Review for Atmospheric Chemistry and Physics

Title: 15-year variability of desert dust optical depth on global and regional scales

Authors: Stavros-Andreas Logothetis, Vasileios Salamalikis, Antonis Gkikas, Stelios Kazadzis, Vassilis Amiridis and Andreas Kazantzidis

General Comments: The authors have analyzed dust optical depth (DOD) trends from the MIDAS dataset for all major global dust source regions in some detail in this paper. I think that this paper makes a useful contribution to the literature and provides a generally comprehensive global view of trends in dust optical depth. However, I have a few issues with the manuscript that the authors should explore in order to make the analysis more convincing and also easier to read and comprehend. First, how are we to gauge the reliability of the DOD trends without showing some verification versus AERONET sites for example? For example the Gkikas et al. (2021) paper on this dataset shows significant regional biases in DOD for MIDAS versus several AERONET sites in their Figure 4d. You should include at least two trend analyses from AERONET data to compare with your MIDAS inferred trends of DOD. The SDA retrieval from AERONET (O’Neill et al, 2001, 2003) separates the AOD into fine and coarse modes and trends of coarse mode AOD from AERONET could be compared to the MIDAS trends of DOD. Second, throughout the manuscript I found it difficult and confusing to have to look at the supplemental document file in order to confirm principal conclusions made in the primary manuscript. The way this paper is currently written it requires quite frequent switching between reading the 2 document files in order to understand the author's arguments. I strongly urge the authors to move some/many of the figures that are in the supplemental section into the main paper. Third: The very localized (relatively small area) of the large negative trends in DOD shown for the Bodele Depression (Figure 9) suggest the possibility of a surface reflectance artifact in the satellite retrievals and therefore possibly a trend in surface reflectance and not of dust concentrations. Since this is the largest single dust source on the planet then why is the strong negative trend in DOD over such a small area? It is well known that the dust plumes from the Bodele are advected by winds over very long distances. You need to explain in the text why the strong decrease in DOD is so limited to the very site of this dust source itself. Also the white areas around the blue negative trends are assumed to be near zero in Figure 9, although the color bar at the bottom shows yellow color for near zero trends, not white. Can you explain the discrepancy in the color bar and the map?

Specific Comments:

Lines 141-142: Some quantification is needed here rather than calling the MIDAS dataset 'trustworthy' and rather than just saying '…a quite high level of agreement' exists with other datasets.

Lines 206-207: It seems that this sensitivity analysis is important. Therefore I think the authors need to include some discussion and figures in this paper in order to support their conclusions.

Lines 210-213: This is not obvious in Fig 1 below. As a reader it is difficult to have to look at a different supplemental document in order to confirm conclusions made in the primary manuscript.

Line 221: Please define the blank or white areas on the Fig 1 maps since white is not shown in the color bar scale.

Line 228-230: It seems strange and unlikely that for such a large dust source as the Bodele that the decreasing trend is confined only to the exact location of the depression itself since it is well known that this dust is advected in various directions out of the basin.

Line 236-238: Please explain here what spatial resolutions were examined by de Meij et al (2012).  Again I find it extremely difficult to have to switch back and forth between the main and supplemental documents in order to follow some of your main points in this paper.

Lines 253-261: I think that a table comparing the two data sets (Voss and Evan (2020) and the current study) is warranted here.

Lines 265-266: Please be clear here what the filtering criteria are. Again here, the readability of this paper is reduced significantly by frequently referring to supplemental figures in a different file.

Line 282-283: Yes this is obviously true, however you did not really present any evidence for erroneous trends. Why conduct this exercise if you cannot show clearly why a lack of temporal filtering is problematic? Or maybe I have missed your point, in which case it would be worth trying to clarify this.

Line 288: Please give a short summary of this methodology. As this paper currently is written a reader needs to jump from one document to another (either previously published papers or your supplemental document) to understand what is being done or discussed.

Lines 322-323: Every satellite data set easily detects trans-Atlantic dust transport. Why reference only your current database?

Lines 326-327: OK I applaud the honesty here but this decreases my confidence somewhat in the trend analysis maps. Comparison of these trends with  some verification versus AERONET sites would be useful.

Line 339: I think you might mean “tends to be getting dustier” since Table S1 is a table of trends.

