This study uses LES to simulate idealized mixed-phase clouds in a relatively clean Arctic boundary layer environment. Sensitivity tests are conducted to explore the impact of higher tropospheric aerosol concentrations to cloud properties. Aerosols in the boundary layer is kept 20 per mg while it is increased up to 50x in the free troposphere. Results show that entrained aerosol from the free troposphere can suppress precipitation and help to sustain clouds for longer time. I think the simulation results are clear and make sense to me. My major concern is how those simulations are relevant to real clouds in the atmosphere.

I’m skeptical about the relatively high droplet number concentration at the cloud top from the simulations (Figure 6). Are those simulations realistic? Therefore, I read some MOSAiC-related papers cited in the manuscript to learn more about the observations. I might miss some other papers, but I do not find observational papers to support model setups and conclusions in this study. I recommend the authors to add more observations to justify their model setup and/or conclusions. Some specific comments are listed below:

1. Figure 1 is the only figure related to the observation. I have several questions: do all those days have clouds? Where is the cloud layer on each day? What about the profile of large aerosol particles (those can contribute to droplet formation), instead of all aerosols larger than 12 nm? I recommend the authors plot similar figures like Figures 6, 13, 14 in Lonardi et al. (2022), for all cases chosen in this study. The model setup and conclusions would be more convincing if the authors can show observational evidence of the existence of aerosol layers above the cloud layer, and/or the potential impact of aerosols in the free troposphere to clouds in the boundary layer.

2. Based on Lonardi et al. (2022) Figure 6, clouds on July 23 and July 24 are all above the top of the boundary layer. Therefore, I think Figure 1 might be misinterpreted by the readers that clouds are in the boundary layer and they are affected by aerosols above during the MOSAiC campaign.

3. The authors said that "Here we extend the analysis presented by Lonardi et al. (2022)..." However, based on Lonardi et al. (2022) Figure 6, the large temperature inversion on July 23 and July 24 are at about 600 m and 900 m, respectively. However, in Figure 1 of this study, the top of the boundary layer on these two days are at about 300 m and 100 m. Please explain why they are so different.

4. Figure 2. Are the initial profiles of potential temperature and relative humidity based on the observation durign the MOSAiC campaign? Based on Lonardi et al. (2022) and also Figure 1 in this paper, surface temperature is very close to 0 C and the cloud temperature is just slightly below 0 C. But the initial temerapture profile for simulations in this study is much lower. Please justify the model setup.

5. Please provide formules of the profiles such that simulations can be rerun by others. What about the initial wind profiles? Do you nudge those profiles?

6. Figure 6. Is there any observational evidence to show that cloud droplet number concentration is maximum near the cloud top?

7. Page 4, Line 102, what is the CCN size distribution in the model? Is it fitted based on observations? I think results are sensitive to the CCN distribution. If the authors do not test its sensitivity, it should be clearly stated.

8. Entrainment rate is critical to bring aerosols from the free troposphere to the boundary layer. It would be nice to plot the time series of entrainment rate for different cases.

9. Page 9, Line 180: “This is indicative of a decrease in the amount of aerosol being entrained into the cloud…” Do you know the amount of aerosols entrained from the free troposphere as a function of time?

10. Page 10, Line 215: “Figure 7b shows domain-average time series of \sigma_w^2”. People usually calculate \sigma_w in the boundary layer where turbulence is vigorous. Just want to make sure \sigma_w is averaged in the whole domain or just in the boundary layer? It makes more sense to only average in the boundary layer. It would be better to show the \sigma_w profile.

Other comments:

1) Page 7, Line 127: delete “is”

2) Page 7, Line 130; Page 11 Line 219: “not shown” is not acceptable. Please consider adding it in the main text or supplement.

3) Page 12, Line 230: “Mauritsen et al. (2011) also found that…” This sentence is not clear to me. Please check.

4) Page 12, Line 246: “In these simulations, the boundary layer is …” This sentence is not clear to me. Please check.

This paper describes a hypothetical numerical Large Eddy Simulation, to test the hypothesis that Arctic clouds are sensitive to entrainment of free troposphere aerosols from remote sources. There is nothing fundamentally wrong with the paper; I just can’t help wonder why this study was perceived, why now, and what new we can learn from it? There are a lot of detailed results that seems a bit in lack of a plan how to use, and I think the study needs to develop before it can be published; hence “major revision”.

Main concerns

The study essentially falls a part into three stories: 1) one that deals with the entrainment of aerosols from the free troposphere into the boundary layer; 2) a second that deals with the effects of those aerosols on the liquid water clouds in this boundary layer, and; 3) third, the results, as a consequence of 1 & 2, on the boundary-layer structure.

