Analysis of the Transport of Aerosols over the North Tropical Atlantic Ocean Using Time Series of POLDER/PARASOL Satellite Data

Round 1

Reviewer 1 Report

The goal of this paper is to use fine and coarse PARASOL-satellite based aerosol optical depth (AOD) retrieved from the POLAC algorithm.
Satellite products covering 9 years of data were examined to investigate the
transport of aerosols over the North Tropical Atlantic (NTA) Ocean.

The paper has merit in using a unique dataset focusing on the NTA ocean, which represents an area of particular interest to study spatial and temporal variability of aerosols which is known to be the largest and most persistent dust source on Earth. I find this paper suitable for publication in Remote Sensing after addressing some comments:

Keywords should be shortened to keywords instead of sentences References are not in order (e.g. first reference is [29]) It is not clear why the authors decided to compare their results with MODIS products and not with MISR products which also has size-resolved AOD products within the multiangle framework of MISR with a finer spatial resolution of 4.4km. MODIS fine mode product over the ocean is known to be relatively accurate while the fine mode product over land is not recommended to be used in research and as a result, it was removed from the latest MODIS collection. It's not mentioned in the paper what data collection has been used for MODIS data, and not clear where was the MODIS land fine mode aod used for, if at all. There is no discussion on the fact that the PARASOL dataset is limited in time (ends at 2013) and how can this analysis support future work or how can it be extended past 2013 ?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Report on Fréville et al.:

The paper by Fréville et al. presents an exhaustive investigation of the North Atlantic Ocean by means of AOD and it can be divided in three sections. The first compares multi-annual POLDER data – processed with the POLAC algorithm – with AERONET and MODIS data. The second examines the spatio-temporal variability of total, coarse and fine AOD over the study area. Finally, the third discusses the differences of the satellite-based data with model reanalysis. The paper highlights the importance of PARASOL-like observations from space and the need to constrain models.

Broad Comments

The study describes the major aerosol corridors over the North Atlantic Ocean and points out the significant coarse mode transport towards the Americas. This finding coupled with the observed underestimation of the model reanalysis indicate the need to correct for the coarser dust. However, in many parts, the paper is worded poorly and I suggest the English brushed out. Nonetheless, I have several technical comments addressing this in the Specific Comments section. The paper is not yet suitable for publication unless the specific comments are covered.

Specific Comments

Important and minor comments are given for consideration in this section. In the end of this section, I have a series of technical comments. P1Ln1 means Page 1 Line 1.

P1Ln1: Please be sure to explain the acronyms as soon as they first appear in the text. Furthermore, be consistent with the acronyms POLDER, PARASOL, POLAC and their combination as they are often mixed and confuse the reader.
P2Ln80–81: Please rephrase.
P3Ln99: The first element that strikes me by observing Figure 1 is that I do not observe high aerosol values in the bay of Mauritania and Senegal, where lies the main pathway of dust during the summer season. For instance, see Tsamalis et al. 2013 (Reference 112). Could you comment on that? By observing Figures 7g and 7i, that area yields high AOD values. Is it because
of the averaging this underestimation I see in Figure 1?
P4: Make the discrimination between fine and coarse clear in Table 1.
P6Ln214–215: Is this an additional procedure done apart? Are you referring to the quality assurance AERONET program?
P8Ln305–307: Why is this “remarkable”, considering the proximity of the stations to the source? According to Table 2, Calhau exhibits the highest AERONET AOD values among the 4 stations. Therefore, the statement is incorrect. Please clarify. Moreover, Table 2 shows that the inter-seasonal means of Ragged Point are comparable to the Spanish stations for AERONET and higher for PARASOL. Can you comment on this?
P8Ln321: Is the bias range -0.2–0.2 significant? Could you give some examples from other studies?
P8Ln323–325: It would be helpful to see the number of samples in Table 2.
P9Ln331 & P9Ln369–372: Where this overestimation come from? Is it because of the different definition for AERONET and
PARASOL? Please expand.
P9Ln352–360: The scatter plot in Figure 5 shows low AOD values for Ragged Point, whereas the inter-seasonal means are comparable to Tenerife and La Laguna. Why is this happening? Also, do the values reported in Table 2 compose Figure 5?
P9Ln367—369: How does this statement add to the discussion? Are you referring to the underestimation of the coarse mode transport across the Atlantic? Please clarify.
P10Ln383: I think a map with the difference PARASOL–MODIS would amplify the discrepancies between the data. Also, a couple of lines could be spent to explain the effect over the Canary Islands that is visible through all the PARASOL plots.
P10Ln395–398: What is the mechanism that produces this effect?
P10Ln398–413: Is it a MODIS overestimation or a PARASOL underestimation?
P12Ln434: North Africa comprises the Sahara Desert in its entirety. Shouldn’t it be the “Sub-Saharan Africa”?
P12Ln453: Is “minimum” correct?
P12Ln466–458: If you mean figure h and j, I do not agree. Please clarify.
P12n468–470: I cannot understand. Please rephrase.
P13Ln488–490: Why is this happening? It is evident that the biomass burning production is the cause. Please acknowledge this factor in the paper.
P13Ln498–501: Please rephrase.
P13Ln514–516: Please rephrase.
P14Ln564–565: Please rephrase.
P14Ln552–554: Maybe I missed it, can you comment if the reason of the discrepancy is the data assimilation or the model architecture itself?
P15Ln595–597: Please rephrase.
P16Ln606–609: Please rephrase.
P16Ln618–619: I think that this is the case for all seasons. What were your expectations for winter? Also, why fine particles do no travel as far as coarse particles? Is it the height of injection, the mixing with other particles, other mechanisms? Can you comment on that?
P16Ln619–620: Please rephrase.
P16Ln626: To which aspects are you referring?
P17Ln651: “(e.g. the free troposphere)”? Can you explain?
P17Ln677: Could the contribution be given in %?

