Attached please see my comments.

General comments

This paper investigates aerosol-cloud interactions of warm clouds. A huge number of Lagrangian trajectories (> 1,500) are extracted from various datasets (ARM site observation, satellite retrievals, and earth system model simulations) to understand aerosol-cloud interactions, especially precipitation-aerosol and aerosol-cloud radiative effect relations. I think this work is interesting and scientifically valuable. Some comments are below.

Scientific comments

1. As authors already know, aerosol-cloud interactions depend on cloud characteristics. This paper provides some features about the clouds over the studied region, but those are scattered. It would be helpful if the authors could add one paragraph for some information about the clouds over this region.
2. It is difficult to find a clear reason to reduce N_d dependency and increase Q_c dependency for autoconversion constants. It will be helpful to describe that in Section 2.4. Same as for autoconversion parameters, the reason for the accretion factor change could not be found. Also, it is good to explain how the specific parameter values in Table 2 were selected.
3. Explains about N_d derivation from the CERES and MODIS data are missing.
4. For CERES data, I am concerned about using the dataset during the less trustworthy time (Figure 7 and Table 4, 5, 6, and 7). Figure 7 R_e and N_d figures show significantly different values between trustworthy and less trustworthy time. I hope that the authors discuss this.
5. I understood that this paper considers low-level clouds. If this is correct, I guess that 500 hPa could be too high to check the large-scale vertical motion affecting low-level clouds. I think this confusion could be automatically resolved if comment 1 is considered.

Technical comments

• I would like to recommend using the words “CCN” and “CCN number concentration” more precisely (one example, “increase in CCN concentration” seems better in line 24). Because, as authors already know, concentration is not the only CCN property that could influence cloud (e.g., radius, hygroscopicity, …).
• Both “MERRA2” and “MERRA-2” are used. It seems better to choose just one of them.
• line 25: constant changes -> no change
• line 42: geophysical -> physical
• line 89-90: complete a complete cycle -> complete one cycle
• line 153: aerosol optical depth (AOD): AOD is already mentioned in line 64
• line 188: oceanic -> marine
• line 220: start -> center
• line 306: 3b -> 3d
• line 358: 500-hPa vertical velocity, free tropospheric vertical velocity -> Does it mean “500-hPa free tropospheric vertical velocity”?
• line 393: The change -> The difference
• line 398: from (Chen et al., 2014) -> from Chen et al. (2014)
• line 446: global-based observational -> global observational-based
• line 450: lagrangian -> Lagrangian
• line 497 and 506: equation 5 -> equation (5)
• Figure 5a caption: rain rate -> rain rate (blue) or CCN concentration (red) -> CCN concentration
• Figure 5c caption: Is Figure 5c for A1R0 simulation? If so, why A1R0 simulation result is displayed here instead of the control simulation (A0R0)?
• Figure 7: description for c and d should be checked.
• Table 1: geophysical -> physical
• Figure S4: Is this figure for relative frequency or precipitation rate?
• Figure S9a and b: W/m^-2 -> W/m² or W m^-2
• Figure S10: Description for pink shading is missing.
• Figure S14: LWP unit should be checked.

The proposed paper evaluates the E3SM model skill at representing aerosol-cloud interactions (ACI) in warm rain clouds. The E3SM model is compared against the satellite as well as the ground based measurements from the ARM site in the Azores using a Lagrangian framework. The paper is well written and skillfully uses many different observations and an impressive number of trajectories to evaluate the E3SM model. I suggest that the paper should be published after some minor revisions. 

Main comments:                                                                  

1. I think that the amount and synergy of the observational data used to evaluate the E3SM model is really impressive, and it would be great to highlight it even more. Would it, for example, be possible to add an infographic summarizing all the lines of evidence used in this work?  If not, then maybe a table summarizing all the used data could be included in the main text?
2. It would be helpful to the reader to explain in more detail how the ACI are represented in the E3SM model. The references to the papers describing different parameterizations are provided, but the description of how the coupling between them works is missing. That would be useful to understand, especially when later trying to reason about the E3SM results for different aerosol scenarios and the lack of the desired connection between aerosols and precipitation.
   
   Could the following be described:
   
     - How is the microphysics scheme coupled to CLUBB, deep convection parameterization and the large scale flow?
   
     - And a clarifying question: is the KK2000 scheme part of the Gettelman and Morrison 2015 scheme?
   
     - How is the MAM4 aerosol model coupled to the microphysics scheme? This includes describing (i) how the available aerosol number concentration is translated into CCN and cloud droplet number concentration in the CLUBB+KK2000+MAM4 framework; (ii) how rain affects the aerosol number concentration in different MAM4 modes in the CLUBB+KK2000+MAM4 framework?
   
     - For completeness could the formula for accretion rate also be provided (around page 6 line 185). 
   
     - Could an example of the current clean, polluted and pre-industrial aerosol size distributions be plotted as seen by MAM4 model?
3. How were the values of the autoconversion and accretion parameters chosen for the sensitivity experiments?  What is the (micro)physical interpretation of the different setups?          
4. As shown in Figure 5 (b) and (d) and discussed in text: The ARM observations show more gradual rate of CCN decrease before t=rain. In contrast, E3SM shows some CCN buildup and then a very sharp decrease before the t=rain. After t=rain the ARM measurements show a sharp but steady CCN rate of increase. In contrast, E3SM predicts a very sharp initial CCN increase, followed by a much slower growth. Is the CCN buildup in E3SM before the t=rain explained only by the lack of efficient wet deposition? Are there other possible issues related to the MAM4 model, and the way it is connected to KK2000 and CLUBB, that would explain this? What is the possible explanation for the differences after t=rain?  

Minor comments:

1. Would it be possible to keep the y-axis the same between the observations and E3SM model results? (For example Figures 3, 5, 8)? Maybe providing additional zoom-in for cases where the model and observations are too far apart? I think it would make the comparisons between the model and observations easier.
2. Fig 4a: The legend font is too small.
3. Fig. 6 and following: It is hard to tell the pink and red colors apart.
4. Fig. 6 and following: It is hard to tell the stars and circles apart, especially in the areas where they are clustered. Maybe some different shape or solid vs not-filled shape would be better?
5. Fig. 7: The yellow color is hard to see.
6. First line in section 3: - Should it be 3 exceptions?
7. Page 9 line 281: Either "are" or "may"?