In the paper "Prognostic simulations of mixed-phase clouds with model 1D-AC v1.0: The impact of freezing parameterizations on ice crystal budgets" by Sun et al. 2025 the authors present an idealized simulation of an Arctic case with a 1D model, where they investigate the role of different freezing parameterizations, aerosol types, size distribution, and cloud parameters. An extensive set of sensitivity studies is presented. The results are well-described (with the limitations mentioned below), and the outcome of the study is interesting and relevant for the community. The study is well-suited for GMD, and I suggest publishing it after major revisions.  

Major comments:

I should first apologize that I did not read the preceding study of Knopf et al., which might explain some of my struggles. However, the submitted article should explain the major concepts to be a stand-alone article. 

1.) My biggest issue with the study is that I have some conceptual problems with understanding how the time-dependency was done for the CNT parameterization scheme (for me, right now it seems not correct how the implementation for the CNT scheme is done, but that could be related to some missing information or misunderstanding on my side). The following points need more clarification in order to be able to understand the study:

- CNT is a time-dependent scheme, but nowhere in the manuscript is it clearly explained how this time-dependency is reflected in the model (if there is a time-tracking or time counting for each INP etc.). 

- It is assumed that the whole aerosol distribution is effectively available INPs (INP reservoir) at the start (e.g. line 281-284 and line 481-482). That makes only sense if the aerosol particles had been exposed to the conditions in the cloud for a longer time (= the case of a long-existing cloud is studied). I wonder after how much time the whole aerosol distribution is activated for each aerosol species, and if this assumption is realistic? Can you please add calculations for each aerosol type and some representative/average temperatures of the case how long it takes to activate the whole aerosol distribution?

- The CNT formulas described in the manuscript (line 944) seem strange. Why is it not N_INP^CNT=N_aer * J_het * delta t ? From your framework here it is not obvious at all where the aerosol-specific nucleation rates are used. In Appendix C it is explained how the heterogeneous nucleation rates are used, but I did not fully understand it and I think this should be clear in the main text.

- In line 522, it says that for the CNT based approach, the entrainment directly adds to the total aerosol particle production = INP reservoir. How can that be true? The entrained aerosols are "fresh" in the cloud (so starting at timestep 0) and therefore have not been exposed to the conditions for such a long time that all are activated as INPs (if not, the calculation requestion above shows times below 10 s). Otherwise, the time would need to be taken into account. How is the time dependence dealt with in case of the entrained aerosols for the CNT approach - how is the time-exposure tracked?

I would suggest adding some explanations to the basic concepts used for CNT (including the requested calculations). It should also be highlighted that the case is reflecting a long-lived cloud, and the calculations are not done at the initialisation of the cloud but at a later stage. It could help to add a sketch of how the CNT scheme works.

2.) It seems that the basis of the different parameterization types (INN, INAS, CNT) is different, i.e. they don't use the same dataset to derive aerosol-specific parameters etc.. This is probably a necessity given the availability and nature of different INP measurements and judging S6 the slope of the specific types is similar, which means that the used dataset had similar freezing characteristics. However, discussion on this aspect is totally missing in the article. Especially for the mixed external aerosol population, it could have implications that the parameterizations can differ a bit from each other for specific aerosol types. 

Besides, aerosols like mineral dust are as such poorly defined. There are many different types of dust, and they come with a variety of ice nucleation activities. That should at least be mentioned in the manuscript if not critically discussed.

3.) Since the parameterizations are all based on measurements, their range can vary, which is also mentioned in lines 147-149. However, it is not really discussed anywhere if the temperature range of the parameterizations (derived from the temperature range of the measurements if no extrapolation is used) fits the case. What is the temperature range of the observations the parameterizations are based on? I would suggest adding that in S6 (and add this information and discussion to the manuscript). In case a parameterization relies on extrapolation of the dataset, is that justified?

4.) This study is highly idealised (for example, by not having any precipitation mechanisms etc.). That is no problem at all. However, I find the context of an Arctic case a bit confusing, or I don't really see a big meaning in that focus. I would suggest to rather reduce this to being an idealised study with a certain input of meteorology and aerosols (which happens to be an Arctic case). 

If the Arctic context is kept, I am missing some information that should be added. That is:

- Was the used SHEBA case a spring or autumn case (the temperature seems to be a bit low for summer)? A reference to the used case (radiosonde profile etc.) should be added for more clarity.

- How were the observationally constrained LES results used, and what exactly?

- Was the LES simulation compared to observations and produce a realistic cloud case?

Additionally to that, I think more limitations of the idealised setup should be discussed - is the relative importance of freezing a trustworthy result or could it be related to the idealised setup (or is there no way to answer that question). One example here that should be mentioned/more clearly discussed is not having a WBF process or any precipitation mechanisms.

5.) There is a comparison and critical discussion in relation to previous studies missing - this is not the first study comparing a CNT-based parameterization scheme with other schemes (and/or different aerosol types). What is different about the findings from this study from other studies (or are there similar findings)?

6.) In the summary and conclusion, it should be explained and discussed more what the consequences are of the findings of this study, especially in hindsight of implementing the CNT scheme into different models. How could this implementation look for less-idealised models (global models are mentioned, but how to treat the time aspect is not captured)? I think this part is crucial since it is a paper for GMD, where the model development and specifics should be of interest.

Minor comments:

- Title of the paper: A large part of the paper is on the impact of different aerosol types on ice crystal budgets (and not only the freezing parameterization), as well as cloud parameters. That should be reflected in the title.