Lines 404-405: It seems that some mention should be given to the statistical significance and uncertainties of these measured/computed global AOD and DOD trends.

Line 468: Please provide some explanation of the strong land-sea gradient in the DOD over the southern portion of the Red Sea shown in Figure 11b.

Summary and major comments:

The authors describe trends in dust aerosol optical depth and aerosol optical depth as calculated from the MIDAS fine resolution dataset. They conduct several sensitivity studies including assessing differences between the geometric and arithmetic mean, variation in start year and timeseries length, filtering, impact of spatial resolution, and regional trends. Overall, I believe that this contributes meaningfully to the existing literature on this topic as it evaluates a new dust dataset and conducts sensitivity tests that have been under-utilized in previous satellite-derived dust climatology studies. However, I believe some additional analysis and revision to the manuscript is necessary.

While the authors present comparisons to several other dust datasets, many of them are also based on MODIS AOD and are therefore not independent from the MIDAS dataset. Additional comparison with datasets that do not incorporate MODIS data are warranted eg. CALIPSO DOD, AERONET, ground-based dust measurements. At a minimum, more acknowledgement in the text of the lack of independent comparison is necessary. The discussion of the differences resulting from the use of the geometric vs. arithmetic mean of DOD is important and that analysis should be presented completely in the body of the paper. In the methods section, more discussion of the uncertainty, strengths, and weaknesses of the MIDAS data should be included. Some of the methods were not completely described. I also believe the authors should elaborate more on the motivation for this work. There have already been several studies on dust optical depth trends using MODIS-derived datasets. Please discuss why this one is a valuable contribution beyond being a different spatial resolution.

The manuscript needs substantial editing for grammar. There were uncapitalized proper nouns, improperly used or missing articles, and generally confusing sentences throughout the text. I have included some of these in my detailed comments but this list is far from exhaustive.

Detailed comments:

Line 21: Changed to “The effect of DOD on the total aerosol optical depth (AOD) change is determined bycalculating the…”

Line 23: change to “…Gobi Deserts) the ratio value is approximately 0.6.”

Line 28: remove “which sounds a critical aspect when satellite-based measurements are utilized.”

Line 31: Change to “are a major contributor to the atmospheric aerosol…”

Line 48: Capitalize “Van der Does”

 Line 52: South America

Line 108: remove “it” in “it is reasonable”

Line 111: change language “taking advantage that” to something grammatically correct

Line 141: Please elaborate on the “highly accurate” MODIS AODs by providing information on error statistics and validation studies

Line 141: Instead of saying “trustworthy”, consider providing specific information about the uncertainty in the product, based on uncertainty in MODIS AODs and MDF

Line 143: note that validation with MERRA-2 is not independent since the MDF is used in the calculation of MIDAS DOD

Line 145: Describe how you minimize contribution of non-dust aerosol species

Line 152: Change the language of “it has been justified the reliability of …” for grammatical reasons

206: Please show this analysis in the supplementary material

Figure 1: Consider showing differences in trend between geometric mean and arithmetic mean instead of showing values

Lines 267-269: Please rephrase for clarity.

Line 298: Please indicate, with grey shading for example, where seasonal trends were not calculated due to data gaps

Line 310: Please provide some values in the text for these trends in order to make it easier to understand without flipping between the manuscript and the supplementary material. For example, “increase trends of up to X yr^-1” or similar language.

Line 327: Rephrase the final sentence of this line to reflect that the MERRA-2 dust fraction in MIDAS overestimates dust as compared to the DOD values derived from CALIOP retrievals.

Line 337: Consider discussing the relationship between cyclone activity and dust eg. Pan et al. (2018) Journal of Climate; Evan et al. (2006) GRL

Line 400: Why not state the global trend over ocean for MIDAS DOD here?

Line 413: 5 years is a very short period over which to calculate a trend.

Line 460: change “do not contradict with” to “are consistent with”

Line 462: “resulting in”

Lines 567-568: Overestimate as compared to what? What are you considering to be the “truth” dataset/value? Also, please rephrase to correct grammar

Line 574: “and globally”

575: Redundant sentence

Figure 5: explain hatching in figure caption.

Lines 582-583: Please rephrase for clarity and grammar

Lines 588: Please explain further how these trend results could be incorporated into chemical models and why this is valuable