This is a neat paper, a very practical application of data and modeling efforts that the authors and others have been pursuing for years.  The study reaches clear recommendations, which is especially gratifying. 

Lines 183-184.  This is not a complete sentence.

Clearly the near-surface dust concentration is especially important, and I know that CALIPSO sensitivity tends to diminish within the lowest 75 – 100 m of the surface. Is the reason that CAMS produces lower dose than CALIPSO for near-surface dust concentrations the assumption that the CALIPSO extinction at 100 m is extrapolated to the surface?  The confidence with which you can assess the elevation of a near-surface concentration peak seem especially relevant based, e.g., on Figures 4c, 4e, 5c, and 5e, and it comes up again in the discussion of Figure 8.  (I now see some discussion in lines 495-502. And I agree it is surprising in light of CALIOP being lower than other measurements. I’m wondering whether there is any EarliNet data that might help here.)

Section 2.3.3. I think you do as good a job as one can with available satellite data in constraining the dust mass concentration. But it might be worth also making some assessment of the uncertainty in mass concentration, e.g., associated with the uncertainty in the MEC. (I realize you compare the model and two lidar estimates with each other, which might represent a rough estimation of uncertainty due to s in Equation 1, but I think you are using the same MEC to obtain concentrations for all three.) (Again, I now see you say a bit more about MEC uncertainty in lines 503-513. Ok, so is the recommendation that we need better measurements of dust MEC?) Further, is there any available data on actual engine wear, that might be used to test or constrain the overall results?

Figure 3.  I know there are some seasonal and altitude differences, but if you could reorder the entries in the legend for this figure so they are generally in the order of dust concentration magnitude, it would be easier to discern the lines associated with the lowest few, which seem to be Phoenix, Hongkong, and Sydney.

I understand that you are effectively using seasonal background dust levels for these calculations, and I know that for some phenomena such as dust transport in general, extreme events dominate. So, I’m wondering (a) how well the limited CALIPSO sampling and the CAMS model simulations capture sporadic larger dust events, (b) whether the airports in question shut down when dust loading is unusually elevated, and (c) how the dosage for even one elevated event for which the airport might remain open might compare with the typical seasonal averages.

Line 387. The end of the sentence is missing.

Line 576. Might visibility still be an issue in the vicinity of a busy airport, even after an aircraft has left the ground?

General comments:

The work provides valuable information about the ingestion of atmospheric mineral dust at 10 global airports by using one reanalysis dataset and two observational datasets derived from lidar measurements. The authors compare climatological and seasonal features of dust dose and find substantial differences among the datasets. These differences are discussed in detail. The research design and methodology is considered appropriate. The results are explained in detail and the presentation of the results is adequate, but the discussion of the results could be improved. Please see detailed comments below.

I recommend the article for publication in NHESS after addressing the following comments and recommendations.

Major comment:

The differences between the datasets and therefore dust dose uncertainties are rather large at some airports and I understand that it is important to explain these differences. However, a paper with the title “Aircraft engine dust ingestion at global airports” should not only discuss differences between datasets and data retrievals in the “Discussion” section. I would rather expect some discussion about the results and their implications.

My suggestion is the following:

• Move the detailed information about the model and data retrievals to the supplement and condense most important information in the results or the discussion section.
• Compare your results to that of Bojdo et al. (2020) (see comment below) and other studies if available.
• Discuss the model results in terms of magnitude: Since CAMS < CALIOP < AERONET < MODIS, CAMS might significantly underestimate dust concentration at several locations. What does that mean for aircraft engines?
• Discuss the implications of a jet engine core ingesting dust (text from conclusion section, lines 594 to 610). Could you give more detailed information about the specific types of damage? Is there any critical mass of dust? Please give more information, if possible.

Minor comments:

Lines 21/22: You explicitly mention Beijing’s dust dose in the abstract. Why? I suggest either removing the sentence or explaining why this information is important.

Lines 65 to 72: While the work of Bojdo et al. (2020) should definitely be mentioned in the introduction, I recommend moving the description of their work including the discussion of their findings to your discussion section (see major comment above).

Line 101: Did the authors also investigate dust dose at the airport in Singapore? If not, please rewrite this sentence.

Line 124, Line 661: Please consistently write “analyse” or “analyze”.

Line 147: Please consistently use data as singular or plural noun (plural noun: e.g., lines 116, 153, 287, 357, 657, 680; singular noun: e.g., lines 147, 187, 541, 570).

