Based on data collected at 62 ground-based sites around the world, spatial distributions of aerosol number concentration and particle number size distribution are presented. The data are segregated by site type (e.g., polar, urban, mountain, etc.) and degrees of seasonality and diel cycles are reported and explained. This paper is a massive effort and the authors should be commended for a job well done. The paper is well written and will provide a lasting resource for modelers to assess simulated aerosol number and size distributions regionally and globally. I only have the very minor comments listed below.

Page 5, lines 10: There have been many long-term trend analyses of optical properties, number concentration, and PNSD. Does this refer to analyses involving several sites? Please clarify in the text.

Page 6, line 10: Why 2016 and 2017? Please explain in the text.

Figure 1 and throughout: The symbols in the legend box should be larger.

Figure 1 caption: It would be helpful if MPSS were spelled out.

Page 6, line 18: What is meant by “(sub)-urban”? Does it include both suburban and urban? As far as I can tell neither it or “suburban” are used anywhere else in the paper.

Page 7, line 9: What is the definition of “large” particles as used here?

Figure 10: It is difficult to tell the difference between the size of the symbols. Please change so that the size difference is clear. The yellow symbols are difficult to see. Can these be changed to another color?

            Also – up to here the coefficient of determination (r^2) has been used to describe the significance of the correlation between two parameters. Why is the Spearman’s rank correlation coefficient being used here?

Page 40, line 5: change to “RESIDENTIAL”

My apologies to all for the tardy review.  I view the data analyses in this paper as strong, but some of the interpretations and conclusions derived from the analysis are not nearly so rigorous as the methods.  The introduction is nicely written, and the analysis of the fraction of available data that is important to adequately represent the statistics appears to be very well done.   At about 31 pages of text and 13 figures, the paper is too long. 

Section 5.2 is a series of cursory discussions about what the variations in Ntot may represent.  The discussion here does not offer a lot, and some of the inferences concerning POLAR sites are a little misleading.   I t might be helpful if the authors told us how these data should be used to help global models: e.g., which plots might you think the most useful for model evaluation? 

Detailed comments

1. Page 2, lines 43-44 – It is unclear whether multiple transformations include wet and dry deposition. Explicitly mention deposition.

2. Page 3, line 10 – maybe “less evident” rather than “barely marked”.

3. Page 3, line 18 – “>50-100 nm” is unclear. Instead, maybe “(considered here as either >50 nm or >100 nm)”.

4. Page 5, lines 17-20 – The wording here suggests that such closure studies have never been conducted until recently, which is not the case. Please add something like “at the above-mentioned network sites”.  The ACTRIS reference relates to the Schmale study, not this generalized statement.

5. Page 5, lines 31-33 – It is very disappointing to see no PSND data from Alert and no data at all from the Whistler Mountain site (both of which have many related publications) not included in the “most up-to-date information … worldwide…”.

6. Figure 1a – “Pristine” is defined as “belonging to the earliest period or state”, “not spoiled, corrupted, or polluted (as by civilization)”, “fresh and clean as or as if new”. In terms of the aerosol, this definition means that a site is pristine if it is not influenced by anthropogenic sources, rather only by natural sources.  Parts of the Arctic can reach a pristine condition during the summer, but it is incorrect to broadly represent the Arctic stations as pristine.   Arctic Haze is not a pristine condition.  The general use of ‘pristine’ to describe the polar sites is unjustified. 

7. Section 3.2 – You state that you fit two modes to the particles between 20 nm and 500 nm, and the first mode is comprised of Aitken particles (maybe nucleation particles, depending on one’s definition). Is there a typical size range for the minimum between the two modes, and can this minimum be related in some cases to cloud processing (i.e., Hoppel minimum)?  Also, modes of some urban and rural distributions can be singular.  Is that allowed in your analyses?  Lastly, is 100 nm your definition of the split between Aitken and accumulation, and would you be explicit about that, please.

8. Section 3.3 – Two points are missing here: 1) CCN number concentrations do not necessarily translate to cloud droplet number concentrations, particularly at higher concentrations, due to competition for water vapour by droplets growing at cloud base; 2) particles at 50 nm can only be effective if the cloud base supersaturation reaches a sufficiently high value. The latter happens in one of two ways: relatively few particles larger than 200 nm, such that the rate of water uptake by the >200 nm particles does not dominant the total uptake; very rapid cooling rates, typically associated with relatively extreme updraft speeds.  

9. Page 16, lines 18-30 - There are others in Figure 2 that appear to fall into the category of 95% or more: ZEP and ALT, LLN, WGG. Why were they not used?  Also, I find it odd that data coverage can be essentially 100%, as it appears to be for TRL and NMY.  Do these sites not perform checks on the operation of the instrument, zeroes, etc.?  The filtering for wind direction, referred to for SPO, BRW and MLO, also happens for the ALT data, and it has been shown that that such filtering of the Alert data has little impact on the observations at Alert, except for a near-negligible impact on Aitken-size particles.

