Review of “Review Article: Scaling, dynamical regimes and stratification: How long does weather last? How big is a cloud?” by S. Lovejoy

This article reviews the developments of scaling approaches over the last few decades. Scaling approaches have led to novel insights into atmospheric dynamics and recently provide the building blocks of novel prediction models and climate response models. A review on this topic is needed and will be helpful in spreading the scaling approach to a wider range of scientists.

While such a review is needed, I am not sure if the article in its present form will be able to reach a wider audience. My major issue is with the length of the article. In my opinion the article is too long for a paper, which one could read in one sitting. I could imagine it as a foundation of a book by adding more background material to make it easier to understand the topic.

NPG has no formal page limit for review articles but I encourage the author to shorten it with the reader in mind. The article is well written and I find it hard to point to any obvious location which can be easily shortened. One possibility could be to have a ~20 page overview article and put the remainder into supplementary material.

Some more detailed comments:

1) Line 23-24: This sentence reads awkward.

2) Line 67: Why “lag”? An interval is not a lag. Am I missing something?

3) Line 96: cloud -> clouds

4) Line 125: range scaling -> range of scaling

5) Line 134: levels quantify -> levels to quantify 

6) Line 135: would expected -> would be expected

7) Eq. 3: Hz -> H_z

8) The citation style is often incorrect. E.g. line 277: [Mandelbrot, 1981] termed -> Mandelbrot [1981] termed 

and many other locations

9) Eq. 9: An explanation for 2 in \zeta(2) is missing. 

10) In many parts of the article the author relies mainly on his own studies and of his collaborators. It would be good to include a more diverse set of studies which independently confirms the conclusions.

11) Line 592: that is only true for \zeta(2)<-1

12) Line 597: There is a huge class of wavelets. Which wavelet are you actually referring to?

Later it becomes clear that Haar wavelets are used.

13) Line 604-605: Haar wavelets have some nice properties but in my experience their spectra are more noisy than DFA spectrum for example. 

14) Line 821-822: This sentence is odd.

15) Line 911: It might be good to use H only for the Hurst exponent and another symbol when a more general exponent is implied. That would potentially avoid any confusion.

16) Lines 933-935: How GCMs become effectively stochastic on time scales longer than 10 days needs to be better explained.

17) Line 949 and following: I am having a hard time understanding this part.

18) Section 5: This is a nice summary but a good review article should also point out knowledge gaps and future research directions for the community.