Please see the supplementary PDF.

This study investigates the impact of newly implemented 2-moment cloud microphysics scheme on the simulated aerosols and their interactions with radiation and clouds in a global model at 14-km resolution. The authors find that with the new scheme the simulated aerosol burden is overall decreased, which is (said) mainly due to a faster cloud to precipitation conversion (as suggested by the increased RPCW ratio). Consequently, the direct effects of all aerosols (natural + anthropogenic) are reduced. On the other hand, the indirect effect (forcing caused by aerosol-cloud interactions) of anthropogenic aerosols is greatly increased (about doubled). The authors state that there are two possible reasons: 1) the cloud water adjustment changes; 2) the lower background aerosol AOT (burden, CCN).

Evaluating the impact of cloud microphysics change on the aerosol lifecycle and aerosol-cloud interactions is important for global aerosol-climate model development, especially for high-resolution applications. Results from this study will serve as a reference for model development and help the modeling community to better understand the behavior of this model. Therefore, I think this study fits the scope of GMD well and it could be a useful reference. However, I think the current manuscript needs to be significantly improved, especially in evaluating the simulated cloud microphysics responses to aerosol perturbations and in explaining the differences in the simulated aerosol indirect effects. 

Major comments: 

1. Since the focus of this study is on the impact of cloud microphysics on aerosol simulation. It’s vital to show the cloud microphysics property changes in the simulation. I would recommend the authors to check the cloud water mass and number budgets in the simulations and evaluate the impact of aerosol perturbation on the budget changes. A good example is shown in Salzmann et al. (2010) in ACP. 

2. The authors emphasized the liquid water adjustment (2nd indirect effect) in the abstract and summary. What is the impact of the Twomey effect in the model? The cloud droplet number changes (PD vs PD and PD-PI vs. PD-PI between the simulations) and the impact on effective radius and cloud albedo should be evaluated. 

3. It is unclear to me why the authors include the comparison against HRM and LRM (from the other study) and why the discussions are only for some of the fields, but not the others. If the authors want to include this part, the title should be revised (to reflect the impact of resolution and time stepping changes). 

Detailed comments: 

Page 1, Line 21: It would be useful to report the net effective aerosol forcing and ERFari (Ghan’s method) as well. 

Page 1, Line 23: e.g., -> i.e., 

Page 1, Line 31: Why is the 2nd indirect effect (LWP adjustment) so important? How about the Twomey effect in this model? 

Page 2, Line 32: It would be better to look at the ERFaci vs. CCN relationship, rather than ERFaci vs. AOT. 

Page 2, Line 37: better change “aerosol nucleation” to “aerosol activation” to avoid confusion. 

Page 3, Line 68-69: "difference in the simulated aerosol mass concentrations” Do you mean surface concentrations or mass burden? 

Page 3, Line 72: Please provide the reference. 

Page 3, Line 93: Are they climatological runs, or AMIP-style simulations with transient prescribed SST? How is the model initialized? 

Page 4, Line 109-110: Water vapor is not a hydrometeor. 

Page 4, Line 112: How is the updraft velocity calculated in the model? 

Page 4, Line 116-117: What is the background CCN value? Why is it needed? How is the updraft velocity calculated? 

Page 4, Line 126-127: "although rain does not directly change cloud water” Doesn’t auto-conversion in NSW6 affect lcoud water? Also, could you elaborate why the impact of aerosol on cloud water is overestimated? 

Page 4, Line 134: Better present this (cloud/precipitation observational data) in a separate section. 

Page 5, Line 161-162: How is the aqueous chemistry production handled in the original model? 

Page 6, Line 173: Which year of emission is used? 

Page 7, Line 207: To better compare with other models, it would better to use 1850 aerosol emissions, rather than using zero emissions. 

Page 7, Line 210: Better present this (aerosol observational data) in a separate section. 

Page 10, Line 301-302: “An increase in the RPCW leads to an increase in the aerosols that are dissolved into clouds,” RPCW increase can only enhance the turnover time of cloud liquid. Does the model consider cloud processing of aerosols? 

