|The restructuring of the paper’s theme has improved the readability and scope dramatically. I now understand how the different model runs are important to present to the scientific community and falls under the scope of GMD. |
-The reversal of the colorbar between the plots is confusing (sometimes red is an increase and sometimes it’s a decrease). Is there a reason for this? Typically, readers associate blue with a reduction and red with an increase. It would help if it was consistent throughout the paper.
-It may help if some of the supplementary figures are in the paper as official figures. It is hard to flip back and forth, plus the supplementary figures are referenced more than the included plots. The correlations could state supplementary, but the aerosol type and precipitation / environment plots are important and should be included in the manuscript.
[39-40] “However, there are relevant processes that GCMs usually model dynamically, but
which RCMs usually do not” – my understanding is that GCMs parameterize and specify constant values more than an RCM because RCMs have better resolution? Should GCM and RCM be switched in this sentence? I don’t see how WRF would model fewer things dynamically than a GCM, especially related to aerosol-cloud processes. Please correct my ignorance here if I’m reading this wrong.
[43-45] - There are also longwave and shortwave absorbing aerosol types like dust and smoke that can warm aerosol layers aloft while decreasing surface temperatures, leading to stable layers. This sounds like it’s being discussed right after the part on semi-direct effects, but it is a direct effect. The citations are clumped together for these two things when they are distinct.
 - The RACM-GOCART combination produces SOA, but it may only couple to WRF in the form of radiative effects. I don’t know that SOA is connected as an indirect effect. Please check this.
[159-160] - WRF-Chem does have fully coupled aerosol-cloud-radiation modules, including MADE and MOSAIC, they just were not selected here most likely because they are computationally expensive.
[165-167] - “The WRF-Chem model makes it possible to transform the single- into a double-moment scheme” – this sentence is misleading and needs to be phrased more like “The WRF-Chem model assumes XXX to infer an aerosol number concentration from aerosol mass” and the parts about converting from single to double moment should be removed.
I think my point in the first round of comments was missed. The Ghan et al. (1997) CCN formulas still require an input of aerosol number. GOCART and Lin microphysics does not predict aerosol number because it’s single moment. Thus, some assumption must be made to get at aerosol number with this setup. This conversion between mass and number doesn’t magically transform a single moment scheme into a double moment scheme because there is no value-added information: it is an assumption to make the model work. I’m asking the authors to check in the code or reach out to the WRF-Chem team to understand how this setup gets number information from mass.
I urge the authors to not use the model as a black box and understand what is being represented, what is being assumed, and what is not in the model whatsoever. The authors may think I’m being unreasonable here, but this is important for understanding the limitations of this study and to inform researchers who may look to this setup for future work.
 – The calling of a different autoconversion scheme should be mentioned in the model setup section too and not just at the end. It’s good to know limitations before looking at the data.
[247-248] “With higher aerosol concentrations over most of the domain, reducing it by up to a half” – does this mean that where there are higher aerosol concentrations, the reduction is stronger?
 – More stratiform clouds? More convective clouds? Because lower surface temperatures, despite an increase in RH, can lead to less convection.
[260-262] – The semi-direct effect would be in both the ACR and ACIR simulations, correct? This also applies to discussions of the semi-direct effect and it’s attribution throughout the paper.
 - Large aerosols, or GCCN, can accelerate cloud processes such as nucleation and collision-coalescence. What do you mean by large aerosols would prevent cloud formation?
 – Could it be that because most of the aerosol mass is dust, that the absorption is creating stable layers and preventing convection?
 – The clear sky correlations could be impacted by aerosol-environment co-variability. For instance, dust is associated with dry, hot, cloud-free weather. Those aerosol particles can impact the environment and make it hotter and warmer. Do you think that is at play here?
 - Evidence to support this argument that cloudiness is the most important ?
[Figure 1] – Why is the correlation so low between AOD and RSDS? Even if there is an indirect connection between these two variables via cloudiness, I’m surprised it’s so low.