The manuscript by Danker et al. examines phase partitioning in Southern Ocean low-level clouds and draws very interesting and potentially important conclusions regarding the relationships between phase and albedo and how phase responds to large-scale dynamics and cloud morphology.  These findings are derived through an innovative analysis of the so-called DARDAR cloud classification product combined with MODIS-derived albedo.  I find the paper to be very interesting, generally well written and well organized.  I recommend publication after minor revisions addressing the following concerns.

Major Comment: 

The authors base their analysis on a classification product that was designed to assist implementation of algorithms being developed for the EarthCare satellite.  It was adapted to the CloudSat and CALIPSO data sets augmented by thermodynamic information from ECMWF.  The authors of this paper have more or less adopted this classification scheme as the primary source of information for their analysis without, in my opinion, sufficient critical assessment of it for their specific purpose.  The DARDAR algorithm was developed to be applied globally and was not specifically tuned to the clean maritime environment of the SO where INP concentrations are very low.  Therefore, I think the authors should explain in their methods section 1) what actual information, beyond simple model-based temperature threshold relationships, exists at 60m resolution in the vertical profile of an optically thick cloud that is sampled by CloudSat to distinguish phase and 2) how such an algorithm can distinguish or not supercooled liquid drizzle from ice-phase precipitation? Based on Ceccaldi et al. (2013) Figure 1, the DARDAR algorithm uses radar reflectivity and thus the presence of precipitation-sized hydrometeors) along with how that reflectivity is distributed relative to wet-bulb temperature to prescribe phase in optically thick clouds.  Thus my third question: 3) To what extent has the DARDAR algorithm been validated in a clean maritime environment where there may not be sufficient INP to nucleate ice phase hydrometeors and where supercooled drizzle may be common in clouds that do not have sufficient updraft strength to initiate secondary ice processes?

Specific Questions Related to this Comment:

Line 109:  This logic to derive a column phase type seems reasonable.  However, the authors must explain what in the DARDAR algorithm distinguishes the liquid types from the mixed types?  What, in the actual data, is providing the information?  If it is the wet-bulb temperature threshold combined with radar reflectivity, then it is important to explain whether this threshold can identify the presence of liquid supercooled drizzle in optically thick closed cell stratocumulus. 

Line 115:  Supercooled drizzle has been observed in Southern Ocean stratocumulus clouds to temperatures of -25C. It seems as though the wet bulb threshold might have difficulty identifying the presence of supercooled drizzle.  Mace and Protat (2018 DOI: 10.1175/JAMC-D-17-0194.1) document supercooled drizzle occurrence with a cloud temperature near -10C.  See also Silber, I., Fridlind, A. M., Verlinde, J., Ackerman, A. S., Chen, Y.-S., Bromwich, D. H., et al. (2019). Persistent supercooled drizzle at temperatures below −25 observed at McMurdo Station, Antarctica. Journal of Geophysical Research: Atmospheres, 124, 10878–10895. https://doi.org/10.1029/2019JD030882

Line 151:  Again, what "signals"?  Where is the information coming from?  From Mace et al. (2021) CALIPSO is mostly unable to identify the presence of ice when it occurs in optically thick clouds.

Line 160:  The value of 0.5% is 20 times lower than 10% that is reported by Mulmenstadt.  Mace et al. (2021) analyzed cloudsat and calipso data between 2007 and 2010 and find that  ~25% of the SO MBL clouds are precipitating.  I am aware that Marchand and his student (paper in review) find a substantially larger precipitating fraction from data collected at Macquarie Island.  The magnitude of this discrepancy between what is found in this paper and what has been reported in the past is large and those differences have implications for our understanding of Southern Ocean climate.

Line 165:  Might it also be that optically thin clouds are much more likely to be categorized as liquid because they are optically thin and that optically thick layers are more likely to be classified as mixed because it is impossible to distinguish the phase of the precipitation that is in them?

Line 419:  Might it be that my third question above has been addressed by the authors’ comparison to the in situ data analysis of D’Alessandro et al. (2021) where the aircraft data show a lower occurrence of MPC than what is suggested in the DARDAR product?  I wonder if the authors should reconsider the possibility that the lower occurrence of MPC in the aircraft data may actually be indicative of the inability of the radar reflectivity-wet bulb relationship to distinguish the difference between supercooled liquid precipitation and ice-phase precipitation?

Minor Comments:

Line 87: The asymmetry parameter g is a function of the droplet size distribution (effective radius and width of the distribution).  How might g vary in realistic southern ocean clouds? 

Line 101: Since CloudSat's reported resolution is 240 m, it seems that there may not be much difference in what is known about phase or anything else between 720 and 780 m in an optically thick cloud. Also, the conservative rule of thumb is that clutter begins at 1 km.  However, this could be checked by examining the CloudSat cloud mask product to see if the Marchand et al., algorithm is identifying clutter or not at a particular height since the height where clutter begins varies by 1-2 (240 m) range bins around an orbit. 

Line 104:  I'm not sure I understand "orginal" in this context.  The actual vertical resolution of CloudSat is 480 m and it is oversampled twice to provide return in 240 m range bins.  The 60 m resolution of DARDAR is purely a construct and represents an extrapolation of the CloudSat data. That vertical resolution is a function of the CALIPSO lidar data. Information at 60 m resolution below the point where the lidar attenuates is likely not a function of the actual data but may be due entirely to the model-derived therodynamics and the DARDAR decision tree.   

Line 121:  Problems with the MODIS CTT are a potentially important piece of information for the community.  Can the authors elaborate, provide an example, or provide a reference if this problem has been previously reported?

