Reviewer1

Suggestions for revision or reasons for rejection (will be published if the paper is accepted for final publication)
1. Spatial scale. The revision states “the spatial scale or the extent of the analysis area is identical between the CSRM simulations and the GFS simulations, although the number of grid points in the area or the domain is different between the CSRM simulations and the GFS simulations due to differences in resolution between those simulations.”
The term spatial scale is ambiguous. The analysis area determines the upper limit of scale, but the lower limit (the grid size), is very different in the two configurations. This must be stated clearly, as it has a huge influence on the interpretation of the analysis. I recommend something like “since the resolution of the CSRM and GFS simulations is quite different, the smallest scales analyzed are quite different, which will affect the resolved updraft intensity and hence the interpretation of the results”.
2. Scale-aware schemes. The way this is written, “The implementation of the cloud microphysics such as the two-moment (e.g. Morrison and Gettelman, 2008; Morrison et al., 2009) and scale-aware schemes are intended to reduce these uncertainties.
” it sounds like scale-aware schemes involve cloud microphysics. To reduce this confusion I suggest “The implementation of two-moment cloud microphysics (e.g. Morrison and Gettelman, 2008; Morrison et al., 2009) and scale-aware convection schemes are intended to reduce these uncertainties.

Reiwer2

In this manuscript, the authors ran a cloud-system resolving model (CSRM) at different horizontal resolutions and with different microphysical processes for understanding the effects of model resolution and microphysical parameterizations on the simulations of clouds, precipitation and their interactions with aerosols. By comparing their CSRM results with results from a NWP model, GFS, the authors made further statements about issues with global models. I like how CSRM sensitivity results were analyzed in the paper, but I think the comparison with GFS results may need further work, as the differences between the CSRM and GFS can come from many different aspects, and it is not clear how much you can learn from this comparison. I would suggest the authors to focus on the differences in CSRM simulations instead, but not on the comparison between CSRM and GFS simulations.

Major comments:
1. The GFS does not have aerosol-aware cloud physics and the paper only included one GFS simulation. But in many places, the authors used the CSRM results to refer the deficiency in simulating CAPI in GFS. This can be very confusing and is also misleading. For example, it stated in the abstract that “results here also indicate that the use of coarse resolution in the GFS model lowers the sensitivity of updrafts and cloud variables to increasing aerosol concentrations compared to the CSRM simulations”. The paper only shows how CAPI are different in different CSRM experiments, and applying this to GFS is clearly a stretch. I suggest the authors to remove any reference to CAPI simulations in GFS in the abstract. I also suggest the authors to reduce the discussion of CAPI simulations in GFS in the main text, as most of these discussions are speculative in nature and do not add much to the paper.
2. The difference between GFS simulations and the coarse resolution simulation of CSRM. As results from the GFS simulation and the coarse resolution simulation of CSRM are similar, the authors argue that the difference between the coarse resolution and fine resolution of CSRM can be used to understand the deficiency in GFS results. However, the similarity between coarse resolution CSRM and GFS does not necessarily mean they are caused by the same reasons, as coarse resolution CSRM and GFS are different in many aspects and similarity in results may come from compensating errors. For example, GFS uses cumulus parameterization, but CSRM does not have a cumulus parameterization. As GFS and CSRM are two different model systems, they are also different in many other aspects, such as turbulence and cloud microphysics schemes. So it is unclear to me how much the differences between GFS and the fine CSRM results can be contributed to resolution and cloud microphysics.

Specific comments:
1. Abstract, lines 26-31: See two major comments. It is unclear to me how much the authors can learn about GFS by comparing GFS results with their CSRM results, as they are many other differences between these two systems.
2. Abstract, line 32: “the CSRM simulation can function as benchmark simulations”. This statement is questionable to me. See my major comment #2
3. Lines 37-38: this statement is also questionable to me. See my major comment #1.
4. Section 2.1 and 2.2: the differences between CSRM and GFS. It is clear from the model description in Section 2.1 and 2.2 that there are many differences between two model systems, such as turbulences, cloud microphysics, etc. This has implications for how the authors interpret differences in results between the CSRM and GFS simulations.
5. Section 5.1.1 and Section 5.1.2: Suggest to combine these two subsections, as messages in two subsections are very similar.
6. Section 5.1: 15-km and 35-km resolution CSRM simulations without cumulus parameterization: I understand the authors’ logic for these simulations, but CSRM results without cumulus parameterization at 15-km and 35-km resolution may not be that meaningful. On the one hand, these resolutions can not directly simulate cloud-scale motions, but on the other hand, without cumulus parameterization, cloud scale motions are not parameterized through subgrid cloud schemes either. So it is not clear how much you can learn by comparing CSRM results without cumulus parameterization at 15-km and 35-km with GFS results with cumulus parameterizations.
7. Lines 289-290: The difference between GFS and CSRM can come from other factors, as these two systems are different in many aspects.
8. Lines 309-310: this statement is again questionable to me.
9. Lines 315-318: the statement here regarding CAPI in GFS is clearly a stretch. I do not think the authors provide enough evidence to support this statement.
10. Lines 350-362: the underestimation in mass flux in GFS. As the mass flux in GFS with cumulus parameterization and the CSRM without cumulus parameterization at 35-km resolutions are similar, an alternative explanation is that the cumulus parameterization in GFS may not work properly for the two cases examined here.
11. Lines 363-366: this statement is questionable to me (see my major comment #1).
12. Lines 432-434: Again, this statement is not convincing to me.
13. Section 5.2: the differences in sensitivity experiments from different microphysical processes are quite interesting. I would suggest the authors to further understand how these different processes (saturation, sedimentation, and collection) affect cloud properties and their responses to aerosol perturbations by examining process rates from different microphysical processes.