Line 157: The “Young et al. 2018” reference is missing in the reference list.

Lines 183/184: I do not understand this sentence. Please clarify.

Lines 187 to 189: Did you do this in your study or is this part of the LIVAS processing? Please clarify and add a remark.

Line 203: “two products are not similar”: what does “not similar” mean? How big are these differences? This sentence seems to be the condensed version of Section 5.2 (see major comment above).

Lines 219/220: “We choose to compare the profiles using mass, since this is the metric of interest to the aviation community, though we note that extinction comparisons showed the same results.”: The authors could show these results in the supplement.

Line 223: “Dust dose is defined as the total mass (g) of dust”: When using equation 2, the unit of total mass is “kg” rather than “g”.

Line 241: What does “lbf” mean?

Figure 2: Please use the same y-axis for both panels. Figure caption: add “(blue)” after “altitude” and “(black solid line)” after “w_core”.

Lines 265 to 267: “All airports show the highest mean dust concentrations in JJA (driven by peak solar heating and dry convection over northern hemisphere desert regions) except Beijing and Niamey.”: What about Sydney, which is located in the southern hemisphere?

Line 272: “Sydney, Phoenix, Hong Kong and Bangkok all display mean dust concentrations below 10 μg m^-3.”: The authors should add here that these airports will not be discussed later on.

Figure 4: Is the different number of CALIOP L3 and LIVAS overpasses related to the different dust products (pure-dust vs. dust also from polluted dust products)?

Lines 359 to 383: Compared to the description and discussion of the other figures, this text gives to much detailed information and some aspects are not clear. For example: lines 359/360: “…CALIOP L3 substantially larger than both, are most evident at airports with a low altitude dust plume, particularly Niamey in DJF/MAM and Dubai year-round.” In Niamey, the DJF and MAM medians of both LIDAR datasets are in very good agreement. In Dubai, the DJF and SON medians of both LIDAR datasets are also in very good agreement. Please clarify and shorten the text.

Lines 389/390: “For departure (see supplement), overall the similarities and differences between the datasets are the same as for arrival, with lower doses for departure by 10 to 23%.”: Do these numbers refer to the median or to the lower/upper quartile?

Lines 390/391: “However, in a few cases differences between datasets compared to arrival are sensitive to the overall vertical profile shape and magnitude, particularly if ground concentrations are very large.” Isn’t this always the case because of the location of the hold altitude?

Lines 401 to 403: “This is partly a feature of the larger magnitudes seen in the lidar data compared to CAMS, but also due to the lower sampling rate for CALIPSO compared to the regular 3 hourly model output from CAMS.”: Did you use all CAMS data or only CAMS data coincident with CALIOP measurements? Please clarify.

Lines 437/438: “For the peak dust seasons, a reduction in dose of 41% at Dubai in JJA, 34% at Delhi in JJA and 39% at Niamey in DJF could be achieved.”: I try to understand how you calculated these numbers by looking at Delhi: According to Figs. 6 and 7, maximum dust dose in JJA is 6.6 g and 4.4 g for arrival and departure, respectively. This sums up to 11 g. However, according to Table 1, the reduction of 34 % refers to 6.44 g, which indicates that total dust was approximately 19 g. Where does this difference come from? Please clarify.

Table 1: Please specify only one decimal place for dose reduction.

Lines 484/485: “without mixing from other aerosol types (e.g., Marrakesh and the Canary Islands)”: What about sea salt aerosols in Canary Islands? Please clarify.

Section 6: Conclusion: I’d suggest renaming this section to “Summary and conclusions”.

Lines 633/634: “resulting in a mean underestimate by CAMS of 2.4 over both datasets”: I do not understand the relevance of the mean bias over both lidar datasets. I suggest removing this part of the sentence.

Lines 657 to 660: I suggest moving these two sentences to line 650 and mentioning that the results discussed before were based on the CAMS reanalysis.

Supplement, Figure S5: Canary Island and Delhi: LIVAS data are missing.

Editorial:

Line 21: add “northern hemisphere” before summer.

Line 25: Add “instrument” after (CALIOP).

Line 29: “up to 44% and 41% respectively” rather than “up to 44% or 41% respectively”, I think.

Section 1: remove “1”.

Line 49: “O’Connell” rather than “O’connell”.

Line 64: “(Inness et al. 2019)” rather than “(Inness et al. (2019))”.

