the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 19 Jun 2026
Microphysical Properties and Ice Crystal Habits of Low-Level Arctic Mixed-Phase Clouds from Four Airborne In Situ Campaigns over the Svalbard Seas
Abstract. This study presents a statistical analysis of the microphysical properties and ice crystal habits of low-level Arctic mixed-phase clouds (MPCs) using in situ observations from four aircraft campaigns conducted near Svalbard, Norway. The dataset comprises 45 hours of in-cloud measurements, with 731 600 ice crystal images classified using a convolutional neural network. Observations were categorized into four meteorological regimes using the Marine Cold Air Outbreak index: Marine Cold Air Outbreaks (MCAOs), Warm Air Advections (WAAs), and two transitional states. Vertical profiles of microphysical properties exhibit clear differences between regimes. MCAO conditions are associated with deeper mixed-phase layers, larger ice water contents, and higher concentrations of ice particles. In contrast, WAA cases are associated with shallower clouds dominated by liquid water with lower ice water content. The analysis indicates that temperature inversions influence cloud microphysics: weaker inversions support deeper clouds and enhanced ice development, while stronger inversions limit the extent of the liquid-containing layer. Analysis of ice crystal habits reveals distinct growth patterns based on the meteorological regime and surface type. Rimed particles appear more frequently during MCAOs, particularly over open ocean where increased turbulence likely promotes droplet–ice interactions. Over sea ice, pristine and aggregated crystals are more prevalent, implying that riming plays a lesser role while depositional growth and aggregation are more significant. This combined multi-campaign dataset offers a comprehensive statistical view on the interaction of synoptic regime, temperature inversion, and surface conditions in determining the microphysical structure and ice crystal habits of Arctic MPCs.
Competing interests: At least one of the (co-)authors is a member of the editorial board of Atmospheric Chemistry and Physics.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.- Preprint
(18046 KB) - Metadata XML
-
Supplement
(9607 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-3265', Anonymous Referee #1, 01 Oct 2026
-
RC2: 'Comment on egusphere-2026-3265', Anonymous Referee #2, 21 Oct 2026
Dziduch et al, present composites of cloud microphysical and thermodynamic properties from four different aircraft campaigns over the Fram Strait. The campaigns took place during different seasons and over five years, providing a wide range of conditions to be investigated. They bin the measurements by thermodynamic profile and normalize the in situ measurements relative to cloud base and top. This is a very nice approach and appears to provide a consistent framework in which to investigate cloud microphysical properties. The description of the results is very detailed but whether the difference in the microphysical properties is significantly different across the meteorological regimes is missing. In fact, in many cases the microphysical property is almost identical regardless of the meteorological regime, making one wonder if the regime matters for the cloud properties or if the cloud properties are just governed by e.g. cloud top temperature. At the same time, many of the discussion points in the manuscript could be expanded upon on and compared with previous literature. While other points could be significantly shortened as if the differences are not significant, then there is no need to describe them in so much detail. Removing the description of ambiguous cloud properties across regimes would shorten the text, allow for more discussion on the properties that do truly differ and thereby, make the manuscript much more impactful and easier to read. Therefore, I encourage the authors to address and consider the points below before the manuscript is published.
Major Comments:
- I appreciate that the authors would like to develop the typical profiles that represent these different airmasses, but providing the 99 % confidence interval of the median hides the observed variability across the different clouds observed in a given meteorological situation. Sure, this shows that the median is representative but it doesn’t mean that the profiles in the meteorological conditions are different from each other etc. When looking at the liquid phase, there is likely no significant difference in microphysical properties between these different regimes except for the WAA regime. Similarly, the ice microphysics are very similar except for the CAO case and maybe the Di for Norm-C. But even these differences are not necessarily significant or at least it is not possible to determine this when presented like this.