- Citations in the introduction: often long lists but still not comprehensive, so please add "e.g." in front of the citation list. 

- Citations in the introduction: rarely are primary sources cited, but newer articles/reviews etc.. 

- Introduction: I would appreciate more explanation of how immersion freezing works under sub-saturated conditions, since no particles are immersed in that case?

- line 63: CNT does not use the concept of INAS, but still assumes some aerosol-specific quantity (often expressed as a contact angle) that is a similar concept.

- line 66: Not only constant supersaturation but also a constant temperature would be needed to activate more INP with time (in case of a temperature increase, other conditions would apply).

- line 111: The term recycling could be added here for clarity.

- line 120-122: It is true that laboratory-based parameterisations are limited by the temperature range of the measurements/instruments. However, this is true for all three parameterization types used in this study (even the CNT parameterizations use aerosol-specific information based on laboratory data) - it seems a bit strange to only mention that in the context of deterministic parameterization schemes.

- line 122-123: The last part of the sentence (slowly continuing ide formation) is not clear to me, please split the sentence and elaborate.

- line 138: Liquid phase as a fixed quantity- what does that mean? Constant and not changing by WBF etc., because not represented in the model? What is the limitation of this assumption? (see also major comment 4)

- line 247: Does that mean that the whole aerosol distribution is assumed to be activated/in cloud droplets?

- line 251: You could add one line of explanation here why/how these arrays are needed.

- line 463: What determines the quasi-stable plateau for the CNT param? Why is there a plateau?

- line 473: The aerosol composition is also important for the CNT parameterization.

- line 482: Does subsequent mean recycled (sublimated) and entrained aerosols? How is the time aspect taken into account there (see also major comment 1)?

- line 501: Is it not a colder subset of the total aerosol population?

- line 532-533: Why do the schemes react differently on sedimentation (or if not, why is it written/discussed this way)?

- line 543: Is it not the lower half of the cloud - it looks like the peak is there?

- line 584: "more sensitive" compared to? (all the other schemes?)

- line 537: Since the parameterizations were not derived from the same dataset the aerosol type and chosen parameterization are connected and can not be easily separated?

- line 735: Vast but also time-dependent INP reservoir (connecting to major comment 1).

- line 856: The first sentence is the same as 3.) before?

- line 880: I don't understand the comment about the computational simplicity - it is not simple if the time that aerosol particles are experiencing certain conditions has to be tracked to capture the time dependence?

- line 970: But only settling/sedimentation is accounted for in terms of vertical distribution (no updraft etc.)?

- Figure 1 vs. Table 1: From the table, it looks like dust and organic aerosol are missing the largest mode, but from the values and figure 1 it is clear it is missing the largest mode - leave D1 empty instead of D3 (use the same index for the same mode for all aerosols).

- Figure 3 and many others: the dashed style type used for the entrainment rate (or settling velocity) is very hard to see on a printout (and in the pdf). I appreciate the thought of having one line type for each sensitivity type, but would still recommend having the cooling rate dash for all sensitivities for better visibility.

- Figures 6, 7, 8: It would be helpful to have the parameterization name next to the label (a), (b)... of the plot. 

- Figures 6 and 7: Are there lines missing? Not all plots show all of the sensitivity studies? Or is that because the lines are lying on top of each other? This has to be fixed or explained/mentioned.

- Figures 6 and 7: Use the same scale for all four subplots?

- Figure 6: The red color looks more like brown on the print-out.

- Figure 6: What does the red line with value 0 mean physically? Mention.

- Figure 8: Some lines are on the y-axis and hard to see.

- Figure 8: Since the order of colors is different, does that mean that the schemes are inconsistent between the parameterization schemes (see also major comment 2)?

- Figure 9 (figure and caption): External = Externally mixed?

- Table 2 (and related text): What does the mixing time scale mean/do? Move this information to the Appendix, where the mixing time is mentioned.

- Table 2 (and related text): The text around the entraiment is not very clear. What is it that is entrained? From the discussion etc. I can read that it is referring to particles (and not water vapour, which would be an alternative interpretation, or both), but how many particles are entrained etc.?

- Table 2 (and related text): The value of the sedimentation rate does refer to approx. which ice crystal size/habit?

- Line 397: It makes sense that the depletion of the INP reservoir is seen in the ice production rates, but I don't understand why the cooling rate is not seen at all in the control (it is getting colder/more particles are activated).

- Table 7: Better plot N_i/N_i^(10 s)? The table is difficult to comprehend.

- Table 8: Why is the ice crystal production const. 0 in INN_CTRL - is the cooling rate too small to lead to an activation of more aerosols?

Technical corrections:

- Units: There are sometimes line breaks in between units or between units and numbers. Use protected space to avoid this.

- line 85: The sentence is incomplete.

- line 69-89: It would be better readable if written as a list or with a line break for each aerosol type.

- line 142: Incomplete sentence (or what does the "and statistically" refer to?) (?).

- Figure 1 (c): it should be Large Accumulation in the legend instead of SSA.

- Figure 1: Add the variable D in the x-axis label.

- line 239: Was that noted before?

- Table 4: Make categories (Control Run...) bold or emphasize a different way.

- line 320: Is the comma correct here? The sentence can be read a bit wrong because of that.

- Figure 3: The x-axis is cut from the left panel.

- line 867: Switch . with :.

- line 875: Typo, it should be "freezing".

- line 910: Does it not have to be <?

- S7 figure caption: Typo, should be "N_aer x 0.1".