Wu and colleagues present the novel application of a data treatment technique (fuzzy c-means clustering) that reduces the complexity of a data set for the interpretation of atmospheric mass spectra. I consider the exploration of novel techniques for the extraction of information from increasingly more complex and information rich mass spectra to be of interest to the atmospheric science community and the readership of AMT. The manuscript strikes a balance between being an introduction to the technique and the demonstration of the technique on an established chemical system. I consider the chosen format and presentation as appropriate for the presentation of a new approach, and can see how this manuscript could serve as a reference for future studies using the technique.

I enjoyed reading this manuscript and would welcome a publication after some minor points have been addressed.

Could you touch briefly on the practical implementation of the algorithm? I do not readily find what software you used. What is the code availability? What are typical run times of the algorithm? 

I did not check the correctness of the equations, but encourage the authors to double (or even triple) check the manuscript for typos in the equations before the final publication.

Else, I found few typographical errors only, which will be taken care of in copy-editing, and limit my comments to content observations only.

paragraph 3.1.1: unclear on what data set you "ran the FCM algorithm 50 times", please clarify

paragraph 3.1.3: expand on the accuracy of m*. Is 1.42 and 1.52 significantly different or essentially the same? Generally, what can be considered as different?

Fig.4: I was a little thrown off by the missing x-axis label on the mass profile panels. Can you add a label, and/or label the dominant species in the individual panels? Also, consider changing the time axis label to elapsed time since start of experiment. 

184: consider rewording "and only formulas within an accuracy tolerance of 10 ppm and with reasonable chemical meanings were considered." to "and only plausible formulas with relative m/z deviations smaller than 10 ppm were considered"

290: Incomplete sentence. Missing "be"? 

300: consider rewording "the right choice" to "may not always be appropriate"

428: "mathematically unsolvable". Probably better to say that there is no simple analytical solution.

467: punishing function? while punishing seems to be used in some literature, it should maybe rather be penalty instead? 

493: reword "looks reasonable"

512: drop "As a quick reminder"

699: "mathematically"? Consider making the paragraph (especially lines 699-701) more concise. 

736: marker size: I appreciate the attempt to be specific about what the marker size represents. Please be fully specific by referring to the marker area or the diameter. This comment applies to a couple more figures in the main text and SI.

Fig. 11: Units for k

Supplementary information

page 2: proposed >the< Kwon index

page 2: punishing function, same as main text

page 4: it's --> it is

Fig S1: axis labels cut off

Fig S2: subscripts inO3, NO2, NO3

Fig S4: same comments as main text figure

Please, find my detailed review report in the attached pdf.