Dear Authors,

Thank you for your revised manuscript, which has largely addressed the questions and concerns raised by the referees. Please review the further comments and suggestions provided by Gabor Vali in his most recent referee report, particularly concerning the df/dT analysis. In accordance, please further revise your manuscript to address his comments, or explain why you disagree with the suggested change.

Best regards,
Ryan Sullivan
ACP Co-Editor

Dear Authors,

Thank you for your further revisions to your manuscript. There was a significant disagreement regarding the use of the df/dT analysis and the interpretation of the "cold mode" that appears in the df/dT spectrum between yourselves and the referee Gabor Vali. The referee made strong arguments against the use of the df/dT analysis and that the "cold mode" that appears in this spectrum is artificial. In your response you agree that the cold mode is artificial, yet continue to use the df/dT analysis and did not sufficiently revise your manuscript to address the issues with the use of this flawed analysis.

To help resolve this disagreement I obtained a third referee's opinion regarding the df/dT analysis and the interpretation of this apparent "cold mode". By analyzing a synthetic dataset using both k(T) and df/dT approaches, as well as analysis of the underlying equations, the referee convincingly demonstrates that this df/dT analysis is flawed, for the same reasons provided by Gabor Vali. Incorrect analyses cannot be included in manuscripts to be published in ACP as this will lead to further proliferation of erroneous analysis methods and interpretations of data. The k(T) spectra already provide the analysis necessary for the scientific goals of this field study. Both referees have provided entirely convincing reasons for the artificial and erroneous nature of the df/dT analysis and clear arguments against its use.

Therefore, the df/dT analysis and its interpretation and discussion must be removed from the manuscript entirely. The k(T) analysis is robust and already demonstrates the results required to support the analysis and conclusions presented here.

Please carefully consult the reports from Gabor Vali and Referee #3, and remove all use and discussion of the erroneous df/dT analysis. Should this critical issue be satisfactorily addressed in your revisions, your manuscript will then likely be suitable for publication in ACP.

Best regards,
Ryan Sullivan
ACP Co-Editor

Dear Jessie,

Thank you for submitting your further revised manuscript that removes the erroneous df/dT analysis.

In reading through your manuscript I noted that there are some missing key references that would further support your points regarding separating the "warm" biological IN mode from the colder mineral IN mode. Also the extension of the k(T) analysis to colder temperatures by Vali (2018) was actually made possible by the use of data collected using a microfluidic device, reported by Polen et al. It was that paper that actually caused Vali to write the tutorial in AMT regarding the use proper use of k(T). I am sorry that both of these papers are from my research as I am not trying to emphasize my work here, but these two papers really do support your arguments, and it is important to properly explain what lead to the recently renewed interest in the k(T) analysis. I had pointed these two papers out in my initial Editor's review of your paper for ACPD, which I repeat here:

Analyzing the spectral characteristics of droplet freezing curves to separate the influence of biological and mineral ice nucleants was the subject of Beydoun et al. They showed that varying the amounts of Snomax or illite minerals changes the complex freezing spectrum in a /predictable/ manner. This should be discussed in Sect. 3.2 of your manuscript, as it demonstrates that mixtures of biological and mineral particles have been studied and through careful analysis you can indeed resolve the two modes.

Beydoun, H., Polen, M. and Sullivan, R. C.: A new multicomponent heterogeneous ice nucleation model and its application to Snomax bacterial particles and a Snomax–illite mineral particle mixture, Atmos. Chem. Phys., 17(22), 13545–13557, doi:10.5194/acp-17-13545-2017, 2017.


The paper focused on reducing background freezing interference that contained the dataset used by Vali (2018) for the k(T) analysis that extended to much lower temperatures than achieved previously is below. This should be discussed (briefly) along with Vali (2018) in Sect. 2.2.

Polen, M., Brubaker, T., Somers, J. and Sullivan, R. C.: Cleaning up our water: reducing interferences from nonhomogeneous freezing of “pure” water in droplet freezing assays of ice-nucleating particles, Atmos. Meas. Tech., 11(9), 5315–5334, doi:10.5194/amt-11-5315-2018, 2018.


This was my original comment regarding this paper and suggesting the use of the k(T) analysis:

In our recent paper focusing on reducing background interferences in droplet freezing assays, Gabor Vali suggested using the differential ice nucleus analysis, as you have down here. When we tried this our spectra looked very poor, and we engaged with a detailed discussion with Vali regarding this. He realized that the choice of the temperature interval is key here. It must be tuned to minimize the number of zeros retrieved, while still retaining the high resolution spectral features. This was an aspect not discussed in the original Vali 1971 paper. Some of these aspects are discussed in Polen et al., and in the discussion of this paper. Vali plans to submit a tutorial on the use of the differential ice nucleus analysis to AMT shortly, so do keep an eye for that.


If you can make those two additions then your paper can finally be accepted for publication in ACP.

Best regards,
Ryan Sullivan