Dear authors,

please take these considerations from one of the reviewers into account (see below).

Editor comments:
Furthermore, it should be indicated in the manuscript, whether the changes in CDNC and the radiation feedback cause modifications of the circulations and how strong these changes are among the sensitivity simulations to get a rough estimate on the resulting uncertainty originating from the different activation schemes or feedbacks with meteorology. Please provide also a quantification whether the differences in the analysed quantities are statistically significant, i.e. how large they are compared to the internal variability of the model, e.g. seasonal standard deviation compared with the differences of the mean of the two model simulations, as this would help to estimate the strength of the effect and discern it from numerical noise.


The manuscript has been much improved. However, the following clarifications need to be made before publication can be recommended:

1) Abstract, line 15:
It seems pretentious to use the term 'realistically' here. Is there hard evidence to back up such a wording? I doubt it. Therefore, I suggest skipping 'realistically'. The sentence reads fine without it.

2) Computational cost
In the previous round, I requested some information on this. That has now been added on page 5, line 98. That is good, but not sufficient. First of all, the sentence says "In our global simulations, the FN05 scheme increased computational time by ~10%.", but it doesn't say what the computational cost is when all the updates are included, i.e. comparing FN05/K09/B10/BN07 to AR-G00, rather than just comparing FN05 to AR-G00. Please add this information.

Furthermore, the computational cost is crucial information to anyone who might wish to use these schemes. Therefore, that information needs to be (briefly!) mentioned also in the Abstract and the Conclusions.

3) Figure 1 caption:
The caption needs to explain what clouds are shown in panel b). Only low clouds over ocean? Is the Bennartz (2007) algorithm used for the retrieval? Without this information, panel b) has little meaning.

4) Figure 3 caption:
It would help the reader if 'observed' were clarified, e.g. by adding "(from field campaigns)" or something like that.

5) Figure 4 caption:
It is very confusing that "the increase (warm colors) in the shortwave cloud forcing absolute change represents more negative values."
SWCF is almost invariably a negative quantity, so an increase (decrease) in its W m-2 value must surely mean that it is weaker (stronger). This needs to be changed!

6) Figure 5, left panels:
These incredibly high numbers are basically useless, because the reader can't even count the number of zeros at the color bar.

There are two easy alternative ways to solve this:

a) Express values as multiples of 10**6 as in Figure 4.
or
b) Replace the unit cm-2 by m-2, which would imply a factor 10**4 scaling.

In either case, the same scaling and notation should be used for 'column CDNC' in Figures 4 and 5.

Dear authors,

even though I am personally not happy with the study, i.e. to analyse aerosol effects only on one year of simulation data, the reviewers are fine with the revised version, and some of my concerns are taken into account in the revised manuscript. Therefore, I recommend publication.