General Comments

This is a very interesting study using the automated HINC instrument, which is packed with useful results and honest (refreshing) caveats, limited only perhaps by the assessment being made at a relatively low temperature (243K, for conditions that emphasize water droplet co-activation and expected immersion freezing). That is a modest critical comment, really, as it does not terribly limit the insights reflected in the deep analyses conducted. The study reveals important links between ice nucleation at lower temperatures and Saharan dust transports in this region, an apparently ubiquitous and strong influence, made further relevant by the observation that higher altitude dust appears to be drawn down via virga following likely heterogeneous ice formation at cirrus levels. The authors emphasize that this raises the potential issue of pre-activation and the known inability of most INP instrumentation for assessing this without special preconditioning of sampled air (needs more discussion). I wonder if this behavior does not also point to an important potential link between dust and precipitation in storms where deep clouds form over the region under dust transports. Precedence for this occurrence already exists in the literature over other regions that are more distant from major mineral dust sources. If so, transported dust INPs at high altitudes could more importantly impact precipitation (via ice sedimentation effects) than one might guess based on their characteristics of activating only at lower temperature, versus some other INP types that are imagined to have a greater impact on mixed-phase clouds due to their higher activation temperatures. I feel that a potential missed opportunity in this paper is relating the INP data more directly to aerosol size distribution measurements. An important relation to aerosol surface area is only inferred in the paper, and indirectly supported via the relation shown with attenuated backscatter. Closure with actual measured aerosol surface area to show relation to INPs could be useful (not required, just a suggestion). Presently it is only stated that some of the surface area (35% estimate) is missed by the measurement limitations on capturing all sizes. That seems a minor amount to miss, something that I think should be emphasized as a positive aspect of the real-time, automated methods.  I have only an assortment of other specific comments added to this, which I do below in order of appearance.

Specific Comments

Abstract

Line 4: I suggest that this statement is not true in general, and so requires more generalization without attribution of a specific temperature. There are many reports of other influences on ice formation dominant at this temperature that are not only mineral dusts. For example, also soil dusts, which made me think about whether (or why not) this site does or does not see influence for more local/regional soil sources.

Line 8: I suggest to mention the method for derivation of Ångström exponent here (nephelometer), just as satellite retrieval is mentioned for mass concentrations.

Line 15: Can you distinguish mineral from soil dust particles (due to attribution scheme for SDEs)?

Introduction

Line 52: This is obviously not critical, but just to note that DeMott et al. (2003) was revised in 2009 to indicate a correction to maximum INP concentrations, reducing them to a few hundred per liter (https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2009GL037639).

Lines 52-55: I feel it is necessary to say that Zhao et al. was a modeling study, depending both on proper modeling of aerosols and parameterization of INPs.

A general comment on the introduction: It could be better introduced how you plan to identify SDEs specifically.

Methods

Lines 92-93: “There is no appreciable natural source of mineral dust in proximity of the site.” Is it true that there are no regional sources through the year of other dusts, such as agriculture?

Lines 109-110: Similarly, “If the plumes appeared not to originate from within the planetary

boundary layer (PBL), the period for which the plume is observed between 3200 - 4000 m a.s.l. is marked as SDE.” Is it always so that origination in the boundary layer is so easily determined, for the entire history of a parcel? This statement confused me because it becomes clear later in the paper that SDEs entail air reaching the site both from the boundary layer and the free troposphere.

Section 2.2.2: Could you add to say a little more about the Ångström exponent calculation, and why a negative value clearly marks an SDE? I think it would help the general reader.

Lines 174-179: This is an important discussion, perhaps for reasons that are not highlighted. Although there is size limitation on transmission of particles to the HINC, the losses of INPs are not so great. That is, if indeed mineral dust INPs scale with surface area, then a 35% loss is not so significant in consideration that different INP measurement methods easily disagree at times by numeric factors up to 10. Hence, it would be great to see documentation of this statement with a size distribution typical of SDE periods. Not a requirement, but a suggestion. And could you also state if an upstream impactor is used for the HINC or not as part of INP detection?