Page 10, Line 322: Please define HRM. 

Page 12, section 4: the ARI and ACI represent different aerosol forcing/effects. Better report both the effective aerosol forcing (ERF_aer_total, ERF_aer_ACI, ERF_aer_ARI) of anthropogenic aerosols and the direct effect (IRF_ARI) of total aerosols. It would be also useful to report the surface forcings. 

Page 14, Line 451: The discussion here is a bit vague. Some further analysis is needed to explain the difference. 

Page 15, Line 483: What do you mean “performance improvement” here? Computational performance or better represented physical processes in the model? 

Reference: 

Salzmann, M., Ming, Y., Golaz, J.-C., Ginoux, P. A., Morrison, H., Gettelman, A., Krämer, M., and Donner, L. J.: Two-moment bulk stratiform cloud microphysics in the GFDL AM3 GCM: description, evaluation, and sensitivity tests, Atmos. Chem. Phys., 10, 8037–8064, https://doi.org/10.5194/acp-10-8037-2010, 2010.

In this study, the authors quantify the impact of a double-moment cloud microphysics scheme on aerosols in a high-resolution global model, comparing to its single-moment counterpart in the same model. The description of results and comparison between the two and with published results in the literature are quite comprehensive, but the explanations of differences are often handwaving. Particularly interesting result is the much higher ERFaci in NDW6 than in NSW6. In my opinion, more in-depth analyses in terms of how the two schemes treat aerosol indirect effects on clouds and precipitation in the model are needed.  Also, what’s the implication of such results for the use of double-moment microphysics in high-resolution climate models?

It’s hard to find specific information regarding the setup of experiments (NDW6, NSW6, HRM, LRM), except for the resolutions and microphysics schemes in Table 1. Which specific simulation years? What are the aerosol emissions used for the simulation years, mean or year-specific? Such information is important to determine whether the comparison of simulations results with observations and other models in the literature is valid. Please include the details in Table 1.        

At many places, IRFari and ERFaci are referred to as “shortwave” aerosol forcing. If the longwave component is not considered at all, I don’t think they are comparable to the cited values in the literature, which mostly include both shortwave and longwave components and are referred to as net radiative forcing. Please confirm and clarify.  

Both NDW6 and NSW6 appear to simulate a too weak SWCRF and LWCRF (shown in Table 2), as compared to observations and historical mean of CMIP models. Is this due to interannual variability or model bias? How does this affect the evaluation of aerosols and their associated forcings in NDW6? Please include a discussion on this issue.

Minor comments and technical corrections:

L118-119: If CDNC is fully prognostic in the NDW6 double-moment scheme, I assume it’s always updated with source and sink tendencies. Why is there an additional constraint by CCN that depends on supersaturation as well? please clarify.

L123: Please clarify on “which” is updated in this study. NDW6 or NSW6?

L132: Is there no shallow convection parameterization at 14-km grid spacing? Please justify.  

L157-159: This might be relevant to the comment for L118-119. This treatment needs more clarification and justification. If the aerosol scheme and the NDW6 are fully coupled, I don’t see why this constraint is justified for certain conditions only. CCN number, which is only meaningful with supersaturation specified, can be larger than CDNC before the activation tendency is updated to CDNC. Otherwise, CCN diagnosed from interstitial aerosols should be mostly smaller than CDNC in clouds. I wonder whether partial-grid clouds matter here.

L174: Change anthropogenic “materials” to “sources” or “activities”.

L207: Does that mean there is no BC, OC and SO2 emission in the extra experiment except for fire and volcanos? I wonder how this experiment is different from the preindustrial-condition experiment, which should also have emissions in the anthropogenic sectors (e.g., residential, agricultural waste burning, etc.)

L247-249: the use of “products” and “results” for model simulations and satellite, respectively, should be the other way around, i.e., referred to as simulation results and satellite products.

L261: What does the “horizontal biases” mean? Spatial or regional biases?

Figure 7: “AOD” is used in the figure labels, but “AOT” is used in the figure caption, tables (e.g., Table 3), and the main text. Please make them all consistent.