That said, it is not bad to separate the cloud profiles by thermodynamic regimes but the conclusion that there are clear differences is potentially overstated and the drivers of these differences may in the end just come down to cloud top temperature and not the “meteorological regime.” For example, w2 is basically the same except for during the CAO regime, so can it really be said that the instability profile is actually changing the cloud microphysics between these other regimes or is it just the CTT?
- When presenting the Zn profiles, it would be worthwhile showing how many observations are observed for each meteorological regime and on how many different flight days these come from. This would provide more information than the general time in cloud etc, as it matters how well the profile space was sampled etc when developing these statistics.
- Consider leaving the extinction out as it will shorten the discussion and it does not really show more than what is shown by the mass profiles.
- When discussing the habits of MCAO events, consider discussing SI/MIZ zone first and then the OO as that is the evolution of an MCAO. Additionally, more emphasis can be made on the habit distribution and what causes the OO MCAO clouds to be rimed effectively throughout the cloud while over the SI/MIZ rimed fraction increases near the surface. This is a cool find and thus could be linked more to the differences in turbulence etc.
Minor comments:
Line 45: For surface coupling you could include Griesche et al., (2021)
Line 52: Kanji et al., (2017) seems like the wrong ref here. Consider including it for the next sentence and INPs should maybe be introduced a bit better.
Line 51-58: could be reformulated for an easier understanding of the mechanisms involved in ice formation and their drivers.
Line 66-67: the start of the sentence doesn’t really make sense as liquid water was not necessary for some of the mechanisms listed to influence the crystal shape.
Figure 2.b: Are the numbers in the plot really hours decimal minutes? Or is it the fraction of hours? The latter would make more sense when using a decimal.
Line 176: It is not immediately clear why an event has to last 48 hours?
Line 177: von der Lippe et al., (2026) showed that the threshold for cloud occurrence is closer to 3.75, which may be worth checking.
Line 180-181: Should there be an atmospheric transport component included in this criteria as well e.g.Lackner et al., (2023)?
Figure 3: ACLOUD RF11-13 do not appear to meet the WAA criteria defined above when looking at the climatological mean shown in panel c. Does this have something to do with the 48-hour mean? Either way, it does not look like any of the RF would meet the -10 K anomaly threshold. Similarly, when looking at RF7 and 14 during AFLUX and RF1 and 7 during HALO(AC3), it doesn’t look like the CAO threshold is met.
Figure 4: Are these dropsondes only over water? It looks like some of the CAO soundings would not produce a positive MCAO index, is this something to consider? Also are the dropsondes normalized to MSLP if yes, then should the plot go to 1012 hPa?
Line 210-212: Where is the EIS analysis? I don’t quite follow this here.
Line 259: please check the Korolev et al, (2011) reference as this may not be the correct the reference for the interarrival time algorithm description.
Line 274-279: These sentences need to be restructured as it is unclear what is being discussed here.
Lines 280-288 are also jumping around and it is hard to follow the logic, please reformulate to be more concise and structured.
Line 290-296: It is great to try and merge the PSDs but the thresholds for where the merging are a bit unclear. Using the CIP down to 33 µm doesn’t make sense, in reality it shouldn’t be used below ~75 µm or higher due to its 25 µm pixel size. In Table 1 it looks like the CIP was only used during ACLOUD but two resolutions are listed in the manuscript. Which resolution was used during this campaign? On a similar note, the 2DS can almost never sample a 1280 µm particle due to it completely covering the optics so it is not clear why the 2DS measurements would be used to such large sizes? Perhaps I don’t understand what is meant by weighted averaging here.
Figure 6: When looking at the merged PSD, it is clear that the weighting is not working optimally as the distribution has abnormal features (e.g. at 50 µm). It is not immediately clear why the PSD would not just follow the higher of the two OAP probes (once at least 4 or 5 pixels are occluded and the probe is measuring robustly) in the overlap region. This would likely lead to a more natural distribution as the probe reporting the higher concentration (once sizing well) would have the higher and more representative sample volume.