General comment on Section 2.3: The nature of the correction for frost background is unclear. Simple subtraction? Counts exceeding the 1 sigma value of the filtered air value? And in general, it would be useful to repeat how INPs are distinguished from other large particles in the HINC, especially during SDEs. I realize that an entire instrument paper was recently published, but some of these things bear repeating in short form.

Fig. 4: Would one infer from this figure that the SDE encompassed primarily boundary layer air transport? Again, I was confused because there was such an emphasis placed on indicating FT origin, and I was expecting that FT would encompass most SDE. Also, in this Figure caption and in the main manuscript text, I suggest writing out the meaning of hcSDE and lcSDE. Please note, I could not find anywhere else in the paper where these were specifically defined. That is, what makes for attributing hcSDE versus lcSDE?  

Line 250: Particulate matter means mass concentration? Or total counts in a certain size range?

Section 2 general comment: It seems like the CAMS, FLEXPART modeling and CALIOP data are the key to assuredly attributing dust to a Saharan source. That is, an aerosol layer in ceilometer data and Angstrom exponent will only indicate larger particles present from somewhere. I am curious if there were ever indications of arable dust sources reaching JFJ. It seems common in other locales, including mountaintop sites that are perhaps not so elevated. Unfortunately, I cannot offer a reference for that because the separate works I am aware of are still in preparation. It is simply an honest question about whether any such influences were inferred during the non-SDE periods or if any lcSDE events could have captured such.

Results

Line 338-339: I appreciate the strong qualifications added to the conclusion on the major role of mineral dust outside of SDEs for the conditions examined, but I wonder if somewhere here or in the conclusions you could speculate on what other types of measurements (or extensions) could be done to support the major role of mineral dust that you hypothesize here for this site (and by proxy perhaps, these altitudes over the region)? I say that because, in boundary layer measurements, my impression is that mineral dust cannot be assumed as the only influence at 243K, though I will not list the references.

Lines 355-356: Do natural INP concentrations scale with aerosol surface area? Were particle distribution data co-collected to test this assumption for this situation that appears to reflect one primary INP type? Does CAMS provide such predicted information or is their size distribution fixed?

Line 434: Regarding these results apparently linking cirrus virga to descent of mineral dust plumes to levels where they can impact ice formation at lower levels, the mystery to me is why the virga connect to regions of apparent higher dust concentrations than ever seem to appear aloft. Is it the case that the higher regions are more typically shrouded in clouds because ice will so readily nucleate ice there when ice supersaturated conditions exist? It may be useful here to note that during a study intended to assess Asian dust impacts on INPs at a mountain site in the western United States, Richardson et al. (2007) inferred that dust frequently was concentrated well aloft on the basis of aerosol modeling (c.f., Fig. 6 of that study), and often this was predicted to descend over time although connection to virga was not possible to investigate. It seems possible that dust descends with the virga in your case as well, perhaps not only as activated ice crystals? Hence, it may not all represent pre-activated particle populations. However, other studies have indicated the critical role of elevated dust plumes on precipitation formation under circumstances where that occurs (Creamean et al., 2013). The seeding impact from higher altitudes with longer inherent ice growth times and fall velocities could play a critical role in precipitation for orographic locations. I think your data suggests that this could be common due to the ubiquitous presence of dust aloft and its propensity for freezing there.

Lines 440-442: You say as much in the abstract, but I expected a few more words here. It is possible I suppose to explore this topic in a laboratory setting, or by exposing instruments to the natural environment. I suspect that unreported efforts have already been attempted. It will be immensely more difficult to do this in an aircraft setting, where the actual nature of the inlet outside the aircraft would have to be modified in order to not heat particles due to pressure. In any case, I think you should expand the discussion a little bit, for the sake of those who may not have deeply considered how this can be done (or how hard it may be).

Editorial notes

Line 25: It is unusual to start a sentence with a number. Perhaps write it out.

