The manuscript contains numerous changes, which are only partly motivated by reviewer comments.

I do not find the answers to the reviewers questions sufficiently convincing. The major point mentioned by both reviewers was the underlying assumption that the models variability is representative of the atmospheric variability which CARIBIC samples. The main conclusion reached is that after smoothing the data to yield a similar 20-min (short term) variability as the model, the long term variability in this smoothed data set is still about twice as large as the model variability. I do not think that this supports the statement that "The representativeness evaluated from the model data alone is also valid for the real atmosphere and the measurements taken by CARIBIC".

Please be more careful with the conclusion drawn.

Second, the answers to the reviewers are insufficient. Please explain for every comment in detail how you have reacted, e.g. not by saying "the wording has been adopted", but by explaining how it has been changed.

Third, please explain the "numerous other changes" made to the manuscript in detail, i.e. every change (beyond wording) needs to be clearly shown and explained. This includes the inclusion of two new appendices, only one of them motivated by a reviewer as far as I could see.

If these explanations are not given, the manuscript cannot be accepted and must be resubmitted.

dear Johannes Eckstein,

thank you for the revisions and the explanations of the revisions. I have one remaining major issue, which is related to the model-measurements comparision of variability and in particular to the use of Meas_smooth_CARIBIC, which by defintion has the same variability as Mod_sampled_CARIBIC on the 20 min time scale. It is thus not an observed variability, but a model variability. Therefore the term used here is misleading and should be eliminated. It has no relation to a measured variability.

In my view, the original question posed by both reviewers, if the model can represent the observed variability is thus not answered. This would only be achieved if an objective way of smoothing would be used for the measurements (e.g. use of 20 min averages). Right now the new section 5 in my understanding compares model variability on longer time frames with model varaiblity on the grid scale of about 200 km (which actually corresponds more to 15 min of flight time than 20 min).

The question posed by both referees was, however, how well the model represents true atmospheric variability as measured by CARIBIC. This question is not answered.

Please correct me if I have misunderstood the approach of the new section 5. Otherwise, please explain how well the model represents the observed variability.