The authors conducted large-eddy simulations to explore the impact of seeding on boundary layer supercooled clouds. The model setup is based on observations in the CLOUDLAB project. They first demonstrated the capability of the model to simulate and reproduce the seeding experiments at different environmental conditions. Then, they investigated the WBF process in the model by changing the INP emission rate. One conclusion is that the WBF process seems to be less efficient in the model than in the field. The conclusion is striking and interesting. One inconsistency is that the seeded cloud is expected to be above the site at 10:30 UTC (see Fig.6). However, in-situ measurements show that ice particles exist at around 10:35 UTC (see Fig. 9). So the apparent less efficient WBF process in the model might be due to some other reasons, e.g., underestimation of the growth time or advection time.

In general, the manuscript is well written and easy to read. I have some minor comments listed below.

Line 113: “both experiments are identical in their setup”. Since both seeding experiments are at the same location, is there any physical reason why there are two seeding experiments on that day? For example, I can understand S25-2, S25-2.5, S25-3 can test the impact of distance (growth time), but what about S26-2.5a and S26-2.5b? What can we learn from these two experiments?

Line 149: “The frequency of model output was set to 5 min”. The ice growth time is between 6 and 9 min (Table 1). Please comment on whether the relatively low output frequency would affect the comparison between observation and simulation.

Table 1: Please also add the seeding height in the table. It is difficult to accurately read the seeding height from Figure 4.

Line 178: “seeding particle emission rate”. Please add more justification of the choice of emission rate. For example, is it based on the estimation of the real seeding experiments, or is it chosen to match the ice number concentration. I find some discussions about it in the later part of the manuscript, but it is better to add some justifications here.

Line 235: “There is a good qualitative agreement between …” What is the scanning frequency of the radar? How does reflectivity from the scanning radar look like e.g., 5 min before and 5 min after 10:30 UTC? Can the radar observation show the impact of cloud seeding?

Line 296, 345 “(not shown)” is not accepted nowadays. Please consider adding the figure in the supplementary material or rephrase the sentence.

Line 296. “The ice crystal number concentrations are in good agreement (within +-0.3 cm-3) with observations in 4 out of 5 simulations.” What I see is that the simulated median ICNC is one order of magnitude smaller than the observation, while the maximum value is similar. Even if the median ICNC is 0 from the model, the uncertainty is still within 0.3 cm-3. So I think this statement is not accurate.

OVERALL COMMENTS:

Omanovic et al. evaluated the WBF process rate in their model by performing ice seeding simulation and comparing the simulation results with the observations. A few major issues may affect the value of this paper and need to be addressed before I can recommend this paper for publication:

- Assuming the main conclusion of this paper (that is, the WBF process rate in the model is much slower than those inferred by comparison with observation) is correct, is it merely a manifestation of the well-known fact that spherical ice assumption underestimates growth rate by vapor deposition?

- While the main conclusion is not surprising, the method leading to it is problematic: Given the uncertainties in simulated meteorology and microphysics and observations, it is concerning that the authors essentially attributed all discrepancies between the simulations and the observations (regarding cloud droplet depletion) to WBF. For example, the authors dismissed several discrepancies between the simulations and the observations as unimportant without sufficient evidence; they used Jan. 26 simulation as a testbed for seeding and compared that with observations from Jan. 25 without enough justification; they tuned the seeding rate to match observed ice number concentration, inviting compensating errors.

- The evidence and reasoning supporting the main conclusion are not very clear. See detailed comments below.

DETAILED COMMENTS:

L26: During WBF, it is the evaporation of liquid that maintains supersaturation w.r.t. ice.

L27: How is the ice crystal radius defined?

L28: What does "which" refer to?

L65: Why "on the statistical significance"?

L74: This is confusing. Did the cited works use a weather model or LES?

L85: "CLOUDLAB aims to ...": This is unclear. How to improve precipitation forecast skill by evaluating seeding method?

