consider baroclinic to barotropic transition. In their abstract, they state:

“The cyclone of 2012 is characterized by the structure change from the cold core to the warm core at the lower stratosphere, indicating a shift from the ordinary baroclinic cyclone to the typical Arctic cyclone.”

Later, they state:

“In the early stage [of the 2012 case] (Fig. 5), the vortex shows clearly the baroclinic structure."

Therefore, they do discuss this transition for the 2012 case. Still, I think it’s fair to say that they do not make a transition from baroclinic to barotropic structures part of their overall conceptual model. In other words, I think a few tweaks to text are in order – in particular, pointing out that the 2012 case exemplifies this finding about transitions

10. Summary and Conclusions: I think bringing up the Clancy paper one more time here is important. Perhaps their different results indicate that the extreme storms differ from the average storms in the Arctic. But as the authors pointed out earlier, differences between the studies might also explain some of the discrepancy in results. There might be some “future work” statements around this, but at the very least a comment about the implications is worthwhile.

B. Proofreading Comments

1. Throughout text and figure captions: The authors are using propagation-relative grids but they also often use north, south, east, or west to describe positions relative to cyclone centers. It’s not always clear whether these are earth-relative or map-relative (e.g., “north” = top of map). For clarity, the authors could a) always specify earth-relative or map-relative (i.e., propagation-relative) or b) describe the position as left, right, ahead, or behind the cyclone (as is sometimes done already). Note: I say this all recognizing that because cyclones generally move west to east, map-north and earth-north are fairly close to each other. So it’s a small thing.
2. Line 15: Replace “National Snow and Ice Data Centre” with “National Snow and Ice Data Center”. (The USA apologizes for the inconvenience of its disparate spelling.)
3. Line 19-20: Since the Stroeve et al. (2007) is primarily about model validation, not model projections, I suggest using either Notz & SIMIP Community (2020) or Årthun et al. (2021) instead to make this point.
4. Line 105: I believe “system centered” should be hyphenated as “system-centered”.
5. Line 107, 124: An apostrophe is needed to make a possessive, changing “cyclones” to “cyclone’s”.
6. Line 124: The phrase “common direction” is unclear because previously, the terms “geographical orientation” and “oriented according to the cyclone’s propagation direction” were used – but not “common direction”. By default, then, I’d assume that “common” means “absolute”, or, in this case “geographical. However, the authors then specify “propagation direction” in paratheses, showing my default assumption is incorrect. It might be clearer, then, to just say outright “An advantage of rotating each cyclone relative to propagation direction…”
7. Line 215: Change “typically” to “typical”.
8. Line 253: A closing “)” is missing.
9. Line 279: Replace “at the after” with “at and after”.
10. Line 305: Based on the figure concept, vertical velocity is inverted to be positive upwards, so this ought to say “negative vertical velocity”.
11. Line 306: Change “Only a weak” to “Only weak”.

Full citations for references in the review that are not in from submitted manuscript

Årthun, M., I. H. Onarheim, J. Dörr, and T. Eldevik, 2021: The seasonal and regional transition to an ice-free Arctic. Geophys Res Lett, 48, https://doi.org/10.1029/2020gl090825.

Notz, D., and S. Community, 2020: Arctic Sea Ice in CMIP6. Geophysical Research Letters, 47, https://doi.org/10.1029/2019gl086749.

Simmonds, I., and I. Rudeva, 2014: A comparison of tracking methods for extreme cyclones in the Arctic basin. Tellus A, 66, 25252, https://doi.org/10.3402/tellusa.v66.25252.

In recent times the Arctic region, and its tendencies, have attracted much interest in a wide range of respects. This submission uses a composite approach to explore the nature of extreme Arctic-basin storms, how they differ between summer and winter, and how their revealed behavior differs from their midlatitude counterparts. Also revealed is the complex roles played by low-level baroclinicity and upperlevel processes.

The work builds on earlier research by these and other authors. It presents some new insights, but does require significant revision.

Line 15: National Snow and Ice Data Center

Lines 15-17: Also to reference here recent Arctic sea ice analysis of Li et al (2021) Trends and variability in polar sea ice, global atmospheric circulations and baroclinicity. Ann. NY Acad. Sci. 1504: 167-186 doi: 10.1111/nyas.14673.

Lines 51-54: Topical and relevant papers of Takuji Waseda, Takehiko Nose, Tsubasa Kodaira, Kaushik Sasmal and Adrean Webb, 2021: Climatic trends of extreme wave events caused by Arctic cyclones in the western Arctic Ocean. Polar Science, 27, 100625, doi: 10.1016/j.polar.2020.100625, and Vernon A. Squire, 2020: Ocean wave interactions with sea ice: A reappraisal. Annual Review of Fluid Mechanics, 52, 37-60, doi: 10.1146/annurev-fluid-010719-060301 valuable additions here.

