The manuscript presents a comprehensive study in southern Greenland of ice nucleating particles (INPs) in surface material of glacial outwash plains. In addition, aerosol particles sampled at several locations on and near the plains provide for a thorough assessment of outwash plains as potential sources of atmospheric INPs. Analyses of sampled material include heat and hydrogen peroxide treatments to distinguish between different kinds of INPs. Various parameters, including organic carbon concentration, bacterial abundance and soil particle mineralogy provide further insights on the likely origin of INPs in outwash material. It is a technically sound study describing in great detail (28 pages main text, 73 pages in total) work that has led to new insights. Care has been taken to relate obtained results to those of similar published studies and to provide possible explanations for particular findings.

My reading of the results is that glacial outwash dust is at best a minor source of atmospheric INPs in southern Greenland during summer, for three reasons. First, the PCA analysis (Fig. 9) of cumulative INP spectra suggests that most filter samples (atmospheric INPs) contained an INP population different from bulk samples (glacial outwash dust INPs). Second, neither aerosol particle number (> 0.5 um) nor INP concentration increased at high wind speed (Fig. 8), perhaps not so surprising as Figure B1 shows biological crusts on soil surfaces. Such crusts efficiently suppress wind erosion (Zhang et al., 2006). Although it takes in the Arctic two centuries for a biological crust to develop fully, intermediate levels of development already form within a few decades (Heindel et al., 2019; Tanner, 2025). Only at Narsasuaq plain, where the surface is apparently crust-free sand and silt, as indicated by footprints left in loose material around the solar panels (Figure B1c), there is some overlap between filter and bulk samples along the PC1 axis in Fig. 9. Third, the strongest correlation of dust INPs was with colony forming units (CFU), i.e., viable microorganisms. These thrive better on biological surfaces than on loose glacial outwash material. Biological crusts, vegetation, and leaf litter seem to cover a substantial fraction of the surrounding land area, e.g., at Narsarsuaq col and NIRS (Fig B1c,). Taken together, the evidence suggests that dust from glacial outwash plains is not a major source of atmospheric INPs above these plains in summer, except perhaps within the internal boundary layer (e.g., Fig. 1 in Dupont et al., 2021) above larger patches of bare outwash material (Narsarsuaq plain). From my point of view, this outcome should guide future studies on atmospheric INPs towards focusing more on biological surfaces. Although there is tentative pointing in this direction (lines 512 and 513, 517 to 519, 533 to 535), I would recommend to put more emphasis on this issue.

Minor issues

I would move Tables 1 and 2 to the Appendix.

Line 201: "approximately" is not necessary because 98.4 °C is a precise temperature value.

Line 290: The choice of -15 °C as a good one is further supported by the findings of Hanna et al. (2008).

Lines 321/322: "... higher organic fractions than Tobo et al. (2019)...", change to "...higher organic fractions than reported by Tobo et al. (2019)..."

Lines 341/342: The statement that " ice-nucleating efficiency of TOC may differ across climatic regions" could be supported by referring to Schnell and Vali (1973).

Correlation analysis of INP with soil parameters is convincing in numbers. However, in Figure 5 only panel b (CFU) displays a visually convincing correlation. Perhaps, it is in the cluster of data points with low values that similar patterns are hidden in panels a and c? Transforming the scale of the x-axis to log-scale might render correlations visible.

In Figures 2, 6, and B12 there is a lot of overlap between symbols. Consider using open symbols to reduce cover up.

Figure B11: The morphology of the particles in panels a and b is not as typical for mineral dust as is that in panel c. Are EDX spectra available for these particles?

References

Dupont et al., 2021, https://doi.org/10.1029/2021JD034735

Hanna et al., 2008, https://doi.org/10.1175/2007JAMC1549.1

Heindel et al., 2019, https://doi.org/10.1007/s10021-018-0267-8

Schnell and Vali, 1973, https://doi.org/10.1038/246212a0.

Tanner, 2025, https://doi.org/10.3390/land14091827

Zhang et al., 2006, https://doi.org/10.1016/j.geod

Review of “Ice-nucleating particles in Greenlandic glacial outwash plains” by Bergner et al.

The origin of ice-nucleating particles (INPs) in Greenland is investigated. Ice formation on dust samples collected from the ground and from airborne aerosol samples is measured using the new SPICE setup. Heat and peroxide treatments are applied to the samples to distinguish between organic and inorganic INP fractions. Ancillary analyses include TOC, mineralogy, BET surface area, microbial abundance, SEM, aerosol size distributions, meteorological data, back trajectories, specific events (Foehn, precipitation, and wildfire plumes), as well as temporal and spatial variation. Measurement results are compared with literature data and with samples collected in the Swiss Alps.

General comments:

Two main conclusions are, lower ice nucleation activity of Greenlandic dust sources compared to other HLD sources, and a dominant contribution, likely from biological substances to the ice nucleation activity. Both conclusions would benefit from further verification, taking the following aspects into consideration.