L107: Sorry if I missed this: is the seeding flare mounted on the UAV?

L135: "dynamically almost stable and persistent low stratus clouds": Did the authors mean the clouds reside in a stable layer?

L137: Sections 2.2 and 2.3: Please describe the ice properties used in the model/parameterization, including size ~ fall velocity relationship, density, size distribution, ice shape, and so on.

L140: "80 vertical levels": What is the vertical grid spacing around the cloud layer?

L149: Schmale et al. (2018) reported CCN concentrations at many sites with different characteristics and seasonality. Which site and season did the authors use as a reference?

L151: "highlight the impact of growth time on the seeding plume": This is confusing; please clarify.

L156: "utilize the simulation of the 26 January 2023 for both seeding days": This is unclear; please clarify.

L168: "given the similar response in freezing for the larger particle sizes": But 400 nm is an upper bound for both Henneberger's field experiment and Marcolli's lab measurements. What about smaller particles?

L170: Does the probability of freezing depend on time?

L187: "The seeding plume was defined ..." If the seeding plume is defined by a threshold, why did the authors still need the unseeded simulation as the background?

L191: Again, do AgI particles freeze immediately after the release? If not, the time between the release and the arrival is not the growth time.

L194: There are a few issues in Section 3.1. First, both Figures 3 and 4 showed that the temperature range where the seeding occurred was suitable for secondary ice production. Is this process parameterized in the model? Does it have an impact on the results? Second, the use of Jan. 26 as a testbed for Jan. 25 and then comparing the simulation results with Jan. 25 observations is questionable. Even though the seeding height is adjusted to match what actually occurred on Jan. 26, the meteorology for Jan. 26 seems to be different from Jan. 25 (cloud temperature range, thermodynamic profiles, maybe also liquid water content profile, etc.) Please carefully justify this decision.

L209: "predicted cloud cover": How did the authors define cloud cover from radar observations and simulations?

L210: "a long-lasting low cloud that reaches similar cloud top heights as observed": But previously the authors said that the lower inversion base led to a lower cloud top.

L210: I probably missed it, but would you please described seeding height in numbers in the text or a table or as annotation in Figure 4 in addition to showing triangles in Figure 4?

L220: The contour for t7 in the top view is inconsistent with the one in the cross section. Does the cross section go through the center of the plume?

L228: "Note that the ice crystal number concentrations are spread ...": What is the difference between the ice crystal spreading out in this sentence and the previously described plume spreading out?

L233: "from left to right and back": This is unclear; please clarify.

L235: How is radar signal simulated? Did the authors use a radar emulator? Please describe.

L235: "a good qualitative agreement": The observations and simulations look quite different. It is unclear what features the authors referred to. Please clarify.

L246: "the model configuration can be used to conduct seeding experiments for further investigation of the ice crystal growth inside the cloud": Based on Table 2, the median ice number concentration in the model is way lower than in the observation. Why? And what are the impacts of this discrepancy on the seeding experiments?

L248: What is "the environment" here?

L251: In this paragraph, the authors listed a few factors influencing the ice number concentration. How did they attribute the discrepancy or consistency between modeled and observed ice number concentration to one or all of these factors?

L255: "The model, however, fails to reproduce ...": Wasn't seeding rate tuned to match observed number concentrations?

L256: "due to the aerosol concentration being adapted to the simulations S26-2.5a/b." Lost; please clarify

L278: "emphasizing the high efficiency of the WBF process": Is it possible to be secondary ice production?

L284: "Both processes (WBF and riming) are parameterized in the model": Does WBF require a dedicated parameterization?

