Accurately forecasting fog is currently a challenge for many operational numerical weather prediction models, with important implications in both weather and climate prediction.  It has been shown in previous work (eg. Boutle et al 2018) that improper aerosol activation can lead to poor fog forecasts.  This paper presents a detailed analysis of fog properties in model simulations as compared to observations for 11 fog events, and discusses the impact of aerosol activation as well as several modifications to the aerosol activation scheme.  It is shown that the model, with 500m grid spacing, is capable of producing fog with reasonable droplet concentrations (N_d), and that N_d is sensitive to the aerosol activation parameterization.  The authors include a discussion of the limitations of their study.  Recommendation: Accept with minor revisions

General Comments:

It may be helpful to the reader to include a table or graphic of the various schemes used in the various model components.  I found the model description long and somewhat confusing, due to the 3 different model components, each using different parameterizations (just a suggestion).

Given that the Mod-ARG scheme significantly improves the results, how important are the multiple aerosol modes?

The role of sedimentation is only very briefly discussed, but could be important in removing droplets and aerosols.  Was there any evaluation of the surface deposition due to sedimentation?  Was there precipitation in any of the cases?

It is somewhat surprising to see a lack of sensitivity to aerosol hygroscopicity.  Perhaps, as the authors suggest, that is due to the type of fog that is studied here.

This study included a complex chemistry scheme, which is probably too expensive for typical weather and climate forecasting.  Could the authors comment on the minimum requirements needed in NWP to produce reasonable fog simulations?

The paper is long.  I would encourage the authors to consider ways to shorten the text.

Specific comments, typos, etc:

Line 51: thermodynamic is misspelled.

Table 1: Welas 2020?  Or  WELAS-2000 as in lines 137 and 138?

Line 166: The author should be van Weverberg, I believe.

Fig. 5 (and others): Please clarify if the model profile is a point value or an average over multiple points?

Fig. 12 caption refers to S9, but it looks like that should be S10.

Line 496:  Are the 2 numbers referring to the two events on the 16^th?  Please clarify.

Line 503: Subfigure is misspelled.

Line 506: November is misspelled.

Line 602: “subfigures (a,b,c)” should be subfigures (a,d,g).

Fig. 15 (b, e, h): Retrievals is misspelled.

In the supplemental material:

Fig. S2: Add to the caption the meaning of the shading (showing the observed foggy periods, presumably)

Figures S5 and S6 appear to be out of order.

Fig. S8, line 2 of the caption: Omit either “boxes” or “gridboxes”.

Fig. S12: The legend in the figure doesn’t match the experiment description in the caption.

Review of “High sensitivity of simulated fog properties to parameterized aerosol activation in case studies from ParisFog” by Ghosh et al.

This study presents simulations of multiple fog cases observed during the ParisFog field campaign. Updates to the aerosol activation scheme are shown to improve the simulation of droplet number concentration. These updates include more realistic hygroscopicities and updated parameters that control activation rate. These updated parameters are found to have a much larger impact than the hygroscopicities. The improved droplet concentration prediction does not necessarily improve the liquid water prediction. The improved Abdul-Razzak-Ghan scheme could potentially be used by others. The manuscript is long and could possibly be shortened. But otherwise, I don’t have any major concerns with the study. It is a straightforward comparison of simulations with observations. The comparison is particularly comprehensive in this manuscript.  

Specific Comments:

1. It would be useful to mention at some point how droplet sedimentation to the surface is handled (if at all).
2. I don’t think that section 5.5 adds much to the manuscript. Consider removing it.
3. Figure 1 should have panel labels since they are referenced in the caption.
4. I believe that units should have spaces between components, e.g. on Line 265 ms^-1 technically reads as per millisecond, not meters per second. Use m s^-1 Same for gm^-3.
5. Figure 3 – presumably the temperatures are cloud top temperatures?
6. Line 671 – I wouldn’t state that LWC can be realistically simulated “most of the time”. Even previously in the conclusions the authors only state “some of the time”.
7. A few typos/extra words to note: Lines 51, 135 (“the point at which point” is awkward), 503, 662 (grammatical parallelism).