Technical comments

P1Ln2: Add “the” before “POLAC”. The definite article is missing in multiple occasions.
P1Ln4–5: Replace “with . . . AERONET network” with “with AERONET ground-based measurements”.
P1Ln11: Delete “emission”.
P1Ln12: Replace “AOD satellite products” with “POLAC/PARASOL data”.
P1Ln13: Replace “confronted to” with “compared with”.
P1Ln14: Replace “reanalysis” with “reanalyses”.
P1Ln15: Add “In summary,” before “Our”.
P1Ln17: Correct “at the regional scale”.
P1Ln17: Delete “on”.
P1Ln29: Delete “all”.
P2Ln36: Delete “desert”.
P2Ln37: Replace “dusts” with “dust”.
P2Ln41: Replace “contributes” with “contribute”.
P2Ln42: Please explain “contributing”.
P2Ln43: Delete “hereafter referred to as”.
P2Ln49: Replace “represents a” with “are”. Replace “technique to” with “for”. Replace “monitor” with “monitoring”. Replace
“at a global scale” with “on a global scale”.
P2Ln51: Add “a” before “context”. Delete “the” after “derive”.
P2Ln54: Delete “using remote sensing techniques”.
P2Ln61: Add “the” before “PARASOL”.
P2Ln62: Replace “measure” with “perform”.
P2Ln65: Replace “represents a piece of information” with “are”.
P2Ln67: Replace “improves” with “improve”.
P2Ln69: What do you mean by “new”?
P2Ln70: Replace “For that” with “To this”.
P2Ln77: What do you mean by “abundance”
P3Ln86: Add “the” after “study”.
P3Ln93: Remove the brackets.
P3Ln93: Delete “although”.
P3Ln96: Replace “origins” with “origin”.
P3Ln99–100: Delete “the range”. Remove the brackets. Delete “in altitude and longitude respectively”.
P3Ln101–102: Delete “the range”. Remove the brackets. Delete “in altitude and longitude respectively”.
P3Ln102–103: Delete “the range”. Remove the brackets. Delete “in altitude and longitude respectively”.
P3Ln109: Replace “has been” with “was”.
P3Ln117: Replace “views” with “view”.
P3Ln121: Delete “has been”.
P3Ln122: Replace “lowered continuously” with “has been continuously lowered”.
P3Ln126: Replace “operationnal” with “operational”.
P4Ln132: Replace “at a global scale” with “on a global scale”.
P4Ln136–137: Delete “that from”.
P4Ln147: Add “The” before “POLAC”.
P4Ln184: Delete the second “that”.
P4Ln185: Replace “exploited in” with “employed by”. Replace “is used as well here” with “is used here as well”.
P4Ln161: Replace “whatever” with “independently of”.
P4Ln163: Delete “made about”.
P4Ln164: Add “the” before “POLAC”. Delete “a piece of”.
P4Ln165: Replace “one” with “a”. Replace “in complement to” with “coupled with”.
P4Ln168: Replace “between a case where” with “when”.
P4Ln169: Replace “the case where” with “when”.
P4Ln171: Add “the” before “POLAC”.
P5Ln180: Delete “of these points”.
P5Ln183: Add “the” before “POLAC”.
P5Ln186: Delete “determination”. Replace “Angström” with “Ångström”. Delete “of AOD”.
P5Ln189: Replace “discriminate” with “discriminates”.
P5Ln192: Delete “current”. Delete “the” before “AOD”.
P5Ln195: “In fine”?
P5Ln203: Delete “in presence”.
P5Ln206: Replace “or” with “and”.
P6Ln221: Add “are used” after “(Figure 1)”. Replace “the AERONET” with “continental AERONET”. Delete “that are”.
P6Ln221–222: Replace “inland the Africa continent” with “over the study area”.
P6Ln222: Delete “or the America Continent”.
P6Ln226: Replace “record” with “records”.
P6Ln226–228: Three sentences begin with “Note that”!
P7Ln242: Replace “band” with “bands”.
P7Ln256: “NASA, s”?
P7Ln274: Delete “the” before “land”.