Line 330: mentioning dropsondes here is out of place when the subsequent sentence is about the turbulence measurements again
Line 357: so this asymmetry parameter is coming from the Polar Nephelometer? What size does this measure. So the LWC is only calculated from the CDP/CAS if the asymmetry is in a certain range otherwise it is considered IWC? To be honest, including this does not seem to add much unless it is used to flag the small particle phase and then classify it accordingly. But maybe I am missing the connection now.
Line 376: It is not expected that the CDNC increases with increasing height in the cloud. Does this mean that the cloud is CCN limited (e.g. Motos et al, 2022) and there is an increasing number of CCN with height?
Line 380: An LWC of 0.4 g/m3 seems quite high for such a shallow supercooled cloud and the spread in the values seems quite low. Looking back at the chart where WAA is identified, it looks like this is from two consecutive events/ days. Is the spread between the events really so small?
Line 398-399: Isn’t it just that at cloud top is where the coldest temperatures are observed and the most INPs can nucleate ice? Either way, the way the rest of the ice discussion is formulated is a bit confusing. Consider restricting to explain that Ni is not constant with height and actually smallest at cloud top so that the reader understands why the detectability is important to mention.
The comparison between Norm-W and WAA is not really saying much here, the ice number is the same, the vertical motion at cloud base is the same but the LWC is halved. So what causes this difference? Also, how many different profiles are used to make these plots, the spread in the values are extremely small considering there are multiple flight days and profiles.
Line 424-425: Can the crystal images check this?
Line 475-477: Is this difference at cloud top significant? It looks like even the uncertainty around the median values here are overlapping, so this might not actually be a difference at all. Either way, is it expected that the same profile in liquid properties would be expected under differing inversion strengths?
Line 479-483: Is the higher number just due to more inps capable of producing ice? Why would the Di decrease like this if there is plenty of liquid around? Are these profiles all from below freezing or “in cloud”? Are these differences actually significant?
Line 490-505: Again here, are these differences significant? This is a rather detailed and long description of what was observed for not much of a difference in cloud properties.
Line 511-525: Are the OO and SI measurements from the same flight? It is interesting that over OO where INPs are expected to be available etc there is no ice but only over the SI do you begin to see ice in the clouds. Temperature is mentioned but could it also be more active INPs from melt ponds e.g. (Mavis et al., 2026)? Seems like it would be hard for the INPs to get into the clouds though based on the low fluxes.
Line 534-551: Is it expected that such high LWC values would be observed over the SI? Also, that the Nw is higher over SI than OO is a bit counter intuitive. Is this a blowing snow ccn production situation (e.g. Gong et al., 2023) or how can this be explained?
Figure 13: Looking at the figure, it is quite surprising that such similar updrafts are observed between the two regimes (WAA and MCAO) and that SI typically sees higher values than over OO. Yet one would have expected that the higher fluxes over OO and during CAOs to produce stronger updrafts. Do the authors have an explanation for this? In the end, it is not clear if there is a statistical difference between updrafts ( or downdrafts for that matter) in these situations.
Line 564-566: With the previous comment in mind, it is not clear that such a statement can be said. These are very likely not significant differences and should be checked.
Figure 14: It would be worthwhile presenting how many particles are classified in each of the Zn and temperature bins for the two sizes investigated. I say this as it is quite surprising that > 2 mm particles are observed at cloud top, so it would be nice to know how many of these large particles are actually observed.
Line 624-628: Could add some references here as this is the expected evolution of snow morphology and growth mechanisms in mixed-phase clouds everywhere.
Line 657-658: Could point back to Figure 9c here to support this better
Line 669-670: This is not a new idea and has been known through the literature (e.g. Lohmann et al., 2016; Pruppacher and Klett, 1997; Rogers and Yau, 1989)
Section 4.2: It is very interesting that even though you have similar Nw and LWC over SI and OO the SI particles are much more pristine. Do you suspect this is due to the difference in fall speed/collision efficiency (e.g. Fig 13.d)? Otherwise, the parameters over OO are quite similar. This is worth discussing.
