In my view, the author’s have sufficiently addressed all the reviewer comments. Although reviewer #2 indicated major changes, it appears to me that most of the comments required minor changes to the text and have been adequately addressed. There is one comment, however, that I do not feel the author’s adequately explained:

Page 5, Figure 3. It looks to me like Figure 3b (the area distribution, a(D)) is not compatible with Figure 3d (the PDF, n(D)). In 3b, the area distribution increases as D increases, from D=0.2 to D=3. In that range, alpha_S = 1.1 so alpha_S + 1 = 2.1 so the PDF n(D) scales like Dˆ-2.1. The area of a floe scales like Dˆ2. So the area distribution a(D) should scale like Dˆ(-2.1 + 2) = Dˆ-0.1. That means the area distribution should DECREASE as D increases from D=0.2 to D=3. But Figure 3b shows a(D) increasing as D increases over that range. Is there something wrong with the plots, or with my analysis?

The authors’ response is that by changing the range over which alpha_S is determined, one could get different alpha_S because of the concavity of the area distribution. This is true (for example over D=1.5-3, one obviously), but does not explain why the a(D) increases monotonically from D~0.3-2.3 and the fit of alpha_S over that same interval results in alpha_S > 1. Based on the reviewer’s analysis, it alpha_S should be less than 1 over this precise interval. I believe this may be an issue with how robust the fit in Figure 3f is. From D=1.5-2, n(D) is flat, so a fit over this range would have alpha+1 ~1, and this is consistent with an increasing a(D) based on the reviewer’s analysis. But for D=0.3-1.5, alpha +1 looks to be slightly less steep than the red line fit in 3f, so alpha could be < 1 over this range as well, so that would also be consistent with a(D) increasing, although less steeply. That’s all fine. But since the red line fit is to 0.3-2.3, and the blue line appears less steep than this over both the 0.3-1.5 and 1.5-2.3 ranges, why is the red line, which is fit over the full range 0.3-2.3 steeper than what you would intuitively expect for a fit over either of these ranges? Is it due to the fitting technique? If so, then it would seem that this fit is not particularly robust. The authors’ response does not really answer why a(D) still increases over the full range of 0.3-2.3 yet the fit has alpha_S > 1.

Now what the exact value of alpha_S is is not so important for the main points of the paper, but this does suggest to me that there is a possible issue with the fit. This needs to be explained. The statement added to the text is too short, and doesn’t really explain why the increase in a(D) is inconsistent with alpha_S>1 for the reader.

Minor edits/clarifications:

Line 2 change to “made up of floes”

Abstract – Based on the comments and your conclusions, I wonder if it is worth stating in the abstract that the FSD does not exhibit power law behavior?

Page 2, Line 5 – are you sure that “most” of the observations conform to a truncated power law? I believe Stern et al (2018) only states that several studies show this, , or if only some conform to a truncated power law. Also, my understanding is that the “split power law” could be an artifact of this truncation (or finite size effects) – see figure 3 of Stern et al (2018). Your text as written could be read to imply that previous observations show either a truncated power law or split power law. But I believe many of the prior studies only suggest a single power law. Please clarify this statement.

Page 2, Line 11-12 – despite your response to the reviewer, you still state that the power law scaling has not been demonstrated, and you still make the claim that the power-law scaling is justified by the wide range of diameters. Please modify the former to state that the power law has been demonstrated for only some cases, and the latter to state something more reasonable – isn’t it justified simply because it appears to follow a power-law, even if this is not rigorously verified?

Figure 1, caption – “Environmental conditions”

Figure 1, caption – “sourced from the AMSR2”

Figure 1, caption – “The black dots denote”

Page 4, line 17 – “image processing”

Page 4, line 19 – perhaps to better address the reviewer’s comment, this could be changed to “image processing to improve the fidelity of the shape of identified pancake floes”

Page 4, line 19 – “light and ice conditions”

Figure 3 caption – “floe diameter with two power laws”, “(D > 4m), respectively”. Also I think it would be better if the panels were described in order (i.e. a, b, c, d, e, and f).

Page 5, line 7 – be better to state that this could result in an over- or underestimation of the in situ ice concentration. There is no reason to expect this average would always cause an underestimation.

Page 6, line 28 – “transition between these regimes”

Page 7, line 12-13 – are you sure you can say estimation of the scaling exponent it is not rigorously applicable because of the less than a decade of length scales? I am not certain, but I would guess that for a large amount of well-behaved observations, one could achieve a rigorous fit (at least over that less-than-decade range). My point is “rigorously” might not be a precise word here. Perhaps better to say “not particularly meaningful”? Leave alone if it is correct.

References – please ensure your references are consistent with The Cryosphere style. For instance, you have full journal titles, whereas journal style is to abbreviate.

Page 9 line 7 – should annals be capitalized here?

Page 9, line 17 – Reports should be capitalized.

Page 9, line 20 – Herman et al. is now in The Cryopshere (https://www.the-cryosphere.net/12/685/2018/), so this reference should be updated.

Supplementary material – this supplement is quite short, so could this be better included as an appendix?

The authors have adequately addressed all remaining concerns. Specifically the behaviour noted by the reviewer in figure 3 has been explained and clarified. I pleased that the article can now be published as is.