P8Ln314: Add “The” before “Seasonal”.
P8Ln326–327: Move “respectively” to the end of the sentence.
P8Ln326: Replace “Figures” with “Figure”.
P9Ln331: Replace “surestimation” with “overestimation”.
P9Ln345: Add “the” before “results”.
P9Ln348: Delete “match-up”.
P9Ln349: Delete “time”.
P9Ln350: Delete “that were”.
P9Ln352: Delete “The” before “PARASOL”.
P9Ln354: Replace “from the East to the West” with “from East to West”.
P9Ln355: Replace “sub-region” with “zone”.
P9Ln365: Delete “that” after “while”.
P9Ln367: Replace “surestimation” with “overestimation”. Replace “Differents” with “Different”.
P9Ln378: Delete “previous study made by”.
P10Ln393: Replace “an” with “a”.
P10Ln401: Replace “is” with “are”.
P10Ln404: Replace “in regards with” with “in regard to”.
P10Ln417: Replace “points” with “point”.
P10Ln425: Replace “Aerosols” with “Aerosol”.
P10Ln426: Replace “such” with “the”.
P12Ln432: Add “the” before “aerosol variability”.
P12Ln437–438: Replace “Then, . . . comes from” with “This variability is mainly driven by”.
P12Ln442: Add “side” after “opposite”. Replace “allows” with “allow”.
P12Ln458: Replace “part” with “part”. Add “the” before “aerosol”.
P12Ln462: Replace “Figures” with “Figure”.
P12Ln463: Delete “lower”. Replace “comes” with “come”.
P12Ln471: Replace “could” with “can”.
P12Ln476: Delete “of” after “mode”.
P13Ln477: Replace “Such result” with “This”.
P13Ln478: Replace “relatively” with “relative”.
P13Ln486: Delete “of aerosols” and add “Aerosol” in the beginning of the sentence.
P13Ln500: Delete “, more specifically,”.
P13Ln501: Delete “respectively”.
P13Ln505: Add “with Figure 9” after “parallel”.
P13Ln518: Delete “retrieved”.
P13Ln519: Delete “of aerosol”.
P13Ln521: Replace “in regard with” with “in regard to”.
P14Ln536: Correct the values “-0.3/-0.4”. Replace “in regard with” with “in regard to”.
P14Ln537: Replace “lowest” with “lower”.
P14Ln550: Replace “confirms” with “confirm”.
P14Ln557: Replace “quaterly” with “quarterly”.
P14Ln558: Delete “by” after “processed”.
P14Ln559: Add “the” after “allowing”.
P14Ln561: Add “the” before “aerosol”.
P14Ln562: Replace “relatively” with “relative”.
P14Ln563: Add “the” before “source”.
P14Ln565: Replace “whom” with “which”.
P14Ln569: Replace “relatively” with “relative”.
P14Ln571 & P14Ln573: Replace “concomitance” with “coexistence”.
P15Ln577: Replace “Intercation” with “Interaction”.
P15Ln585: Replace “which” with “who”.
P15Ln586: Replace “concludes” with “conclude”.
P15Ln587: “On its side”?
P15Ln593: Replace “relies” with “rely”.
P15Ln596: Delete “thus”.
P15Ln601: Delete “to be”.
P16Ln605: Replace “its” with “the”.
P16Ln613: Add “the” before “Seasonal”.
P16Ln622: Replace “reanalysis modelling” with “model reanalysis”.
P16Ln623–624: Delete “atmospheric model and” and “techniques”.
P16Ln632: Replace “unability” with “inability”.
P16Ln643: “sensitively function”?
P16Ln644: Replace “this latter study” with “the latter”.
P17Ln654: Replace “whose” with “with”. Delete “the” after “diameters”.
P17Ln659: “are concerned by”?
P17Ln647: Delete “method”.
P17Ln677: Replace “relatively” with “relative”.
P17Ln678: Replace “consistently” with “, consistent”.
P18: Replace “Ragged point” and “La laguna” with “Ragged Point” and “La Laguna” respectively.
P19: Delete “different locations (see Figure 1)”.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