For all parts, one problem is the case and the LES. There are no attempts to illustrate or ascertain that this LES in this setup is capable to simulate this cloud and that the processes that makes it tick are adequatly modeled. The case itself looks a lot like a midlatitude marine stratocumulus but is placed at a specific date and specific location in the Arctic. But there is no information on how the case was designed and why these choices were made. The results are dependent on the rate of entrainment, which is dependent on the model formulations, the setup (e.g. subsidence) and the resolution etc. If the entrainment is sensitive to for example resolution, then all the results here will also be.

And what is new here? The second author already a few years ago showed, using another LES but on a well-documented real observed case, that free troposphere aerosols are indeed entrained into the PBL where they have an effect on the clouds. In that paper, as also in this, one main conclusion was that aerosol observations near the surface does not necessarily provide any guidance as to what goes on in the cloud layer, especially if this is dynamically decoupled. So what does this paper add that wasn't shown already in the previous paper? And why should aerosols not be entrained? At the interface of a turbulent boundary layer to a laminar free troposphere there is always entrainment. How much may depend on how vigorous the turbulence is and how well resolved the stably stratified interface is resolved.

So, what is new here? Sure, there are more runs here and more details but how are those used to shed more light on the delicate balances at the cloud top? It looks to me that the synoptic scale divergence prescribed has a large effect on the result; maybe there should have been less runs with different aerosol settings and more runs with different subsidence? And why was not inversion strength varied or the cloud bulk features(?); none of the simulated clouds are very dense. So, the results show that entrainment of free troposphere aerosols happens (provided there is any right at the interface), but any boundary-layer meteorologist could have told you that! What else did we learn?

Once inside the boundary layer these aerosols have more or less the expected effects. Ice is there but IN is so small that ice plays no important role to the dynamics here. So why have it there? Solar radiation is also present, but so little it (probably) do not have an effect. So why have it there at all? More free troposphere aerosol results in more CCN in the boundary layer – obviously – and this results in denser clouds (larger LWP) with smaller and more numerous droplets. And this in turn leads to more cloud top buoyancy and more mixing; Nothing spectacular here; mostly the expected effects. But lots of details that may – or may not – reveal something. Problem is those are never explored in any detail; the analysis is unimaginative and the result is rather boring. Probably right, at least in principle, but still rather boring. With this effort in numerical simulation, there just has to be something more you can do.

Minor comments:

Line 8: That the cloud top does not move is controlled by subsidence. In reality this mean that the the cloud does move upward by mixing, but that i is immediately advected back down again by synoptic-scale advection. The bulk result may be close to net zero, but mixing and advection are not the same thing.

Line 15: Clouds are indeed important for Arctic climate but I have seen no study that shows them to have a large effect on “Arctic amplification”. If these authors have, I’d like to see a reference here.

Lines 20-21: A reference should be inserted here, as this statement comes from one single study (SHEBA) and most other summer studies I have seen does not show this warming regime, unless the ice has melted completely.

Line 45: “argued” is better than “found”; there is no way in which Matt could have proven this, but it is a good argument.

Line 64: What “scale” is that? LES is a numerical technique and is appropriate for flows with large enough eddies but there is no specific scale; it all depends on the problem.

Lines 73-74: So what happens with coalescence? Each droplet forms on one CCN each, but after having collided, forming a larger drop, there is no way the new aerosol formed by evaporation of that drop can be assigned back to the original CCNs.

Line 69: I’d like to know more about the aerosols scheme. How is formation of new aerosols handled and how are aerosols that become CCN or are entrained replaced? How is size distribution consndered?

Line 81-83: This is a different motivation to that given earlier, and does not rely on assumptions on entrainment.

Line 84: Is this radiation code sensitive to effective droplet radius?

Line 86-87: Setting the heat fluxes to zero is a reasonable assumtions; they are never very large anyway. But what about the momentum flux? Is this a free-slip simulation? How is this “reasonable” for ice surfaces?

Line 94: It looks like the cloud top is decreasing, at least for the less vigorous boundary layers. So, how was this value selected and how is the LES upper boundary and the magnitude of the entrainment affected? Is it a special target to have a constant cloud top constant, and should that then not require different divergence? And how would that affect entrainment?

Line 102: In the whole intro with references and all, the units for aerosol concentreations is cm^-3. Here all of a sudden, in the LES, it is mg^-1. Why?

Line 105-106: Since IN and ice processes are hardly considered, why even bother with IN? This is hardly a prototypical mixed-phase cloud as it is.

Line 156: Delete space between “salt” and “20”.

Line 165: Do you mean “dropets” and not “aerosols” here?

Line 176: What “cloud radius”?

Line 179: What “cloud number concentration”?

Lines 198-199: Why choose such an awkward date? While it is true that aerosol concentrations are low in autumn, this is a hypothetical LES and it would not be more or less hypothetical with the sun switched off completely.

Line 200: The flux divergence cannot have the unit “W m^-2”.