Dear Ruhi Humphries et al.,

thanks for your efforts in revising the manuscript.
The reviewers have some remaining concerns, such that a further revision is necessary before publication. Please, see below.
Best wishes,
Wiebke Frey

Reviewer #1:
Comment
From discussion paper, information of observed data which the authors employed in this study become clear. However, it is still less information. Especially, Figures 1, 3, and 4 are needed the information of observed periods to make these plots. Because, the authors employed several periods data in Appendix Figures. It makes confusion. Although the purpose is to propose the new technical method, it is needed the reliability of data the authors employed.

The plots in Figure 1 was changed from the discussion paper version. Why?
For BC data in manuscript and figures, please recheck the all unit for BC (ng? or ug?)
I could not understand BC criteria was set to 0.07 ug/m3
In showing periods in appendix figures, the authors discuss different periods for the examples, leading confusion. For A1, A3, A11, if these periods was same, different time stamp were strange. Please adjust.
In upper panel of Figure A5, please indicate the date information. It is unclear “older” and “more recent”
In CCN in Figure A12, the new method indicated that several data in CCN was recover from eliminated data by wind-filter. It seems that some CCN data is in around 1000 cc-1. Is it acceptable? Authors mentioned CCN in these areas were few hundred. Authors should describe why these data (in 20min) were available for further analysis.
New method indicated quantitative percentage change. What is total? (time?) (in the discussion paper, the authors indicated the time.) Because there are many case studies in the manuscript. if data were employed all period, please indicate.

Reviewer #2:
With the latest version, the manuscript is improved by point-to-point responses. The authors quantitatively provide the percentage of recovering data as well as of excluding “exhaust” data that the wind filter’s way cannot recognize. Also, the validation test was performed on SMPS data. Accordingly, I think this method may be fine to publish now to give hints on ship-based data filtering. I have some minor suggestions/questions for authors’ consideration:
1. As I understood, the time resolution of BC measurements is 10 min with detection limit of 50 ng/m3. It is a bit strange to discuss/show the BC data below 1 ng/m3 (or 10 ng/m3) as shown in Figure 2 (or Figure A3). Theoretically all these points are noise. Can authors give some more explanation on why they used this parameter even below the detection limit?
2. Page 8, Line2-3, please add reference for the selection of 70ng/m3, or reference for the cases that “significant biomass burning events are absent”.
3. Page 8, Line 21, the algorithm used five minute window, while in Page 9 Line 25 “here set to 20 minutes” of the window width. I did not quite understand this extension of the window width. Can authors give some explanation or transition?
4. Page 11, Line 25, please quantitatively describe “recovering more exhaust-free data”. The algorithm in this study might recognize the exhaust data before Apr. 28, but it excluded some “clean” data (by eyes) around May 28. So it might be tricky to say it saved more data, just a bit more efficient to recognize the exhaust data.
5. Figure arrangement. Figure 4 somehow repeated Figure 3 just in a different scale and provided more cases. I would suggest only keep one in the manuscript and put the other one in supplementary information.

Dear Ruhi Humphries et al.,
thanks for the revision of your manuscript.
There are a few points that should be addressed before publication, please see below.
Best wishes,
Wiebke Frey

Major comments:
One of the reviewers still feels that you do not address the advantage of this new method over other methods enough. You do mention some advantages (e.g. minimizing false positives) in places, but I think it should be possible to add some more text to discuss and make the advantage of this method more clear. Please also add text to the conclusions, where the advantage is not mentioned at all.


Minor comments:
page 5, line 8-9Can you be more specific about the 1 Hz signal detection limit ("much higher")?

Figures:
A1 - Please indicate the periods of elevated CCN concentrations across all panels (vertical colour shadings), mention the according range of wind directions in the figure caption. This will guide the reader to the parts where data are affected.

A2: Check references to Figures, Fig. A3 (and A4, A5) are mentioned in the text before A2. Maybe a reference is missing?

A3: It would be helpful if you could additionally repeat the point you want to make with the figure in the figure caption (add information to the figure caption).

A10 and according text in manuscript: Are the instrument data time synchronised? It does not look like! If synchronised, those peaks would presumably fall on the same time. Such systematic time shifts between instruments should be identified and synchronisation applied before filtering!?!

Dear Ruhi Humphries et al.,

thanks for your revisions!
One final question regarding Figure A10 and related text:
Maybe there was some misunderstanding, I feel that the data synchronisation should also account for differences in instrument residence times (such that an event that is measured at one time shows up at the same time in all instrument data after synchronisation). In that case, the peaks would fall on the same time and the filter would work well?! Please consider revising your manuscript/commenting on this aspect in this regard.
Thanks,
Wiebke Frey