See attached file.

Review of Light-absorbing snow impurities: Nine years (2016-2024) of snowpack sampling close to Sonnblick Observatory, Austrian Alps by Daniela Kau, Marion Greilinger, Andjela Vukićević, Jakub Bielecki, Laura Kronlachner, and Anne Kasper-Giebl

Overview.

It is apparent that the authors are experts in analysing this type of samples, and I can find little to criticise in the methods and protocols used which seem very professional. I think the manuscript can be improved by extending the discussion of the results to describe what is novel and significant in this data set, and what the new results mean in the context of the broader time series of measurements that have been carried out at this site. The paper would benefit from a more cautious interpretation of uncertainties, and by making method comparisons more quantitative.

Scientific.

The paper presents an impressive nine-year continuous dataset and introduces a TOA-based method for estimating mineral dust deposition using Fe as a proxy. These are significant contributions. However, the introduction never fully clarifies what gap in the literature this study directly fills. The authors mention missing comprehensive discussions (e.g., mineral-dust bias in EC quantification), but don’t articulate: 1) Why earlier studies (e.g., Cerqueira 2010; Tuzet 2020) were insufficient. 2) How this work advances monitoring practice or radiative forcing estimates. I recommend adding a short “This study addresses three gaps…” paragraph at the end of the Introduction to anchor the novelty.

Section 3.3 introduces an Fe-based approximation for mineral dust loading from TOA data. While elegant and practical, several limitations appear under-discussed. First, the limited calibration dataset. Only 14 samples were used to derive the ATN–Fe fit (Eq. 1) and the valid range is stated as ATN = 4–13. Yet the paper applies the relationship throughout nine years of snowpacks, without showing how often values fall outside this range. Please provide a histogram of ATN700–450 values for all mineral dust layers to show how representative the calibration domain is. Second, the assumption of constant Fe fraction. The method assumes 4% Fe by mass in all long-range dust, but literature and the authors' own comparison (Figure 5) show Fe can vary from ~2% to >11% depending on source and transport history. Please discuss how source-region variability (e.g., Saharan vs. Middle Eastern dust) could bias the long-term trend analysis. Third, the uncertainty is stated as 65% for mineral dust estimates (Section 3.3.2). This is substantial. A sensitivity test showing whether interannual variability is still statistically meaningful under such large uncertainty would strengthen confidence in the conclusions.

Figure 1 provides a qualitative visual comparison of the TOA vs IC method comparison but the text states “very good agreement” without any numerical metric. Please include agreement statistics e.g. percentage of overlapping layer identifications, Cohen’s kappa, number of false positives/negatives per method. This would make the comparison more rigorous.

Light-absorbing impurities (LASI) strongly affect albedo and melt timing. The Introduction opens with this motivation, but the Discussion does not return to these impacts -- the paper remains methodologically focused. Please add a paragraph estimating or contextualizing the radiative significance of the observed impurity levels (e.g., based on Di Mauro 2019).

The manuscript discusses the issues arising from incomplete snowpack sampling in 2019 (3.2.2 and 3.3.2). However, results for that year are still plotted without any visual cue of lower confidence and results from 2019 are included in averages. Please mark 2019 values in plots as low-confidence (distinct color or hatching) and consider excluding 2019 from interannual statistics, or provide results with/without 2019.

Appendix A is very informative and somewhat buried. It explains convincingly why coagulants were rejected, but some conclusions are speculative without quantitative data. Please summarize key quantitative changes (e.g., how much EC or TC deviated in the test aliquots) to support the conclusions.

Line 14, 'Using the interference introduced by mineral dust, we identify mineral dust layers and find very good agreement with a complementary method based on calcium concentrations and the pH.' I would like some numbers to show the degree of agreement e.g. correlation coefficient. Please present these in the Abstract. Is there an accepted standard for the 'right' value, to which these two methods could be compared? What do the differences between the results of the two methods tell us about the methods or the samples?

Overall, I suggest adding more specific quantitative findings to the Abstract (e.g., “EC median = 11.1 ng g⁻¹; mineral dust deposition up to 2100 mg m⁻²”).

Samples are analysed for the period 2016 to 2024. The data is presented. But I would like more commentary to understand the significance of the results. Are there meaningful trends in the data? How do the data from these years fit into what has been found for the preceding years, are there patterns, trends, anomalies? I think there is a lot more context to the measurements that could usefully be added.

Lines before 283, do you think the decrease in Fe fraction is due to preferential deposition of particles with a high iron content (density? size?)

I like the discussion lines 302 to 310. But when you say 'the uncertainty of the mineral dust approximation is 65%', which result does this refer to? It is unclear, and rather than uncertainty, I think you may mean variability?

line 324, why can they only be underestimated?

Line 416, 'Data used in this work will be uploaded to TU Wien Research Data and the doi will be added here.' I think that at this stage in publication, it is time to upload the data and make the doi. For readers, reviewers, etc. The data must now be in final form if you have analysed it and written the paper, correct? 

Technical.

Advisory bodies (SI; IUPAC) advise that the symbols used for variables should be italicized. Please apply this convention at multiple locations e.g. 'n', 'R^2' and so on.

There are a lot of abbreviations. LAI, LASI, LAP, GOK, GAW, TOA, WinsTC, WinsOC, EC, OC, ICP-OES, IC, WMO, FLK. They are often a barrier to understanding for non-experts. Some are used rarely. Many (ICP-OES, LAI, LAP, WMO and RF) are only used once. TOA is defined at the fifth use, not the first use. RF is undefined. Recommend to use best practice - define an abbreviation on first use and only define an abbreviation if it is used multiple times. Some (WinsOC & WinsTC) are defined twice. Avoid abbreviations in Abstract whenever possible since it shoudl function as a standalone summary.

There are times when the meaning is unclear, please rewrite to improve clarity. Examples: 

Line 131, 'Since the temperature range given previously was not always reached for our set of samples,'

Line 153, 'Considering layers with coloured filters would match the two approaches in 2018,'

Line 197, 'EC concentrations for single layers and changes for layers including mineral dust are shown in Figure 3 exemplary for 2020 and 2024.' -- 'exemplary' is unclear, used in a way that doesn't align with it's meaning -- do you mean 'EC concentrations for single layers and changes for layers including mineral dust are shown in Figure for 2020 and 2024, chosen to exemplify the 

Line 75, 80, 119 and elsewhere, 'filtrated' is not a word, use 'filtered'

Line 79, replace 'Contrary,' with 'On the contrary' or 'In contrast'

Check x-axis label in Figure 4, remove '()'?