This paper observes the movement of aerosols, especially desert dust, over the
tropical Atlantic Ocean. Measurements from the POLDER instrument are processed
with the POLAC algorithm, for which there has been an adjustment of the aerosol
microphysical properties. The satellite observations are compared to four sunphotometer sites in the region, showing a similar level of agreement to that achieved by other satellite AOD products. A qualitative comparison of seasonally averaged AOD
fields shows that the POLDER data reports a greater loading of aerosol, both near
the Saharan source and transported to the Caribbean, than seen in the widely used
MODIS product or two renanalyses.

This paper is acceptable for publication after minor revisions. It is a pleasingly
straightforward demonstration of the results of a satellite dataset, attempting to
address a known disagreement between aircraft observations and weather/climate
models as to how far large aerosols are routinely transported. My only major disagreement with the paper is the framing of it’s comparisons (particularly in §4): the
POLDER product is described as the truth, with all others having a bias relative to it
(for example, line 403). While I relish pointing out the numerous faults and biases
in the MODIS product, it is generated by widely respected algorithms with extensive validation. They have the weight of evidence behind them so, arguably, POLDER
has a bias relative to MODIS. The text should be revised to reflect this, or additional
evidence presented to show that POLDER data is detecting a real signal that MODIS
omits.

There are a few other matters that must be addressed:

There is no evidence provided for the statement on line 355, ‘a weak abundance of aerosols over the study area does not lead to a decreased performance
in PARASOL observations.’ The scatter plots of Fig. 5 show that POLAC agrees
acceptably with AERONET around the NTA. They do not highlight the low
AOD regime and, by the standards of a satellite validation, have relatively
few low AOD points. In fact, the greatest discrepancies shown in the Tenerife
plot are for τ_AERONET < 0.1 and there is a large spread in low AOD observations at Ragged Point.
You possibly meant to point out that the discrepancies shown in Fig. 5 do
not appear increase with increasing AOD. That is noteworthy given that the
MODIS products’ error envelope is proportional to AOD. I would be willing to
believe that your method exhibits superior performance at high AODs (due to
the advantages of multiple views), but there is  insufficient evidence presented
to support that. The conclusion on line 568 is unsupported in this text, ‘confirms the long-range transport of coarse particles over our study area relative to other types of aerosols’. You showed that coarse mode AOD is enhanced across the NTA. You
never quantified that transport, nor calculated its ‘strength’ relative to that of
fine-mode aerosols. One possibility might be to calculate the e-folding length
of the AOD fields you have. I am surprised that the GRASP and MAIAC algorithms are never mentioned.
The former is the only other POLDER aerosol product I am familiar with and
the latter shares POLAC’s heritage as an atmospheric corrections for the ocean
colour community.