Line 728: Is Lackner et al., (2023) the right citation here? Consider using e.g. Williams et al., (2024)
Line 730-734: There are several studies discussing the role of available INPs over sea ice vs open ocean and the potential implications on cloud phase. It is interesting that there is very little difference in Ni over OO and SI. This would be worth discussing a bit here. (e.g. Carlsen and David, 2022; Creamean et al., 2022; DeMott et al., 2025; Raif et al., 2024).
Line 735-739: Consider mentioning some studies who have done this (e.g. Pasquier et al., 2023)
Line 748-753: It is not clear what is meant by the discussion here? Consider removing this or link it better.
Similarly, lines 754-760 are not well connected to what they are referring to and are rather vague here.
Line 780: Whether the differences in the regimes are significant or not needs to be addressed. Showing that the distributions medians are different doesn’t necessarily indicate that the underlying distributions are significantly different. Furthermore, the “significant” differences in the medians were often only for one or two of the variables investigated.
Line 784-788: It would be nice to describe an overview of the same variables for both WAA and MCAO, as now one is about liquid properties and the other about ice instead of contrasting them.
Line 806-809: Where is this shown?
Editorial:
Line 41: “up to” should be “down to”
Line 54, it is not clear what “it is” is referring to
Line 95: could add e.g. Geerts et al., (2022) or Zuidema et al., (2026)
Line 131-133: There is no need to mention HALO here
Line 148: Just limit to the measurements used, so just Polar 6 stats.
Figure 5: CBC and CA particle names are not in English
Figure 11 and 12 use different legend labels, consider making them the same.
Figure 14: the ytick locations/labels could be redone so it is clearer what the bin edges or middles are.
References:
Carlsen, T. and David, R. O.: Spaceborne Evidence That Ice-Nucleating Particles Influence High-Latitude Cloud Phase, Geophys. Res. Lett., 49, e2022GL098041, https://doi.org/10.1029/2022GL098041, 2022.
Creamean, J. M., Barry, K., Hill, T. C. J., Hume, C., DeMott, P. J., Shupe, M. D., Dahlke, S., Willmes, S., Schmale, J., Beck, I., Hoppe, C. J. M., Fong, A., Chamberlain, E., Bowman, J., Scharien, R., and Persson, O.: Annual cycle observations of aerosols capable of ice formation in central Arctic clouds, Nat. Commun., 13, 3537, https://doi.org/10.1038/s41467-022-31182-x, 2022.
DeMott, P. J., Swanson, B. E., Creamean, J. M., Tobo, Y., Hill, T. C. J., Barry, K. R., Beck, I. F., Frietas, G. P., Heslin-Rees, D., Lackner, C. P., Schmale, J., Krejci, R., Zieger, P., Geerts, B., and Kreidenweis, S. M.: Ice nucleating particle sources and transports between the Central and Southern Arctic regions during winter cold air outbreaks, Elem. Sci. Anthr., 13, 00063, https://doi.org/10.1525/elementa.2024.00063, 2025.
Geerts, B., Giangrande, S. E., McFarquhar, G. M., Xue, L., Abel, S. J., Comstock, J. M., Crewell, S., DeMott, P. J., Ebell, K., Field, P., Hill, T. C. J., Hunzinger, A., Jensen, M. P., Johnson, K. L., Juliano, T. W., Kollias, P., Kosovic, B., Lackner, C., Luke, E., Lüpkes, C., Matthews, A. A., Neggers, R., Ovchinnikov, M., Powers, H., Shupe, M. D., Spengler, T., Swanson, B. E., Tjernström, M., Theisen, A. K., Wales, N. A., Wang, Y., Wendisch, M., and Wu, P.: The COMBLE Campaign: A Study of Marine Boundary Layer Clouds in Arctic Cold-Air Outbreaks, Bull. Am. Meteorol. Soc., 103, E1371–E1389, https://doi.org/10.1175/BAMS-D-21-0044.1, 2022.