L285: There are a few issues in this paragraph. First, the authors used Figure 9c as evidence that the model was not able to capture the observed cloud droplet depletion. However, in Figures 9a and 9b, there are no cloud droplet depletion and high ice number concentration around 10:30. What is the exact time period that the observations shown in Figure 9c come from? For this type of comparison, is the small sampling volume by HOLIMO suitable? Are the model data in Figure 9c from ice particle plume in the model domain? Please clarify. Is it possible that there is simply a mismatch between the observed and modeled arrival times? Second, "This discrepancy may originate ...": This is too much speculation. Did the authors perform any sensitivity simulations to prove? Are there any alternative explanations? Third, in Figures 9d and 9e, the authors commented that the model was performing well. Doesn't this suggest the WBF is doing fine? Why does this "further points to the fact that the WBF process in the model in its current form is not efficient enough"? Fourth, please show cloud water content/path, cloud droplet size distributions from observations and model for better comparison.

L293: This paragraph is confusing. It seems that the authors were saying the model performs well regarding capturing observed cloud droplet reduction (starting from "Furthermore, ..." in L298), which contradicts previous conclusions.

L314: What is the difference between these two conditions? Did the authors mean (0 < w < w*) and (0 > w > w')?

L317: There are a few issues regarding the results in Figure 10. First, please clarify the data points going into the third and fourth rows (ice and liquid size distributions). In particular, is the integral of the area below each distribution the same as the total number of grid boxes in the plume? Second, due to the sedimentation of ice particles, the ice and liquid size distributions are for different particles and cannot be directly linked. Third, it seems that the main result is that at t2, (1) the observed ice particles are bigger than those in the simulations and (2) the observed liquid droplets are smaller. If one believes these two facts are linked, it is consistent with a weak WBF. But there could be many reasons that these two discrepancies are caused by different factors. Why can it be attributed to WBF? Fourth, if the main indicator of WBF is cloud droplet depletion, then all the data points contributing to the distributions in the fourth row are from grids that are less or not affected by WBF and these distributions do not support the argument anyway. Fifth, how do the liquid size distributions in the plume compare with background size distributions in the model? Sixth, the first two rows are interesting. Is there any dynamical factor that could lead to ice being too small and liquid too large? Like, does the vertical velocity distribution compare well with the observations? Does the release of the latent heat from seeding create its own circulation? This may affect the "background" cloud properties. There are some papers on this, for both marine stratocumuli seeded by ship emissions and mixed-phase clouds or supercooled liquid clouds seeded by ice, IIRC. Seventh, it seems that the distributions of the updraft vs the downdraft in the second row are inconsistent with those inferred from the first row (i.e., if one naively assumes the line between WBF_up and WBF_down separates the updraft and the downdraft). Is it simply because the vertical velocity in the cross section is not representative of the whole plume volume? Please clarify.

L334: Section 3.3.1: It is well-established that spherical ice particles do not grow fast enough, compared with ice particles with extreme habits. In the temperature range during the two days, the ice particles are likely to be needles/columns. Testing the effect of ice habits is probably more meaningful than increasing seeding rate by brute force.

L343: "This sensitivity analysis also ...": Confused by this sentence; please clarify.

L345: "The default setup may still be ...": Confused by this sentence; please clarify.

L348: "Hence, if the model were colder, we would see a higher ice crystal number concentration.": This contradicts previous statement.

L360: "The observed seeding temperature was nevertheless eventually reached": This is unclear. How was it eventually reached?

L364: If the dilution for Jan. 25 cases (2 to 3 km) is so different from Jan. 26 cases (2.5 km), does it mean one cannot use Jan. 26 as a testbed for seeding on Jan. 25 and compare the results with Jan. 25 observations?

L367: "aerosol concentration": See comments for L256.

TECHNICAL ISSUES:

Some of the comments below may reflect a personal preference in style. Feel free to ignore.

L24: Ice should be sublimating, not evaporating

L24: "The second case ..." The sentence is not well-constructed. Please revise.

L50: AgI does not consist of discrete molecules.

L316: "the serve": "they serve"?

Figure 10: Maybe better to refer to the panels with two-digit labels, one for rows and one for columns. For example, b1 for second row and t1.

L347: "stron temperature dependence": "strong"?