Line 76: “CAMS forecasts and reanalyses” rather than “CAMS forecasts and reanalysis”.

Line 81: Add a reference after “dust properties”, e.g., Mona et al. 2012, doi:10.1155/2012/356265.

Line 83: Add “instrument” after “(CALIPO)”.

Line 84: “(e.g. Liu et al. 2008a; Yang et al. 2013; Song et al. 2021)” rather than “(e.g. Liu et al. (2008a); Yang et al. (2013); Song et al. (2021))”.

Line 86: “12-year period” rather than “12 year period”.

Line 93: “Section 5 concludes”: This is not true. Section 5 is the “Discussion” section, Section 6 concludes.

Line 99: Please exchange “Bejing” and “Bangkok” to follow the numbering.

Line 100: “‘dust belt’,” rather than “‘dust belt,’”.

Line 103/104: add “coarse” before “model resolution” and refer to Section 2.2.

Line 113: Introduce “ECMWF”.

Line 121: “lower/upper diameters” rather than “upper/lower diameters”.

Line 141/142: I suggest moving the sentence “The CALIPSO orbit track….” to line 134 (after the Winkler et al. (2010) reference.

Line 142: Add a break before “Here we analyze…”.

Line 158: “reported by Floutsi et al. (2022)” rather than “reported by (Floutsi et al., 2022)”.

Line 160: add “latitude-longitude” before “grid”.

Line 161: Add a break before “We use the extinction coefficient”.

Line 161: “extinction coefficient profiles at 532 nm” rather than “extinction coefficient at 532 nm profiles”.

Lines 165, 207: “CALIOP” rather than “CALIPSO”.

Line 166: “an altitude of 90 m” rather than “an altitude 90 m”.

Line 172: (LIVAS, Amiridis et al., 2015) rather than “(LIVAS, Amiridis et al. (2015))”.

Line 174: “EARLINET, Pappalardo et al., 2014; last” rather than “EARLINET, Pappalardo et al. (2014); last”.

Lines 180 and 197: “‘pure dust’, ‘dusty marine’,” rather than “‘pure dust,’ ‘dusty marine,’”.

Line 191: add “as input” after “profiles”.

Line 202: “sr” rather than “Sr”.

Lines 202/203: “(Amiridis et al., 2013; Marinou et al., 2017; Proestakis et al., 2018; Floutsi et al., 2022, see supplement)” rather than “((Amiridis et al., 2013; Marinou et al., 2017; Proestakis et al., 2018; Floutsi et al., 2022), see supplement)”.

Line 224: “from Clarkson (2020)” rather than “from (Clarkson, 2020)”.

Line 226: add “(dimensionless)” after “regime”.

Line 235: “thermodynamics” rather than “theromodynamics”.

Line 252: “discuss” rather than “test”.

Line 301: “are” rather than “were”, I think.

Line 304: “show” rather than “showed”, I think.

Line 313: “greater than 2 g in winter”: it is even greater than 3 g in winter.

Line 316: “contributes to at least 50%” rather than “contributes at least 50%”, I think.

Line 350: remove “to those calculated from”.

Figure 8: Please use the same y-axis for all panels; please capitalize the first letter of each airport name. Does it make sense to use an exponential y-axis?

Line 387: The Section number is missing.

Line 433: “for the six dustiest airports” rather than “for each airport”.

Line 438: “an” rather than “a an”.

Line 439: Maximum reduction is 19 % in Canary Islands.

Line 482: remove “the reason” after “explain”.

Line 489: “(e.g., Zhao et al., 2022)” rather than “(e.g., Zhao et al. (2022))”.

Lines 495/496: “(e.g., Schuster et al., 2012; Sayer et al., 2018; Song et al., 2021)” rather than “(e.g. Sayer et al. (2018); Schuster et al. (2012); Song et al. (2021)”.

Line 500: “DOD” rather than “DAOD”.

Line 509: “O'Sullivan” rather than “O'sullivan”.

Line 526: “sr” rather than “Sr”.

Line 578: Add a break before “This study…”.

Line 599: “representative of the region around” rather than “representative of region around”.

Line 613: “spent there, which” rather than “spent in the hold pattern, which”.

Line 625: “the vertical structure at Beijing” rather than “the vertical structure shape at Beijing”.

Line 642; remove “both” before “dust properties”.

Line 652: add “by increasing hold altitude from 1 km to 3 km” after “can be reduced by 41%”.