Review of Li et al. with the title: ‘Physicochemical Characterization and Source Apportionment of Arctic Ice Nucleating Particles Observed in Ny-Ålesund in Autumn 2019’

In the study of Li et al. a comprehensive campaign-based study of Ice Nucleating Particles (INPs), physicochemical properties of aerosol particles sampled in Svalbard/Arctic and their source apportionment is presented. It was found that INP composition is different compared to those of aerosol particles and hence aerosol particle composition is not an adequate indicator for INP abundance. Further, heat treatment test of INPs and fluorescence analysis suggest a biological INP origin of an INP subset. The experimental findings in general are sound although not surprisingly new. The mere fact that few atmospheric INP data are available, especially in the Arctic, and the INP sources are not well constrained, justifies the publication of the actual manuscript after careful consideration of the following comments. Generally, the addressed topic fits into the scope of the journal Atmospheric Chemistry and Physics.

General comment:

In general, it is not motivated and explained why the two different measurement sites were used to measure aerosol particles and INPs and why the different INP measurement methods DRINCZ and WT-CRAFT operated after different inlet systems. It seems to follow a randomly thrown together measurement set-up rather than a clear concept with a clear vision. I am sure that is not true, but the paper has to be revised in this regard. In addition, the applied measurement methods raise further questions as the comparison of INP concentration determined from impinger and PM10 sampling is quite different, see comment below. It might help to clearly formulate the concrete scientific questions to be answered first and then explain the approach. Furthermore, a short paragraph what is known so far about Arctic INPs would be valuable in the introduction.

Specific comments:

P2, L39: Homogeneous ice nucleation has not a fixed onset threshold, it depends on droplet size in addition to temperature. Please adapt wording accordingly. Further - more important, since this manuscript is about a comprehensive study of Ice Nucleating Particles, a brief paragraph can be written explaining how an INP is defined and which freezing modes are investigated in the actual study. 

P3, L57:  To which measure or quantity do refer ‘the range of activities as INPs’? The sentence has little meaning in this form.

P3, L59: Dust as a ‘significant terrestrial source of INP’ could also result from long-range transport as opposed to or in addition to local dust sources. Without further knowledge, this seems unbalanced in the descriptive introduction.

P5, Fig.2: Which aerosol inlet had been used at the aerosol container? PM10 or TSP inlet? Please specify.

P5, section 2.2/ P7, L173-174: A brief introducing paragraph of the complementary INP measurement methods would be helpful to motivate the need and application of the different instruments instead of a very short sentence at the end of the section. In general, statements of which freezing modes are analyzed (only immersion freezing or also deposition nucleation with HINC?), which INP concentration and temperature range can be covered with the applied methods would be needed.

P5, L108-109: It is described that the impinger is refilled with clean water continuously in order to compensate the reduced amount of water due to evaporation which would influence the sampling efficiency. This is understandable as well as method to subtract the unwanted INP background from the measurements. However, as impurities might concentrate using the described impinger sampling procedure, it is needed to present the background INP measurements in the appendix of the paper for all applied INP measurement methods for good scientific practice.

P7, L165: What is meant by ‘frost particles’?

P7, L175; P12, Fig. 3 caption, elsewhere in the manuscript: The authors use formulations like ‘freezing temperature’, ‘activation temperature’ and similar formulations when they refer to e.g. temperature dependent INP number concentrations. I suggest to use the more general formulation of ‘temperature’ because otherwise the other terms would need a clear definition in the manuscript.

P13, L323: What is meant by ‘different sensitivities of the instruments’. It needs a specification or an explanation.

P14, L335-336: The storage of the INP samples might be one explanation for the observed difference in INP conc., but I could image also other reasons for example different inlet systems, different sampling efficiencies of the different methods etc.

P14, L338-339: This conclusion is not obvious for the referee as no measurement uncertainty is given in the plot. What would be the implication? More obvious and striking is that the different methods impinger, PM10 and WT-CRAFT are quite different regarding INP concentration at the same temperature (e.g. -15°C one order of magnitude). Are the INPs collected during the same period? Partly. What are the main sampling differences? Why are these measurements compared? This needs to be more clearly described in the manuscript.

P14, L343: What is the respective lower limit for PM10? The pore size? And what is the respective upper limit for the impinger? Is anything known of the sampling characteristics in dependence of particle size?

P17, L415: What is a ‘peak case’? Do the authors refer to a higher INP number conc. or fraction? Please specify.

P18, L418 and elsewhere in the manuscript: It would be helpful to motivate the investigations or analysis which have been done and are described in each section.

P20, L448-450: I cannot agree with this statement in its current form that the heat-labile INPs would be overshadowed by heat stable INPs. It is more likely, in my opinion that no heat sensible INPs are present in that sample. Maybe the heat sensible INP are not present in the air during that period or a longer sampling period would have been required to collect even very rare INPs. In principle, it can be possible that very frequent INPs overlap with less frequent INPs and it would not be possible to discriminate between both populations except for the case that the less frequent are more active in nucleating ice, i.e., initiate freezing at higher temperature, because the most efficient (regarding temperature) INP determine the freezing of a droplet. Please rephrase this statement accordingly or clarify your statement.

P21, Fig. 9: Measurement uncertainties are missing in the plot. Please add those. To classify the PM10 and the impinger method regarding high, moderate and low INP load for comparison seems to reasonable at first. However, it distracts from the question why different methods lead to different INP concentrations and which one are more representative for atmospheric INP concentration? Which one can be used as model input data? Or is this effect an artifact from different measurement sites? But than the question arise why the measurement set-up was used as it was, i.e. measurement containers with different inlets at different locations with different instrumentation. What is the scientific question underlying this approach?

P22, Fig. 10: Can we conclude from the bulk aerosol chemical composition that different aerosols are analyzed at the 2 different measurement sites?

P25, L518: Please add a statement if long-term observations would be needed to obtain a better statistical base for investigation also regarding evaluation approached which are based on correlation methods.

Technical correction:

P2, L38: Change ‘temperatures are higher’ to ‘temperature is higher’.

P3, L62: ‘sediments from runoff’ to what area do you refer ‘runoff from the watershed’?

P6, L139: Change ‘estimate’ to ‘measure’.

P7, L163: Remove ‘Fig. 2 of’.

P20, L451: Remove ‘despite the increased labor intensity’.

P25, L510-511: What is meant by ‘high wind speed of marine origin’? Please reformulate!