Gong, X., Zhang, J., Croft, B., Yang, X., Frey, M. M., Bergner, N., Chang, R. Y.-W., Creamean, J. M., Kuang, C., Martin, R. V., Ranjithkumar, A., Sedlacek, A. J., Uin, J., Willmes, S., Zawadowicz, M. A., Pierce, J. R., Shupe, M. D., Schmale, J., and Wang, J.: Arctic warming by abundant fine sea salt aerosols from blowing snow, Nat. Geosci., 16, 768–774, https://doi.org/10.1038/s41561-023-01254-8, 2023.
Griesche, H. J., Ohneiser, K., Seifert, P., Radenz, M., Engelmann, R., and Ansmann, A.: Contrasting ice formation in Arctic clouds: surface-coupled vs. surface-decoupled clouds, Atmospheric Chem. Phys., 21, 10357–10374, https://doi.org/10.5194/acp-21-10357-2021, 2021.
Kanji, Z. A., Ladino, L. A., Wex, H., Boose, Y., Burkert-Kohn, M., Cziczo, D. J., and Krämer, M.: Overview of Ice Nucleating Particles, Meteorol. Monogr., 58, 1.1-1.33, https://doi.org/10.1175/AMSMONOGRAPHS-D-16-0006.1, 2017.
Lackner, C. P., Geerts, B., Juliano, T. W., Xue, L., and Kosovic, B.: Vertical Structure of Clouds and Precipitation During Arctic Cold-Air Outbreaks and Warm-Air Intrusions: Observations From COMBLE, J. Geophys. Res. Atmospheres, 128, e2022JD038403, https://doi.org/10.1029/2022JD038403, 2023.
von der Lippe, F. S., Carlsen, T., Storelvmo, T., and David, R. O.: Shortening of the Arctic cold air outbreak season detected by a phenomenological machine learning approach, Atmospheric Chem. Phys., 26, 1565–1585, https://doi.org/10.5194/acp-26-1565-2026, 2026.
Lohmann, U., Lüönd, F., and Mahrt, F.: An Introduction to Clouds: From the Microscale to Climate, https://doi.org/10.1017/CBO9781139087513, 2016., 2016.
Mavis, C. E., Vazquez, M., Bekemeier, C. A., Barry, K. R., Chamberlain, E. J., Schmale, J., Jokinen, T., Laurila, T., Shupe, M. D., Hoppe, C. J. M., Fong, A. A., Oggier, M., Ulfsbo, A., Balmonte, J. P., Bowman, J. S., Hill, T. C. J., DeMott, P. J., Kreidenweis, S. M., and Creamean, J. M.: Meltwater as a Local Source of Ice Nucleating Particles in the Central Arctic Summer, Geophys. Res. Lett., 53, e2025GL118445, https://doi.org/10.1029/2025GL118445, 2026.
Pasquier, J. T., Henneberger, J., Korolev, A., Ramelli, F., Wieder, J., Lauber, A., Li, G., David, R. O., Carlsen, T., Gierens, R., Maturilli, M., and Lohmann, U.: Understanding the History of Two Complex Ice Crystal Habits Deduced From a Holographic Imager, Geophys. Res. Lett., 50, e2022GL100247, https://doi.org/10.1029/2022GL100247, 2023.
Pruppacher, H. R. and Klett, J. D.: Microphysics of Clouds and Precipitation, Kluwer Academic Publishers, 954 pp., 1997.
Raif, E. N., Barr, S. L., Tarn, M. D., McQuaid, J. B., Daily, M. I., Abel, S. J., Barrett, P. A., Bower, K. N., Field, P. R., Carslaw, K. S., and Murray, B. J.: High ice-nucleating particle concentrations associated with Arctic haze in springtime cold-air outbreaks, EGUsphere, 1–38, https://doi.org/10.5194/egusphere-2024-1502, 2024.
Rogers, R. R. and Yau, M. K.: A short course in cloud physics, 3rd ed., Pergamon Press, Oxford, 293 pp., 1989.
