|the manuscript seems greatly improved but some (mostly minor) issues could still be addressed. |
1. a new version of the SI is not available at the moment. at least i could only find the old one which was rather sub-par. i hope the one that will eventually be published will be (much) better than the initial version.
2. p. 4 l. 8: in the Eurasian side -> on the ...
3. p. 4 l. 18: fulfill -> fill
4. p. 5 l. 18: "well above the treeline": of course the site is not _above_ the treeline but north of the treeline. only a mountain site can be above the treeline. north may be up on most maps but that still doesn't mean that north and up or above can be used as synonyms in a scientific text.
5. p. 6 l. 3: isn't the inner diameter much more relevant information than the outer diameter?
6. p. 7 l. 27: the accuracy of the numbers given for the correlation coefficients seems a bit much. please also check the rest of the manuscript and make sure to provide only meaningfully rounded numbers.
7. p. 9 l. 18: "zero-intercept forcing": why exactly? a non-zero intercept can give instrumental insight in the same way that a non-unity slope can. why force it? unless the measured size ranges are perfectly identical, a non-zero intercept is actually quite impossible and i wouldn't expect it at all. how is this forcing justified?
8. p. 11 section 2.3.5: much along the lines of the cited beddows paper, it would be nice to have some information on the quality of the clustering. how uniform are distributions within each cluster? this and similar questions could be addressed. i still feel that this cluster business is more or less dropped on the reader without discussing underlying motivations and choices.
9. p. 12 l. 10f: "also extra-modal coagulation losses": i don't understand this phrase at all. modes are not involved when calculation the CoagS for a given particle size. what then is this supposed to mean?
10. the two clusterings and their analysis are still somewhat confusing. on page 15, the N-cluster 4 is related to sec. part. form. and has mostly marine influences. on the next page, the M-cluster 3 represents sec. part. form. and continental air masses. while i do understand that both those things can be true despite the superficial contradiction, this underlines what i have said in the first review: the benefits of the clustering business are somewhat nebulous. in the clusters, various factors are mixed together and then have to be de-entangled one way or another. compared to continental vs. marine split up by seasons and BB separated, what is it that we learn from this cluster analysis?
11. p. 17 l. 24: reassembles -> resembles
12. p. 18 l. 1: "other arctic sites": how would a comparison to antarctic sites look like?
13. p. 18 l. 8: "Average annual growth rates [...] were only seen in summer months": this doesn't seem to make much sense. annual growth rates in the summer months? what's annual growth rates anyways in this context?
14. p. 18 l. 17: inter-annual -> annual
15. p. 18 l. 20: "regional-scale": how do you know?
16. p. 19 l. 19: "time [s]": no, that can't be right. the unit, i mean. and the respective figure has [h] which makes more sense.
17. p. 19 l. 24: "it must be bear in mind" -> probably "it must be kept in mind"
18. p. 19 l. 25f: "data suggests rather exponential than linear dependence": then why not fit such a function and see how well it works out? would the R^2 be better?
19. p. 20 l. 18f: "beta value of 0.126": since temperature has a unit, beta needs one, too, otherwise exp doesn't make any sense.
20. p. 22 l. 15: "is Tiksi" -> in Tiksi
21. p. 22 l. 17f: "Some sporadic events are also observed during dark winter months.": does that mean that you have nucleation without any sunlight? really? surely that would deserve a bit more discussion than a half-sentence in the conclusions, no?
22. p. 23 l. 1: "all the way up to the Arctic": again, the Arctic is north, not up.
Figure 1: the map is still kinda tiny
Figure 2: fonts generally borderline-tiny. in the bottom two panels on the right far too small font. also: those two panels would probably be much easier to read if e.g. continental/marine went up/down from the same baseline. now the straight line is at the bottom, it might be better in between. finally, the left axis numbers overlap between panels. please fix.
Figure 5: why does the aps part of the lower panel have different crosses for the most common cluster?
Figure 6: could use some color. really hard to read in print.
Figure 7: formation rate needs to be more specific. this is J_7, right? also in the text, this needs to be clear. and, btw, why not calculate the more relevant J_1.5 or J_2? according to the kulmala protocol, this should be standard. and it also makes it easier to compare results from different sites/instruments with different cut-offs.
Figure 8: the text above the panels is ridiculously small. please enlarge.
Figure 10: font size is borderline, again. the color code on the left is hard to make out. please consider larger dots with thinner edges.
Figure 11: all those numbers inside the plot on the left are impossible to read without lots of zooming. doesn't work in print.