Overall the authors made a satisfactory job in revising the manuscript but they are a few points I would still like them to address.
1) Now that you have added tracks to figures 3-5, you need to adapt the captions accordingly. I would also discuss the tracks explicitly in the text, as you do in the reply.
2) In response to Point 6 of Reviewer 1, you computed correlations for detrended time series. As far as I can see this new result has not been included in the text and I would suggest to do so.
3) I would suggest that you spend a few sentences on discussing limitations of ICOADS. (Reviewer 1, point 4)
4) I would suggest that you make the reader aware of the issues with detecting tropical depressions in the pre-satellite era (Reviewer 1, point 2)

I had some follow-up discussions about this paper with the two reviewers and we decided to ask you for a final round of revisions to further strengthen the quality of the paper. Please find the main points we would like you to address below. We will try to look into this quickly after you return the revised manuscript in order to avoid further delays in the publication process.

1) In-season trend: While it is not necessary to analyze the in-season trend in detail yourself, you should expand the discussion on the relationship between in- and off-season trends. State clearly what we know from the literature about long-term trends / decadal variability / relation to climate indices for in-season / total storm counts. Then discuss your own findings in the light of these literature results. Give possible reasons why off- and in-season counts may differ. If you think, that global warming affects the off-season (and thus more marginal) storms in a disproportional way, you need to clearly state this as a hypothesis you are testing in this paper.
2) Trends in climate indices: Clearly explain the differences in your MLR models with the full and detrended climate indices. What is the role of the strong trends we see in Figs. 3–5? Why is the ENSO index below zero most of the time in Figs. 3b and 5b?
3) Pre-satellite era: Be more explicit about the differences in trends over the full period and those since the 1960s. Give numbers and significance levels.
4) Data homogenization: Better discuss the modifications applied to the data in the pre-satellite era. There is literature about this for the North Atlantic (Landsea et al. 2006; Landsea 2007; Mann et al, 2007; Mann et al, 2009) but not for the other basins. Critically discuss the situation for the other basins. What evidence do we have that modifications are reliable?
5) Tropical depressions: Moon et al. (2019) report that there is great uncertainty in detecting tropical depressions, even in the satellite era. Recompute your results taking out tropical depressions to check whether your conclusions are robust.
6) TC tracks and SST change: Better explain the relationship between SST trends and changes in TCs in Figs. 3–5. Were there any changes over time in track or genesis location?
7) Relationship to cloud cover: You show a significant correlation but should give a better explanation of a potential physical link.
8) ICOADS: Add more detail on the limitations of ICOADS, citing relevant literature (Freeman et al. 2016, Eastman et al. 2011). Can you justify the use of ICOADS cloud cover data before the 1950s?

Thank you for the additional revision that has further strengthen the quality of the paper.
Despite the progress made I feel that there remain a few points to address and therefore I would ask for one more round of revisions. I organized my feedback according to the eight points given in my last list of points to address. As already pointed out previously, you often answer criticism in your reply to reviewers/editor but then not to make corresponding changes in the manuscript. I repeat my advice to you to take this point more seriously – now and in future revisions.

1) In-season trend: The text you added is sufficient to put your results into the context of other studies. However, there are some redundancies in this paragraph and you should try to streamline it a bit.
2) Trends in climate indices: It is good that you added the table but you now need to discuss these additional results in the text. Just stating that things are similar is not enough. What can we learn from the difference? I don’t understand your explanation of the negative ENSO index in Figs. 3b and 5b and wonder whether there may be a mistake.
3) Pre-satellite era: Again it is good that you included the values in the table but now you need to discuss these findings in the text. You state in your reply that the MK test does not work for 60 years only. If that is the case, you should not present the results like this. Either use a different test or break down your timeseries in pentads instead. How is it possible that WP has negative trends in pre and post but positive in off? Please carefully check all numbers!
4) Data homogenization: The text you added creates at least some awareness of the problem. I think that should suffice for the moment.
5) Tropical depressions: I understand your point but you should include this justification in the paper! You could state explicitly that it would be desirable to compare results including and excluding tropical depressions. Given the many uncertainties you are dealing with I suggest that you add 1-2 sentences to the conclusions with caveats (worse data quality in pre-satellite era, question of applying data homogenization for North Atlantic to other basins, lack of information about intensity).
6) TC tracks and SST change: Here you did not answer my question. Were there any changes over time in track or genesis location? Please check and include a respective statement in the text.
7) Relationship to cloud cover: The sentence you added is good I think. There is, however, a misspelling in the reference (Klotzbach) here and elsewhere. Maybe just use “better” instead of “more optimum”?
8) ICOADS: What I meant is that you should explain how the ICOADS cloud estimates are produced and then discuss limitations (changes in data availability and methods to correct for that). The quote from Freeman et al. (2016) can help you to do this.

Thanks for addressing this last round of comments. I have only a few more minor points and then we can finally publish this:
L192: this technique
L288: Klotzbach
L342: correlation
L391: a non-statistically significant
L483: worse
L534: Landseer should be Landsea