Line 31: cover clouds  → do clouds cover

Line 69: Suggest to spell out 20 “minutes” for the INP measurement time resolution.

Line 90: hosts → has hosted

Line 119: previously is misspelled.

Line 127: with is misspelled.

Line 272: where → were

Line 425: Probably do not need the word “for”.

Line 461: Suggest rewrite, e.g., “and sediment to lower altitudes where they sublimate in drier air…”

Line 463: loose → lose

References

Creamean, J. M., K. J. Suski, D. Rosenfeld, A. Cazorla, P. J. DeMott, R. C. Sullivan, A. B. White, F. M. Ralph, P. Minnis, J. M. Comstock, J. M. Tomlinson, and K. A. Prather, 2013: Dust and Biological Aerosols from the Sahara and Asia Influence Precipitation in the Western United States,  Science, 339, 1572-1578.

Richardson, M. S., P. J. DeMott, S. M. Kreidenweis, D. J. Cziczo, E. Dunlea, J. L. Jimenez, D. S. Thompson, L. L. Ashbaugh, R. D. Borys, D. S. Westphal, G. S. Cassucio and T. L. Lersch, 2007: Measurements of heterogeneous ice nuclei in the Western U.S. in springtime and their relation to aerosol characteristics. J. Geophys. Res., 112, D02209, doi:10.1029/2006JD007500.

The authors conduct measurements of INP from February 2020 to December 2020 at JFJ.  The INP measurements are constrained to a temperature of 243K and a saturation ratio of 1.04. They classify the INP according to whether or not Saharan dust events were present.  The classification is based on four criteria: single scattering albedo; satellite retrievals of dust mass concentration; modelled tropospheric residence times; the backscatter signal from a ceilometer.   14 dust events of high confidence (hcSDEs) were classified where each of the four criteria for Saharan dust were met and 12 events of lower confidence (lcSDEs) were classified where at least one of the four criteria were met.  The authors show that INP concentrations increase by generally one to two orders of magnitude during the periods of dust events.  They also find some evidence for dust influence in the absence of an identified event.  I find the main aspects of the work to be sound and useful.  I think the interpretation of the results, initially sound, is carried a bit farther than warranted, reflected in my comments 12-17 below.  Overall, I feel the paper could be suitable for publication subject to some revisions.

Comments:

1) Abstract – lines 3-4 – These two sentences might be better reversed in order, as I find the second seems to contradict the first.

2) Lines 33-35 – Again, slightly contradictory statements: Mixed phase exist between 273K and 235K, yet most clouds warmer than 253-258K are ice free.

3) Line 41 – I suggest mentioning the importance of mineral dust as INP before discussing the sources and transport of dust.

4) Line 73 – Remove “Besides”

5) Line 250 – What metric of particulate matter is recorded?

6) Line 256 – Please clarify what you mean by “and involved microphysics of dust”.

7) Lines 264-270 – Perhaps clarify that the median of all single events is the median of the individual event medians, and that the collective refers to the median of all INP concentrations during the SDEs.

8) Lines 277-279 - They clearly differ based on the stated uncertainties, yet you say the difference is not significant.  Please explain.

9) Lines 297-298 – With only 12 lcSDE cases, the statistics for these events cannot be strong, unless one criterion was dominant.

10) Lines 310-311 - Maybe, you can't say this is true, or even implied, without some sort of chemical ID.

11) Figure 6b – You mention having normalized the area under each curve.  I suggest adding that this does not allow the “SDE” plus “no SDE” curves to equate to “all”.

12) Lines 334-339 – This a simple back-of-the-envelope calculation. The inclusion of uncertainties of 0.1% and 0.2% suggests otherwise and I think is inappropriate.  I suggest reducing this discussion to something like "With our assumptions, we estimate that about 23% of the INPs measured during non-SDE periods were dust-related."

