Bennartz and Rausch have answered my minor or specific comments satisfactorily. However, I feel that my two main remarks were not addressed substantially.
1. As for the “Idealized Stratiform Boundary Layer Cloud”, I believe I’d be happy if they explicitly wrote, at the instance where the authors introduce this term, that they mean exactly the same they meant before by the term “adiabatic cloud model”. As such, the reader can understand that it is simply a re-branding of the same thing.
2. As for the “uncertainty” assessment – the authors still call the sub-scale variability “uncertainty”. I believe this is mis-leading. The newly added paragraph does not improve the assessment much. The authors write in this paragraph “… the standard deviation of all Level 2 observations within a given 1x1 degree box … should in any case be larger than the uncertainty derived from error propagation”. This would be the case if all errors are statistical errors. However, it is clear that there may be substantial systematic errors in the retrieval as well.
The authors write that because the subscale standard deviation is larger than the assessed propagated retrieval error, data users may be more interested in this quantity. I believe, in turn, that users should be able to tell apart what is variability and what is error. Only the latter is an actual uncertainty, the former should be a measure of a meaningful physical behaviour.
As such, I maintain my previous remark that if the authors choose to report subscale variability, they should call it “subscale variability”. If the authors choose to do more rigorous error analysis, more detail is needed on how the error is computed and how it is propagated. In any case, the reader needs more details on how exactly Fig. 8 (top panel) was created.
Typos
p2 l28: km²
p16 l18: consistent
p16 l26: “These … themselves”
Caption Fig. 3: “fraction of a typical” - the authors should clarify what they mean by “typical” |