Williams, A. S., Dedrick, J. L., Russell, L. M., Tornow, F., Silber, I., Fridlind, A. M., Swanson, B., DeMott, P. J., Zieger, P., and Krejci, R.: Aerosol size distribution properties associated with cold-air outbreaks in the Norwegian Arctic, Atmospheric Chem. Phys., 24, 11791–11805, https://doi.org/10.5194/acp-24-11791-2024, 2024.
Zuidema, P., Geerts, B., McFarquhar, G., Bailey, A., Cassano, J., Doyle, J., French, J., Perkins, R., Petters, M., Snider, J. R., Wang, Y., Wang, Z., Cai, L., David, R., DeMott, P. J., Dzambo, A., Grasmick, C., Järvinen, E., Romashkin, P., Rosky, E., Ruchti, C., Tjernström, M., Tsai, P., Veres, P., Vömel, H., Woods, S., Wolff, C., Zietlow, D., Allison, J., Amundsen, N., Aquino, J., Arendt, C., Bansemer, A., Beamesderfer, E., Caldon, E., Campos, T., Carlsen, T., Chen, B., Cruikshank, O., Dewerd, C., Elsaesser, G., Ephraim, S., Goodstein, M., Gryskewicz, S., Haggerty, J., Hicks, J., Huang, Y., Juliano, T., Kille, N., Kreidenweis, S., Kruse, C., Lamsma, B., Loehrer, S., Lorenzo, G., Mahant, S., Majewski, A., Martrich, S., Mavis, C., McGinnity, F., McUmber, E., Newman, E., Niu, Q., Nguyen, L., Noer, G., Patil, S., Patnaude, R., Pruitt, E., Ragland, J., Reeves, M., Richter, A., Robertson, A., Rogge, F., Sarkar, M., Schnaiter, M., Seidl, A., Sodemann, H., Svensson, G., Smith, W., Stringer, S., Tatro, T., Tornow, F., Velt, P., Weissburg, E., Wu, J., Xue, L., and Xia, Z.: Flying into the Cold: First Results from the NSF Cold-Air outbreak Experiment in the Sub-ARctic (CAESAR), Bull. Am. Meteorol. Soc., 1, https://doi.org/10.1175/BAMS-D-25-0241.1, 2026.
Citation: https://doi.org/10.5194/egusphere-2026-3265-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 308 | 171 | 63 | 542 | 74 | 65 | 48 |
- HTML: 308
- PDF: 171
- XML: 63
- Total: 542
- Supplement: 74
- BibTeX: 65
- EndNote: 48
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General comments
The manuscript presents an interesting and comprehensive analysis of low-level Arctic mixed-phase clouds based on airborne in situ observations from four campaigns around Svalbard. In particular, the combination of microphysical vertical profiles, thermodynamic and turbulence measurements, and automated classification of a very large number of ice-particle images is valuable. The manuscript presents several interesting results concerning differences between MCAO and WAA conditions and between open-ocean and ice-covered surfaces.
Overall, I find the study suitable for ACP, but several aspects of the methodology and interpretation need to be clarified before publication.
The data and results are very interesting, but the robustness of some analyses is not always clear. In particular, the treatment of uncertainty could be strengthened, especially for the CNN-based habit classification and the derived morphological fractions. In several places, conclusions also appear broader than what can clearly be supported by the data and displayed uncertainties. Examples include the attribution of differences between strong and weak WAA inversions, the interpretation of individual features in the ice-habit distributions in Fig. 14, and the interpretation of morphological mass fractions despite the large habit-dependent uncertainty of the mass–dimension relationship.
Specific comments
1. Introduction
The Introduction provides a useful overview of the relevant processes and previous work, though I found it somewhat long and think it could be shortened.
L. 46: The sentence describing “moist-air entrainment driven by cloud-top radiative cooling” is not fully clear to me. Please clarify the physical process meant here, in particular why the entrained air is described as moist
L. 59/60: The statement that interactions between the liquid and ice phases include riming, the WBF process, and aggregation seems misleading, since aggregation itself is primarily an interaction between ice particles rather than between the liquid and the ice phase.