Line 156:  What does "internally mixed" mean in this context?  If it means that the DARDAR algorithm is identifying the presence of ice and liquid, then, again, I ask where the information is coming from and can it distinguish supercooled liquid precipitation from ice-phase precipitation?

line 445:  Please define in-cloud albedo. Apologies if I missed this definition earlier.

With Compliments,

Jay Mace

Overall, this paper provides a clear and comprehensive description of its methodology, findings and implications. The study evaluates the evolution of clouds over the Southern Ocean, which is directly related to processes and feedbacks over the region that are in desperate need of improved understanding to improve weather and climate models. Results focus on long scale trends such as seasonality as well as trends on the timescales of days/weeks by separately evaluating open and closed cellular cloud regimes. Specifically, the authors evaluate whether cloud phase is correlated with the transition of closed to open cell regimes, which is certainly a pertinent question. The study is well cited and contains ample discussion of previous studies in relation to the study’s findings. I recommend the manuscript be accepted with minor revisions.

Comments:

The phase product is performed as a vertical integration. As I understand it, any combination of liquid/ice/mixed pixels will result in the classification of the column as mixed-phase (as long as there is at least one pixel containing liquid and/or ice). Can you comment on this in relation to potentially restricting liq<<(liq+ice) from being classified as mixed phase assuming both liq and ice are greater than 0? (e.g., for in situ observations, often mixed-phase is classified as 0.9>LWC/TWC>0.1, such as described in Korolev et al. (2017) mixed phase review paper).

You refer to albedo as reflectivity numerous times in the paper which can be confusing (especially since it’s a remote sensing paper). It might be best to just say albedo.

Line 19: “These differences in cloud albedo”*

Line 55-57: This sentence is confusing

Figure 2: Change y-axis label to just “Probability”

Figure 2 caption: Can you specify what category you are referring to? Are they normalized by cloud regime type and respective row variable? Then I don’t think you need to add the category comment.

Line 166: Where did you get 55% from? Did you calculate it by using all the COT values (including different seasons and cloud regime types) in table 1? If so, please specify.

Line 230: Introduce SLCs here (and not at line 238). Also, it’s slightly confusing to introduce this since you introduce SLC (the same acronym except clouds isn’t plural) at line 10.

Line 258-259: Why isn’t the peak at -5C observed for closed cell MCC?

Line 342-343: What part of your results suggest mixing is relevant to CTH? You just mentioned data limitations prevent you from evaluating turbulence/circulations and no significant trends were found between SST and sfc wind speed.

In this work, the authors use the multi-satellite DARDAR-MASK product together with a neural network identification of mesoscale-cellular convective clouds to characterise the relationship between cloud morphology (represented by the MCC type), cloud phase (from DARADR) and cloud radiative properties (calculated from the retrieved optical depth). They show that cloud phase depends more strongly on cloud depth than cloud top temperature and that cloud phase is also connected to the MCC type, leading to potential implications for cloud phase feedbacks.

This manuscript shows a number of clear results, supporting studies based on ground-based or insitu data that were primarily previously conducted in the northern hemisphere. I have a few minor points and suggested, but other than this, the paper is clearly in scope for ACP and I would recommend publication after these are addressed.

L47 - Are favored?

L78 - Some previous studies have only used DARDAR phase at the cloud top. How reliable is the phase deeper into the cloud where the lidar has attenuated (or does this not matter)?

L89 - does g change for ice or is the same value used for all clouds? The connection to optical depth would also presumably depend on the value assumed in the retrieval.

L95 - What is mix? My understanding is that it is a lidar backscatter peak along with a radar return. Is it possible that this is not actually mixed phase cloud, but perhaps precipitation (or just large liquid water droplets)?

L107 - Presumably Ice only is also a vertical 'phase' that occurs - or is that excluded?

L121 - Is there an example of this? I would have thought the MODIS CTT might be a better option, as at least for these low clouds, MODIS is actually observing the temperature, rather than reconstructing it from the p-T relationship in ERA5? Is the uncertainty perhaps due to cloud phase errors?

L165 - I am not sure what is going on here. The paragraph suggests that Mulmenstadt et al find more precipitating clouds the the current study, but also that they rarely find precipitating clouds (although more often than this study). Is it clear why these studies disagree given they both use very similar data)?

Fig. 3 - When the circles overlap (particularly on the top right), it can be difficult to see how they change. It also makes it difficult to determine how the mean phase fraction changes too, as the central points are black for both mixed and liquid.

L254 - 'could be related to' - you have the data to show if this is the case I think? This paragraph jumps about a bit between potential explanations and new results, it might be easer re-ordered slightly (although I leave that to the authors discretion).

L285 - the decrease in mixed fraction for the observed mixed fraction - this sentence is a bit confusing. It might be useful to be more explicit about which direction the mixed phase fraction is changing with temperature (and which bit you are considering here).

Fig. 4 - L291 states that there is not a higher mixed phase fraction is open than closed MCC clouds. This figure appears to demonstrate that the opposite is the case, and that there is a considerably higher mixed phase fraction in closed MCC, once latitude and CTT are accounted for (if I am reading it correctly)? Is it possible that Fig. 3 shows little difference because it is not stratifying by the correct variables?

Fig. 5 - Given you need several levels to identify a liquid phase top, but the cloud must have a CBH above 780m. Similarly, presumably at least one cloudsat layer outside the clutter is required - does that also set a minimum useful CTH? Might this contribute to the sharp shift in cloud phase for the thinnest clouds (CTH <1km)?

L371 - How much of this is due to different sampling of 'cloudy' pixels with the MODIS algorithm? Are enough cloud edge pixels discarded in the open cell regime to make a difference here (as presumably almost all closed cell pixels have a valid retrieval?)