The authors present an analysis of remote sensing retrievals that were collected during several marine cold-air outbreak (MCAO) flights over the Norwegian Sea. Retrievals are based on passive hyper-spectral and polarization measurements in the shortwave spectrum. Translated into a quasi-Lagrangian framework, the retrievals generally show a progressive deepening of the marine boundary layer and an increasing ice fraction with greater distance from the pack ice. The various MCAO strengths covered by the flights show more intense deepening and greater ice fractions for stronger MCAOs. The paper is generally well-written. However, there are a few concerns that the authors should address before the paper is published. I recommend returning the paper for major revisions.

Major concerns

Quality issues because of sea ice – In several instances the authors write that sea ice impacts the retrieved cloud properties, but it’s not 100% clear which data points in the figures are affected. I think the authors should mark suspicious values (e.g., using a different color) in all figures. Also, looking at case “20220404” in Fig. 1, there are substantial flight portions above sea ice, but lines in Fig. 5 and 6 are uninterrupted (and without artifacts); I’m not sure at which times and distances these portions are, but it would be helpful to highlight them.

Retrieval assumptions – The authors retrieve many cloud microphysical properties and it’s not 100% clear if retrieval assumptions are important to do so. For example, are these retrieval look-up-table (LUT) based and what were the assumed cloud vertical structures when generating the LUT (e.g., was liquid always assumed to be above frozen condensate)? The authors should clarify these structural assumptions in Section 2.1.  Furthermore, the authors use several thresholds (e.g., for cloud fraction, the watershed algorithm, and thermodynamic phase, etc.). The authors should quantify the sensitivity to these thresholds. For example, for cloud cover the modelling community often uses a cloud-optical depth > 2 or 2.5; which value was chosen here and would slightly different values substantially alter the results?

Value for the wider community – While cloud-top height and cloud cover are common properties that can be directly used in the modeling community, I’m less sure how to translate the other quantities and think the authors should discuss it. For example, ice index and ice fraction inform on various layers near the cloud-top; how would it be comparable to other measurements (e.g., in-situ cloud probes) or model output? Specifically for model output, would a forward simulator be needed?

Minor concerns

ll. 35-38 I’m not sure that this vertical structure applies to MCAO clouds that are rather convective.

ll. 40-41 I’m not sure how relevant the WBF process is inside MCAOs but would probably guess that riming is the dominant mixed-phase process.

l. 107 Do both instruments cover the same range?

l. 124 Are the results sensitive to this threshold (also see second major point).

Fig. 2: It would be useful to list the case date in the caption or somewhere in the figure.

Fig. 3 and 4: Maybe add uncertainty around each line (e.g. from percentiles).

l. 288 Cloud-top height in “20220401” does not appear that high?

Fig. 5: The cloud radius of “20220329” seems small. Is the mean value perhaps a poor representative here?

ll. 230-231 Related to the exclusion of high solar zenith angles: Do higher angles also mean less vertical penetration into the cloud? I think it would be good to provide approximate penetration depth for all cases (and all distances and times).

Typos

l. 244 Please check this sentence. Can uncertainty be negative and does the uncertainty have an uncertainty?

General

This manuscript analyzes observations and retrievals of cloud macrophysical and microphysical properties during the HALO-(AC)³ field campaign, focusing on their temporal and spatial variations. One case out of six is used to demonstrate detailed cloud evolution, while the statistical analysis shows the spread and general tendencies across all six cases. The presented dataset is valuable for improving our understanding of MCAO cloud structures and their evolution.

However, the selection of the demonstration case needs stronger justification. In addition, when comparing the aircraft-based findings with previous studies, especially satellite-based analyses, the manuscript should provide clearer explanations for why the results differ. Simply stating that satellite resolution is coarse or retrieval uncertainty is high is not sufficient to explain contrasting trends.

I also find that the statistical analysis section does not add much new information beyond what is already shown in the case study, except for illustrating the spread among cases. Based on these concerns, I recommend returning the manuscript to the authors for major revision before it can be considered for publication in ACP.

Major Comments

1. Case selection (Section 3.1)

Please justify the choice of the 20220401 case as the main demonstration of cloud macrophysical and microphysical evolution. Based on Figure 1, this case does not appear to be the most representative of a classic MCAO. In this case, cloud streets are oriented roughly parallel to the ice edge, implying continuous advection of fresh cold air from the ice to the open ocean. As a result, cloud properties are persistently influenced by cold, dry air.

A more typical MCAO would involve cold air flowing off the ice edge and clouds forming with roll structures oriented roughly perpendicular to the ice edge (e.g., 20220329). While the 20220401 case may still provide useful insights, its findings may lack generality. This might explain why the cloud fraction evolution in this case differs from the statistical behavior reported in Murray-Watson et al. (2023). In your Figure 5, the 20220401 case appears to be the only one with nearly constant cloud fraction during the first several hours, and it also has the shortest flight time over open ocean.

2. Comparison with previous studies

Both the case study and the statistical analysis are compared with previous work, especially Murray-Watson et al. (2023, hereafter MW23). However, it is not a fair comparison to contrast a single case with a statistical composite from many cases, particularly when the selected case is not a typical MCAO example.

For cloud fraction, you attribute differences between your results and MW23 to instrument resolution (L212–215). It is not clear how resolution alone would systematically change cloud fraction estimates and the temporal/spatial trend, and this needs clearer explanation. Later, you note that stronger MCAOs tend to have higher cloud fractions, consistent with MW23.

For cloud particle size, MW23 found smaller sizes for stronger MCAOs, which is opposite to your result. You attribute this partly to satellite retrieval uncertainties, but it is unclear why such uncertainties would specifically lead to smaller effective radii in stronger MCAOs. Wouldn’t retrieval uncertainty affect strong and weak MCAO cases similarly? This point needs more careful physical and methodological explanation.

3. Relationship between Sections 3.1 and 3.2

I understand that Section 3.1 presents a detailed case and Section 3.2 presents statistics. However, Section 3.2 does not seem to add much new information beyond Section 3.1, other than showing the large spread across cases. These two sections could potentially be merged, with individual cases discussed in the context of the statistics.

I briefly looked at your Part 2 manuscript on vertical cloud structure. Combining horizontal and vertical structures into a single comprehensive observational paper could be very valuable. As it stands, the current manuscript alone does not seem sufficient as a standalone paper. I will leave it to the authors to decide.

If you prefer to keep this as a separate paper, I suggest adding more discussion of the physical mechanisms behind the observed temporal and spatial evolution. Currently, the manuscript mostly presents observations, with limited interpretation. Alternatively, you could emphasize which of your findings are truly new compared to previous satellite or modeling studies.

Minor Comments

L50: can you elaborate briefly on why true Lagrangian observations are challenging.

L78: “aircraft” -> “aircrafts.”

L105: The full name of the instrument should be given at its first appearance.

Figure 2: What does “WGS84” on the y-axis of the left column mean? Has this been defined?

L208–209: You state that cloud fraction remains almost constant during the first hours. To me, the variability in the first half looks comparable to that in the second half, as seen in Figures 2b and 2e. Please clarify how this conclusion is reached.

L212–215: MW23 presents statistics from many MCAO cases, while you compare their results with a single case. I suggest avoiding direct comparison of one case with statistical composites. Also, while specMACS may provide more accurate cloud identification due to higher resolution, it may also have a narrower field of view than satellites, which could affect cloud fraction estimates.

Figure 4: There is a distance near the beginning where no data appear in the bottom row, while data are shown in the top row once the aircraft is over open ocean. Please clarify why this occurs.