Lines 111-112: The algorithm used here identifies cyclonic systems based on 850 hPa relative vorticity, rather than mean sea level pressure which is often used in other studies. The former parameter does have a number of advantages. However, I was puzzled as to why the central value of MSP rather than the (filtered) vorticity was used to identify the 100 most intense cases. Some words should be devoted to the rationale behind this change of focus. (A few lines below (at line 131) the authors state that when cyclones’ MSLP reach their minima is ‘when they are at their most hazardous’. One would have thought this would occur for the 850 hPa rel. vort. reaching its extreme.

Lines 120-121: Please to be explicit as to the determine of the propagation direction. I am assuming that the propagation ‘vector’ is taken to the displacement of the MSLP minimum. Similarly, I would guess that the direction here is determined by an (unbiased) centred difference of the central location at + 6hr minus -6hr. Important to spell out.

Similar issue for composites prior to, and after, the most intense time (lines 132-133, …). Are these directions determined from the 12 hr location differences centred on the relevant time, or is same propagation value used for all of these?

Lines 124-125: A good point is made here in connection with retaining the integrity/identity of conveyor belts when performing the required rotations. These features are known to be intimately involved with various extremes associated with cyclones and fronts. Emphasise this very important message by also referencing the article of Jennifer L. Catto, Erica Madonna, et al., 2015: Global relationship between fronts and warm conveyor belts and the impact on extreme precipitation. Journal of Climate, 28, 8411-8429, doi: 10.1175/JCLI-D-15-0171.1.

Lines 179-180: Perhaps reword this sentence. ‘In this region, the system-relative winds would be enhanced by the cyclonic (anti-clockwise) propagation velocity of the cyclone’ could be read as saying the system-relative (as distinct from ‘earth-relative’) winds would be STRONGER, whereas I am sure the authors want to say the exact opposite.

Lines 192-196: While they had a larger testing set (lowest quartile of all Arctic cyclones), these comments perhaps do not fully convey the procedure and results obtained by Robin Clancy et al. As distinct from the measure used here they made use of two measures of intensity, namely (i) the local Laplacian of SLP and (ii) the maximum SLP within a closed contour around a cyclone minus the minimum SLP within this contour. The former is closely related to the geostrophic relative vorticity, while the latter is very similar to the cyclone ‘depth’ (make reference in paper to Murray et al. (1995), Responses of climate and cyclones to reductions in Arctic winter sea ice, J. Geophys. Res., 100, 4791-4806, doi: 10.1029/94JC02206). (I had made some comments earlier on whether the definition of ‘intensity’ in the present paper is really appropriate.) As such, the Clancy method may be seen as more appropriate to identifying extreme systems and could, notwithstanding the different sample sizes, be seen as one of the reasons for the differing conclusions reached here.

The authors need to provide a more complete analysis and discussion on the apparent discrepancies on this central point.

Lines 207-209: It would be helpful and insightful to express these omega ‘speeds’ in terms of approximate m/s (or cm/s) making use of hydrostatic equation etc.

Line 256-260: It is not clear to me how the summer composites at 96, 144 and 192 hrs (i.e., 4, 6, and 8 days) in Figure 7 were constructed. From Figure 1b 22 (6+16) of the 100 intense ACs last 6 days or less. So, e.g., for the +8 day composites (Figure 7c) were only the systems that survive to 8 days considered, or was the atmospheric structure somehow incorporated into the composites. Either way, there will be some bias in those later lags. The procedure followed should be made clear here.

Also, at line 256 ‘From’ is potential ambiguous. Would suggest replace with ‘After’.

Lines 290-320: This section should be more carefully written to reflect the different development roles of low-level baroclinity and upper level support. A key aspect of intense Arctic cyclones is known to be the presence of Tropopause Polar Vortices (TPVs). These points were made very strongly be Tanaka, Yamagami and Takahashi, (2012) and Aizawa and Tanaka (2016). Make reference also here to study of Rudeva et al., 2014 (‘A comparison of tracking methods for extreme cyclones in the Arctic basin. Tellus, 66A, 25252, doi: 10.3402/tellusa.v66.25252) who showed that, in all months, the vast majority of extreme Arctic cyclones were associated with a TPV. TPVs as such are mentioned in the paper, but only in the last few lines. This important structure and concept must be introduced much earlier, and more clearly integrated in to the analysis.