L. 88 vs 110: Please use the introduced acronyms consistently. In particular, “warm air advections” and “marine cold air outbreaks” should use the corresponding acronyms once defined. Please also check the consistency between WAI and WAA.
2. Data, methods and meteorological settings
One general concern regarding the combined dataset is the comparability of the different campaigns. The campaigns differ not only in season and meteorological conditions but also partly in instrumentation. For instance, the 2D-S and CIP are used in different campaigns and have different resolutions and classification thresholds. Please discuss more explicitly how these differences affect comparability and whether campaign, season, meteorological regime, surface conditions, and instrumentation may partly confound each other.
2.1 Airborne campaigns
I found Section 2.1 difficult to follow because the manuscript repeatedly moves between campaigns, meteorological conditions, surface conditions, and instrumentation.
An overview table including campaign period, aircraft, number of flights, in-cloud measurement time, main instruments, and dominant meteorological/surface conditions would substantially improve readability and would allow some of the descriptive text to be shortened. I also suggest reconsidering the organization of Sections 2.1–2.3 to reduce repetition.
2.4 Processing microphysical data
The processing methodology is generally described in detail, but several aspects of the CNN classification remain unclear to me. I also could not find a report of the uncertainty for the CNN classification here (see comments to Section 4 for more details).
L. 265: It is unclear how the additional manually classified images were incorporated into the CNN training. Was the existing model fine-tuned, or was it retrained from scratch? Please clarify the training procedure. This also ties in with the uncertainty reporting for the CNN classification. Were some of these images retained as an independent test set?
265–270: The habit classification is restricted to particles larger than 300 µm for the 2D-S, 450 µm for the CIP, and 2000 µm for the PIP. Please discuss more explicitly what information is lost by excluding smaller particles from the habit analysis and whether the different thresholds affect comparisons between campaigns.
L. 280–285: The manuscript states that particles smaller than 50 µm cannot be reliably separated by phase and are therefore all considered liquid. I understand the practical reason for this assumption, and the comparison with D’Alessandro et al. (2019) provides support for it. Nevertheless, I would like a discussion how this assumption affects the derived LWC, IWC, and phase partitioning, and whether this effect is included in the uncertainty assessment. Why are these particles assigned to the liquid phase rather than excluded from the phase-partitioning analysis?
Figure 5 caption: Please check whether Bazantay et al. (2025) is introduced appropriately before it appears in the caption.
2.6 Profile selection and normalized altitudes
The normalized-altitude approach is useful, but the determination of cloud boundaries is still not entirely clear to me.
L. 355: “Figure 7 illustrates the method used to identify the top and the base of the cloud from the aircraft observations.” I do not clearly see from Fig. 7 how the actual boundary-selection method works.
L. 355–363: Please provide a clearer and more reproducible description of how the final \(Z_b\) and \(Z_t\) are selected for an individual profile, particularly when LWC, asymmetry parameter, radar, and lidar do not indicate the same boundary.
3. Microphysical profiles
The profiles show interesting differences between the regimes. However, in several places I found the interpretation stronger than what I can clearly identify from the figures and displayed uncertainty ranges.
L. 474: “Figures 10 (a–d) show the associated vertical profiles of cloud microphysical properties.” Please add that panels (a–d) refer to WAA conditions.
3.2 Influence of temperature inversions
The influence of inversion strength is interesting, but other cloud properties also vary between the inversion categories.
Table 3: In the WAA cases, stronger inversions are also associated with lower cloud-top temperatures. Could part of the difference in the ice phase therefore result from the lower cloud-top temperature rather than inversion strength itself? Please discuss this possible confounding effect.
3.3 Sea ice influence
The comparison between sea ice and open ocean is interesting, but the strength of the interpretation should reflect what can actually be distinguished in the profiles.