The lower ice-active surface site densities may be influenced by a normalization artefact. When comparing n_{s, geo}, normalized by geometric surface area, with n_{s, BET} normalized by BET surface area, n_{s, BET} is generally lower than n_{s, geo}, even for spherical particles (see, for example, Hiranuma et al., 2014, Fig. 3). For non-spherical dust particles, the difference can reach up to two orders of magnitude.

There are several aspects that may complicate the second conclusion. Dust constitutes most of the surface area of the ground samples and is known to be ice-nucleation active in the tested temperature range, and therefore likely contributes substantially more to the observed ice formation activity than small amounts of biological substances. Ambient INP concentrations can be parameterized using D15 for dust. Furthermore, the evaluation of the heat test in Daily et al. (2022) indicates that, for the minerals present in the Greenland ground samples, heat treatments can produce false positives by deactivating the minerals ice nucleation activity. In addition, peroxide treatment increases pH, and it has been shown that acid treatments of dust decrease ice nucleation activity. These aspects could be investigated more explicitly in relation to the second conclusion.

The manuscript presents a large amount of material (data dump), including data that are only marginally related to the topic of the paper (e.g., Swiss Alps samples and SPICE performance validation). In addition, many ancillary analyses are presented but not fully developed into robust findings.

I recommend major revisions of the manuscript because the material would benefit from substantial trimming and deeper analysis to support more robust and clearly articulated conclusions.

Specific comments:

Move Appendix B and C to a Supplement. Appendices should be understandable on their own. Appendix A meets this requirement.

Line 70, 72: Define the units of “ice-nucleation activity” referred to.

Line 98: Specify what is meant by “sources” in this context.

Table 1 and Table 2: Consider moving to the Supplementary Material. Add sample volumes to Table 2.

Line 150–151: The size ranges of the OPCs are very specific and differ from those in the manuals. Please clarify how they were determined.

Section 2.2: This section contains repetitions from Appendix A1. Consider referring to Appendix A1 for general information and highlighting only aspects specific to the experiments presented here.

Line 165–166: Specify how much dust is used per sample and how many 10-fold dilutions are prepared (as described in Appendix A1).

Line 168: Clarify why 8 mL instead of 12 mL are used to wash procedural filters. Using the same volume for background estimation could avoid the need for scaling.

Line 169: Specify that 15-fold and 225-fold dilutions are measured in addition to the undiluted suspension.

Line 171: Specify how the 192 PCR wells are distributed among samples and treatments.

Line 184: Provide the BET surface areas of the transect samples.

Line 193–194: Explain why the water background was used instead of the procedural filter handling background. Testing the difference would help confirm that the contribution is minor.

Line 198–199: Daily et al. (2022) may not be the most supporting reference for this statement, as it suggests nuances. Consider Hill et al. (2016) instead.

Section 2.4: Explain in more detail why a component analysis of slope and log concentration ratio provides information on sample similarities. Differential INP spectra could be considered as an alternative approach.

Line 250: Explain the grain size effect.

Line 265–266: The percentages used for Fig. 2 do not appear to be averages calculated from Table B1. Instead of an average, consider providing a range.

Line 266: The surface area of K-feldspar is constant along the yellow line in Fig. 2(d). The difference in slope suggests that K-feldspar does not affect the temperature spectra of the dust samples at any temperature.

Line 272: Consider using the XRD results to further investigate differences in sample activity.

Figure 2: The BET surface area of dust can be up to two orders of magnitude larger than the geometric surface area. For a consistent comparison of n_s, either BET or geometric surface area should be used. The data in panel (d) may align with desert dust parameterizations when using a consistent surface definition. The polynomial fits could be removed, as they are not used in subsequent analysis. Consider taking specifications out of the citations.

Sections 3.1.2, 3.2.2, 3.2.3: Consider moving these sections to the Supplementary Material, as they do not directly contribute to the main conclusions.

Line 314, Fig. B4: Most samples show an increasing heat-stable fraction from −15°C to −20°C, but sample N1GT shows the opposite behaviour. A more detailed analysis of such features would be helpful.

Line 316, Fig. B5: Clarify how Fig. B5 was constructed. The colour coding appears opposite to Fig. 4 and Fig. B4. Using consistent colours would improve clarity. Clarify whether the peroxide resistant fraction is too small to be visible. Use consistent labels in captions and legends (also for Fig. B13).

Line 317: Suppression of dust activity by treatment can be expected (see Daily et al., 2022). Consider incorporating this into the interpretation.

Line 320: The only significant correlations are weak and negative. Please provide an interpretation.

Line 323: ΔT₅₀ values in Barr et al. (2023) range from −1.6°C to −3.6°C, while Fig. B8 shows values down to −6°C (closer to Daily et al., 2022). Please discuss possible reasons for this difference.