Lines 301-302: I’m not sure where the ‘4,400 km’ figure for the surface system came from – it certainly was not mentioned in the 2016 paper of Takuro Aizawa. They suggest, for their 2008 case, a diameter of 3000 km (their Fig. 2 and Page 193), based on the radial and tangential 10 m wind. Their 2012 storm had a similar diameter (their Figure 3a). (Perhaps the present authors are confusing that Aizawa and Tanaka refer to diameter of up to 5000 km at the UPPER LEVEL (an enlargement that would be expected from the dynamics).  Please write this, and the conclusions from it, more carefully.

Lines 309-310: No needing to include the statement ‘… which was based on the analysis of just two past Arctic cyclone case studies’. This has already been made clear a few lines above (lines 290-291).

Lines 322-372, Section 4: The focus of the paper has been on the composites of the 100 winter and summer cases. The authors have not presented the dates of these individual events; this is quite appropriate in that that level of detail might needlessly distract the reader’s attention from the main theme. Having said that, it would be of interest to know whether there have been any trends in the frequency of these events. Maybe a timeseries (by year) would be of interest. Or more simply, the technique I saw used by Rudeva I. and co-authors (2014) A comparison of tracking methods for extreme cyclones in the Arctic basin. Tellus 66A, 25252, could be used to establish statistical significance or otherwise of any shift in the ‘centroid’ of yearly occurrence of extremes.

Lines 416-417: Please note this paper has passed the ‘the ‘Discussion’ phase, and has now been published. Details are …

Gray, S.L., Hodges, K.I., Vautrey, J.L. and Methven, J., 2021: The role of tropopause polar vortices in the intensification of summer Arctic cyclones. Weather and Climate Dynamics, 2, 1303–1324, doi: 10.5194/wcd-2-1303-2021.

Line 470-471: Please append correct details to reference:

Waseda, T., Webb, A., Sato, K., Inoue, J., Kohout, A., Penrose, B. and Penrose, S. 2018. 'Correlated increase of high ocean waves and winds in the ice-free waters of the Arctic Ocean', Sci. Rep. 8, 4489, doi: 10.1038/s41598-018-22500-9.

Summary:

Overall very interesting and well-written paper. The paper studies the climatology of intense Arctic cyclones, something that has previously been done mostly based on case studies. The authors utilize a system-centered composition analysis following the cyclones through their lifetime allowing for complete and easily comparable climatologies to be made. The paper extends our knowledge on summer Arctic cyclones and makes the important point that Arctic cyclones can be structurally different depending on the season. Some minor corrections could be made into the text to enhance readability and clearing out some of the method choices made.

Comments:

1. Line 69: Any newer references on this topic? Maybe Varino et al. (2018) or Wicktröm et al. (2017)?

2. Line 99: How are the identified cyclones grouped as cyclone tracks? The authors mention the nearest neighbor approach, but I assume there has to be some sort of limit, as to how far connected cyclone instances can exist from one another in order to be considered part of the same cyclone?  

3. Line 111: Why only North Atlantic winter storms were compared, why not compare winter storms to winter storms and summer storms to summer storms?

4. Line 155: Adding another reference would be good.

5. Line 173-180: The discussion on 850hPa winds is interesting, but it made me curious how different the results would be if you looked at wind speed closer to ground (and maybe over land vs. ocean). A paper by Valkonen et al. 2021 showed that over different sea ice regimes the cyclone wind speeds were different and one would assume even larger differences between land and sea surface. Any comments?

6. Line 219-222: Please rephrase this sentence, it is very long and a bit difficult to understand.

7. Line 301-302: As the size of the cyclones in this paper is compared to that of the Aizawa and Tanaka (2016) cyclones, it would be important to know if the size of the cyclone was determined in the same way in both papers. If it was not, it would be good to give an explanation as to why the last closed isobar was chosen as the area of the cyclone in this paper.

In text references:

Varino, F., Arbogast, P., Joly, B. et al. Northern Hemisphere extratropical winter cyclones variability over the 20th century derived from ERA-20C reanalysis. Clim Dyn 52, 1027–1048 (2019). https://doi.org/10.1007/s00382-018-4176-5

Wickström, S, Jonassen, MO, Vihma, T, Uotila, P. Trends in cyclones in the high-latitude North Atlantic during 1979–2016. Q J R Meteorol Soc. 2020; 146: 762– 779. https://doi.org/10.1002/qj.3707

Valkonen, E., Cassano, J., & Cassano, E. (2021). Arctic cyclones and their interactions with the declining sea ice: A recent climatology. Journal of Geophysical Research: Atmospheres, 126, e2020JD034366. https://doi.org/10.1029/2020JD034366