13) Lines 351-363 - The potential for correlation of INP with dustcams is based on Figure 6 showing consistency between the INP and dustcams distribution.  However, the dustcams distribution for the SDE cases does not exhibit a log normal, which suggests that the sizes of dust particles vary, perhaps substantially.  One consequence of that is the number concentrations of dust particles will not necessarily scale with dust mass.  The authors note that INP scales with r^2, but that is largely related to the process of ice nucleation.  Actual INP concentrations likely also scale with simply r or just the number of viable dust particles.  Significant improvement of this discussion is needed.

14) Line 378 and Figure 9 – The “R^2” in Figure 9c looks to be 0.826.  Is that correct?

15) Lines 393-394 - Please explain the evidence for a connection between the dust and the virga, which to me looks tenuous at best.  Are you suggesting that the dust is responsible for the virga or that the dust may be modifying it?

16) Line 415 – Where are the dust particles between 10200 m and 11200 m?  It looks like noise above 8 km.

17) Lines 420-442 and Figure 12 - I don't see any evidence for lifting of the dust above 7 km at best.  Why can't the "connection" be simply the virga settling until it hits the drier air containing the dust? The virga does not correlate with the dust, which suggests that, if the dust is involved, it may only modify the virga.  One might expect more and smaller ice crystals associated with more INP, which might result in a shorter fall streak.  However, the shortest fall streaks appear to be farther from the dust and completely unconnected.

This study provides interesting insights into the linkage between mineral dusts from the Sahara and ice nucleation activity at low temperature at the Jungfraujoch station. By using a wide array of instruments, techniques, and model data the authors were able to assign Sahara dust events and link it to INP concentrations.  Even though many assumptions were necessary for the data curation, the authors clearly stated the limitations of each technique. The authors further hypothesize possible dust transport pathways within the atmosphere. Generally, the article is well written, concise and the results are of general interest for the scientific community. The provided data will improve the knowledge of the field and are important for further model studies on clouds and climate. However, I suggest minor revisions before publication:

General comments:

Introduction:

Can you explain in the introduction in more detail why you have chosen 243 K and Sw of 1.04 as the parameters for your measurements? I assume it is due to general mixed phase conditions. However, why is this particular temperature area so important for ice nucleation including mineral dust and not e.g. 250K? Please elucidate in more detail so a general reader can better follow the content.

Materials and Methods:

Line 116: I think it would improve the manuscript and makes it clearer to the reader if you could explain the Single scattering albedo Angström exponent in more detail.

Line 203: The classification of air masses comes a little bit abrupt. Could you mention that at the of the introduction where you summarize the investigations? (just a suggestion)

Figure 4: The figure gives a good overview of the collected data to assign a dust event and compare it to INP concentrations. The caption reveals much information. However, I miss an explanation of the figure in the continuous text.

Results:

The results point out that further characterizations are necessary in order to estimate the role of mineral dust and other INPs on microphysical processes in the atmosphere. You mentioned in the manuscript in one sentence (line 321) that you cannot draw any conclusions on the influence of biogenic coating. However, I wonder whether this would be an important information? I think the manuscript would benefit if this would be discussed and commented in more detail. In addition, how could you improve future studies to target this issue?

Technical Comments:

Line 82: Jungfraujoch (JFJ) – abbreviation was already introduced above

Line 88: space is missing between the number and the unit (check throughout the whole paper)

Line 89: m/s or m s^-1 – only use one style

Line 119: […] more ‘than’ 6 consecutive […]

Line 119: This is longer than the ‘previously’ […] (?)

Line 127: […] MERRA-2 ‘with’ […]

Line 135/136: meters were abbreviated before

Line 227: introduce the abbreviations and delete them below (line 231/232); Carbon dioxide (CO) and reactive nitrogen (NOy)

Figure 4 caption: superscript ‘222’-Radon

Line 253: ‘85°’ is missing a degree sign

Table 1: All INP ‘concentrations’ missing the letter s in the end

Line 302: Figure or Fig.? chose one style

Line 349: I guess at the end of the sentence it should be ‘can’?