Line 327: As the previous section suggests that the type and composition of organic matter are relevant, it could be noted that TOC is not sensitive to these differences.

Line 332, Fig. B5d: Clarify why sample K3 is used as an example.

Line 332: Explain the rationale and hypothesis behind this analysis.

Line 354: Microbial abundance, including the presence of Pseudomonas, does not necessarily imply a contribution to ice nucleation if ice-nucleating proteins are not expressed.

Line 359: Specify which organic compounds are ice-nucleating.

Line 396–398: Clarify how much of the difference in INP concentration is due to sample volume limiting detectable concentrations.

Line 400–401: Explain how this conclusion is reached.

Line 401–402: This appears to repeat previous statements and could be streamlined.

Line 407: Provide supporting references for INPs emitted by each mentioned source.

Line 410: Clarify why increases could be attributed to pollen and whether pollen concentrations at the site would be sufficiently high (>0.1 L⁻¹).

Line 421: As McCluskey et al. report observations from the Southern Ocean, comparison with Arctic marine concentrations may be more appropriate.

Line 425: Explain how the >0.5 µm concentration was determined.

Line 439: Clarify how the ice-nucleation mechanisms differ; both appear to involve immersion freezing.

Figure 6(a): INP concentrations in a cumulative spectrum should not decrease toward lower temperatures. Please clarify the cause of this artefact.

Line 461: Consider “low INP content” instead of “ice-nucleation efficiency.”

Line 487: Specify which differences suggest a local influence.

Line 489: Explain why slopes and ratios in 2°C bins were chosen.

Line 494: Specify the original variables.

Line 496: The clusters are not clearly visible; consider highlighting them.

Line 498–500: Explain how the analysis supports this interpretation. A more detailed discussion of clusters, separations, and outliers would be helpful.

Line 510–513: The statements are cautious; but consider clarifying which findings are well supported and which remain speculative.

Line 645: Quality check 3 enforces the assumption of Poisson-distributed INPs. If data fail this check, clarify whether this indicates experimental issues or contains useful information.

Line 659: As experiments are conducted at even faster temperature changes and down to -35°C, the calibration should be performed mirroring these specifications.

Line 695: Define minimum and maximum deviation and clarify whether a temperature gradient exists across the setup.

Line 699: Quantify how sensitive temperature offset and variability are to dry air flow.

Line 703: Reporting maximum temperature spread may be more informative than standard deviation.

Line 724: Clarify that filtered airflow refers to airflow over the pane.

Eq. A9: Clarify whether uncertainty includes the range between lowest and highest sample temperatures.

Line 748: Clarify how the detection limit is applied, especially since measurements below −24.71°C are reported.

Figure B12: n_{s, geo} is expected to be larger than n_{s, BET}, even for near-spherical particles (e.g., Hiranuma et al., 2014). Direct comparison should be avoided.

Technical corrections:

Line 130: Sampling sites are shown in Fig. C1, not C2.

Line 152: Handix

Line 156: Should it be filter “samplers”

Line 323: Fig. 3 in Barr et al. (2023), not Fig. 4.

Figure 8: The figure should fit on one page (print out) including the caption. Remove or move panel titles inside the boxes. (a) bar plots on a log-scale are misleading as their area does not reflect the values. Consider removing the bars.

Line 718: Fig. A5c

Line 726: Fig. A5a

Table C1: The coordinates appear inconsistent. L2a and L2b should have the same coordinates, coordinates from Ferpècle are all the same as L2b, LG2 is 48°. Please verify.

Figure C4 appears identical to the top panel of Figure C2.

References:

Barr, S. L., Wyld, B., McQuaid, J. B., Neely Iii, R. R., and Murray, B. J.: Southern Alaska as a source of atmospheric mineral dust and ice-nucleating particles, Sci. Adv., 9, eadg3708, https://doi.org/10.1126/sciadv.adg3708, 2023.

Daily, M. I., Tarn, M. D., Whale, T. F., and Murray, B. J.: An evaluation of the heat test for the ice-nucleating ability of minerals and biological material, Atmos. Meas. Tech., 15, 2635–2665, https://doi.org/10.5194/amt-15-2635-2022, 2022.

Hill, T. C. J., DeMott, P. J., Tobo, Y., Fröhlich-Nowoisky, J., Moffett, B. F., Franc, G. D., and Kreidenweis, S. M.: Sources of organic ice nucleating particles in soils, Atmos. Chem. Phys., 16, 7195–7211, https://doi.org/10.5194/acp-16-7195-2016, 2016.

Hiranuma, N., Hoffmann, N., Kiselev, A., Dreyer, A., Zhang, K., Kulkarni, G., Koop, T., and Möhler, O.: Influence of surface morphology on the immersion mode ice nucleation efficiency of hematite particles, Atmos. Chem. Phys., 14, 2315–2324, https://doi.org/10.5194/acp-14-2315-2014, 2014.