10. Page 17, line 8 - I tend to agree that 4-5 weeks shows little impact, but visually there is not much difference between the medians at 12 weeks and those at 20. The choice here of 12 weeks here appears to be more a matter of convenience than rigor.  How do you justify 12?

11. Page 17, line 10 - PAL and VAR are both close to the Arctic Circle and relatively close to each other. Alert is much more remote, offers another perspective and appears to qualify, based on Figure 2.  Why was it not chosen?

12. Page 19, lines 7-10 – Although I find the diel cycle analysis interesting, I don’t really see how it helps global models. Somewhere in the text, can you offer some explanation for the importance of this on a global scale?

13. Page 23, line 5 – Should 6.a be 6.c?

14. Page 24, line 12 – The use of “generally” and “around” to characterize something so precisely defined as 42+/-14 seem a little out of place.

15. Page 24, lines 17-18 – Freud et al. (ACP, 2017) should be referenced in connection with this statement.

16. Page 24, lines 19-20 – The Arctic and Antarctic are quite different in some respects, most notably the strong influence of southern-latitude anthropogenic sources on the Arctic in the winter and spring, also known as Arctic Haze. Arctic Haze is significant, and the number concentrations of accumulation-mode particles can extend through significant depths of the Arctic troposphere.  Your generalized statement, based on measurements at the Antarctic, should not include the Arctic.

17. Page 24, lines 28-29 – I don’t understand the statement beginning “Larger diameters ... during JJA”. One of the reasons that Ntot is higher at Arctic sites in the summer (JJA) is because the accumulation mode concentrations are much lower, which helps NPF (e.g., Freud et al., ACP, 2017; Croft et al., ACP, 2016; Leaitch et al., Elementa, 2013).  This statement appears to say the opposite. 

18. Page 24, lines 32-33 – “Transport…” – Yes, but for accumulation-mode particles the transport in Croft et al is accompanied by stronger wet deposition, which significantly reduces those particles.

19. Page 25, lines 6-8 - Arctic Haze, which has been studied for over 50 years, is pervasive during the winter and spring in the high Arctic, largely because of the reduced deposition of these particles in the Arctic during those times of the year. Your statement makes Arctic Haze sound like an occasional factor, which it is not. 

20. Page 28, lines 13-14 – It may be fair to say that the higher Ntot at mountain sites during summer are similar to polar sites, but the reasons may be (and likely are) very different, and this should be noted.

21. Page 29, line 14 – remove “instead”

22. Page 30, lines 1-2 - Can't the lower correlation be related to the fact that the TopoIndex is a calculated quantity, whereas altitude has far less uncertainty?

23. Page 37, lines 9-11 – You suggest here that the diurnal variation at polar sites during the transition seasons is responsible for NPF events, but Nieminen clearly shows that the frequency of observation of NPF is predominantly a summertime phenomenon in the Arctic. This needs correction.

24. Page 39, lines 5-6 – You need to note here that for 50-80 nm particles to be effective at nucleating cloud droplets, the N100 must be significantly lower. Competition for water vapour among the larger (and soluble) particles will not permit supersaturations to reach sufficient values to activate 50-80 nm particles unless the larger particle concentrations are quite low.

25. Page 39, lines 27-28 – It would be interesting to see how Ntot-N100 and N100 compare, and if there are locations/situations where they exhibit some degree of mutual exclusivity or correlate.

26. Page 39, line 34 – interstitial spelling.

27. Page 40, lines 1-3 – The beginning of this sentence seems to be missing something. Do you mean “in contrast to those observed”?  Also, here you mention “when clouds are less prevalent and transport is most favoured”, referring again to Croft.  As above, Croft does indicate higher transport of accumulation-mode particles, but they also show substantially increased wet deposition.  This sentence should be re-considered.

28. Page 40, lines 14-15 – “additional source” relative to what?

29. Page 40, lines 19-22 - Here you mention "contribution", but a slope is not necessarily a contribution. In Figure 13, you show the true contributions of N100 to Ntot to be typically less than 30%.  Please explain why the contributions are so much higher here.

30. Page 40, lines 26-27 – I don’t understand how the summer and winter “behaviour”s are close; perhaps ‘behaviour’ is the problem here. In winter, the size distributions are for particles mostly larger than 80 nm.  In summer the size distributions are skewed towards smaller particles, e.g., see Freud et al. (ACP, 2017). 

31. Page 40-41, lines 34-4 – You overlook the issue with the impact of particle concentrations on cloud supersaturation, which is just as important as the numbers of CCN.

32. Page 44, lines 4-6 - Figure 5a indicates relatively uniform global coverage, although, as you mention, Africa, Russia, South America and oceans are clearly under-represented. Do you need all 39 European sites?  Why not introduce a bit more global balance by conducting the analysis using only the sites in Figure 5a plus a smaller selection of European sites?

33. Page 44, lines 22-23 – It may be true for some of the forest sites, mountain sites and polar sites, but it does not appear to be true for the urban or rural background sites.

34. Page 45, lines 2-3 – Based on section 3.2, it seems as if your analysis essentially dictated the bimodality. Is that false?

35. Page 46, line 10 – Remove “Slight”.