Figure 11: Here Nw is shown, whereas Deff, w is not. Looking at the three quantities in parallel would help the interpretation of the liquid phase changes. It would be interesting to see how Deff, w behaves near cloud top where the LWC decreases. Similarly, also in Fig. 10, including Nw would provide interesting additional context.
L. 529–530: The sentence “This weak vertical increase suggests a limited supply…” is duplicated.
4. Ice crystal habits
This is one of the most interesting parts of the manuscript, but it is also where I miss an uncertainty analysis most. In particular, I could not find uncertainty estimates for the CNN-derived morphological fractions, making it difficult to assess how robust some of the reported peaks and differences are.
Please clarify how particles are identified as “precipitating” particles. If the classification is based primarily on size (>2 mm), “precipitation-sized particles” may be more precise.
4.1 Overview for all campaign
Given that the four campaigns cover different seasons, meteorological conditions, and partly different instrumentation, please discuss more explicitly how representative and comparable the combined distribution is.
Some statements in the description of Fig. 14 are difficult for me to identify in the figure itself.
L. 580: “for” → “of”.
610: The wording that compact particles may grow under “environmental conditions that limit the development of pristine ice crystals” is unclear to me and could be explained more explicitly.
L. 611: The statement that compact particles occur somewhat more frequently in the upper cloud layers does not appear very clear visually. It is also unclear whether this refers to small particles, large particles, or both.
L. 616: The two distinct peaks in rimed aggregates around −7.5 °C and −25 °C are not obvious to me from the figure. If these features are quantitatively robust, please make this clearer or consider softening the wording.
L. 620: The identification of spherical particles as drizzle droplets (see comment above to L. 280–285).
4.2 Comparison of ice crystal habits between MCAO and WAA regimes
The manuscript changes here from morphological number fractions mass fractions. I assume that the mass fractions are calculated using the mass–dimension relationship in Eq. (1) but stating this would make the analysis easier to follow.
More importantly though, I could not find an uncertainty analysis for the mass fractions. Given the earlier discussion of very large and habit-dependent uncertainty in the mass–dimension relationship, a clarification of how these uncertainties affect the interpretation of Figs. 15 and 16 would be helpful.
4.3 Surface influence during cold air outbreak
L. 691-693: In the discussion of Fig. 16, the statement that the fraction of rimed particles is approximately 30% lower over ice-covered surfaces appears somewhat general, as this difference seems strongest only over part of the vertical profile. Please consider making the wording more precise.
5. Discussion
The Discussion provides a useful comparison with previous work. Given the uncertainties and possible confounding factors discussed above, I suggest distinguishing more consistently between relationships directly demonstrated by the dataset and interpretations that are consistent with the observations.
6. Summary and conclusions
Several conclusions are stated more strongly here than in the corresponding Results sections. Short summary statements at the end of the main Results subsections would help connect the presented evidence more clearly to the final conclusions.
L. 791: “Stronger temperature inversions promote favorable conditions for cloud glaciation.” I do not think this conclusion is sufficiently clear from the presented results. There is a potential confounding effect from the lower cloud-top temperatures that occur together with the stronger inversions. In addition, the statement is more strongly worded here than in the Results section.
L. 786–788: High ice crystal number concentrations alone do not demonstrate riming or secondary ice production. In particular, secondary ice production does not appear to be established anywhere in the Results, so I am not sure why it is introduced here.
L. 796–799: The reported 50–70% rimed mass fraction during MCAOs is not obvious to me from Section 4.2, where values of approximately 50–60% are reported, while the higher values appear to occur in the open-ocean data.
L. 800/807: Statements such as “the underlying surface strongly influences” and “jointly controlled” appear somewhat stronger than the results support.
L. 800–803: The statement that stronger surface turbulent fluxes favor both riming and aggregation is not entirely clear to me from Fig. 16. The enhancement of rimed particles over open ocean is clear, but large aggregates also appear abundant over ice-covered surfaces. Please clarify which result supports the conclusion regarding aggregation.