This work compares the performance of different WRF simulation configurations on the accurate depiction of supercooled liquid water by comparing a particular event to flight observations. Overall, the manuscript is well written, it presents the methods and results clearly, and gives a good and thorough description of the model limitations that may explain the biases in the resulting performance. It seems like a novel simulation configuration using CAMS data is performed, thus its benchmarking is the main contribution of the study. The manuscript could benefit from a better explanation of its novelty and proposing future work to overcome the reported biases of the new approach.

Minor comments

-Please revise the title, as the core of the manuscript does not reflect its focus

- The abstract does not clearly explain which simulation configuration is novel, and it also does not clearly explain which are clean and polluted conditions. Also, it is not clear if the synoptic scale is compared with a reference

- The research gap  (L79-L86) is presented in terms of a previous study of these authors. This should be improved by informing the state of the art of all relevant works that have similar research initiatives. Similarly, while the topic of interest is mentioned, the simulation configurations are not justified

- If I understand correctly, only the initial conditions are varied within all the experiments. Does this mean that there are no emissions during the simulation? Could this lead to specific biases when comparing to the observations?

- When presenting the results, ERA5 reanalysis data performs poorly. What could be causing such poor performance in a reanalysis product?

- The last paragraphs in the discussion try to explain the differences between simulations and observations, but the discussion is mainly descriptive. Based on the features that contribute to each difference, are there ideas that could improve future research? For instance, other microphysics schemes.

- Finally, these conclusions are all for a single case study. Can we consider this case study "normal" in order to generalize the conclusions? If not, how can you caution the readers about particular features that may not occur in other cases?

Line-by-line comments/suggestions

L108 measurements "were"

Fig. 1: The mentioned arrow is not clear

L176 This is a generic description; it'd be more useful to close the sentence explaining what will be used in this work

L215 derived by whom?

Fig. 4: Domain-averaged means, Flight path averaged here, correct?

Table 2: Do these values correspond to statistics of all data in space and time? Or at particular timesteps? Or at specific domains as the flight path?

L277 Specify what the above speculation is

Fig. 5. What is QCten? Not explained in the manuscript.