A few more minor comments referenced to the appropriate line number:

L25 While aerosols are an important part of the atmosphere, I believe nitrogen
and oxygen are it’s ‘main’ constituents. L35 Your comments about fine and coarse mode aerosols may be true in terms of
mass, but are misleading when thinking in terms of particle number. A significant source of natural aerosols are SO2 emissions from volcanoes, which can
dominate the fine mode for hundreds of miles. (For example, the enhancement of AOD downwind of Kilauea is seen in every satellite product.) Similarly, on L37, ‘number’ may be a better word than ‘amount’. L39 The Sahara is the world’s primary source of dust emissions, not the first. L137 MODIS Collections 6 and 6.1 made numerous improvements to their method. Do you have any intention to repeat Bréon’s analysis? L184 This paragraph should make clear that the uncertainties given refer to the algorithm’s performance in simulations, and are not representative of real-world
performance. (I sadly don’t read French, and so struggle to follow the argument in the thesis cited; I apologise if my request is covered there.) L206 Why use a 25 km grid when the MODIS product is reported for 10 km super-pixels? Also, what sort of grid is used (e.g. lat-lon, sinusoidal)? L209.3 Though there are only about 300 AERONET sites regularly reporting data at the moment, there are over a thousand Cimel radiometers in use. L230 I presume these times are UTC, as PARASOL is on a sun-synchronous orbit? L281 There is an inconsistency between your definition of fine mode on L34 and
here (< 1 μm vs. < 2.5 μm). There isn’t much that can be done about that, but has it been investigated? L327 While validation of AOD in the fine/coarse mode is uncommon, it has been
done. For example, doi:10.3390/rs10091414 or doi:10.1016/j.rse.2018.12.012. Fig.6,7 Have you considered showing the fine-mode fraction (i.e. AOD_f /AOD)? While I agree that AOD_f is the better quantity to report (as the uncertainty is better behaved for extreme values), given the relatively small values of AOD_f utilise few of the available colours, a map of the fine-mode ratio may make immediately clear the difference between areas dominated by dust and the differential transport you wish to show. L515 This sentence doesn’t completely make sense. I think you mean, ‘The loading of aerosol in the west zone is underestimated by MERRA-2, despite it also overestimating the loading in the east zone. This implies weak transport in the MERRA-2 model.’ L602 While it is technically true that this study performs a more extensive evaluation than the three studies named, I feel it is somewhat disingenuous to imply this study is ‘deeper’ than any previously performed. There are numerous
assessments of Level 3 aerosol products on a global scale that, while not explicitly about the topic, have noted the extent of the Saharan outflow. For
example, doi:10.5194/amt-6-1919-2013 considers a climatology of fine-mode
fraction.

Finally, a few typographic corrections:

Why is AOD rendered in italics throughout the paper? ‘the’ should be added before the following words: L2 POLAC; L5 AERONET;
L61 PARASOL; L122 PARASOL; L164 POLAC; L245 MACC; L246 MACC-II; L284 PARASOL; L470 POLAC. The final ‘s’ should be removed from the following words: L41 contributes; L67
improves; L89 moves. A final ‘s’ should be added to the following words: L226 record; L242 band. L34 Anthropogenic aerosols mostly contribute L43 where the Sahran air layer originates (hereafter L87 Western Africa (Sahara and Sahel) regions L109 which was operational from L112 which consists of a rotating L116 which allowed observation of the entire L117 also enabled observation of a given ground L121 its orbits was lowered L122 PARASOL orbit was lowered L123 PARASOL orbit caused the temporal windows to be delayed by about 1 hour L209.12 The inversion allows retrieval of the wavelength-dependent aerosol L209.14 which enables calculations of aerosol properties [19]. L256 NASA’s L331 An overestimation of AOD_f is L341 consequence of the reduced number of fine mode aerosols L364 get an average close to L410 In particular, different whitecap coverage (a parameter that is a function of the wind speed), the stability L447 AOD greater than or equal to 0.6 L479 transport of coarse aerosols over long distances could L481 aerosols reaching the West zone L557 quarterly L583 constant on both sides L586 counteract gravitational settling L593 other studies rely on L607 layer with tops out L632 It seems that the inability of MERRA-2 to reproduce L647 mechanisms that retain L677 0.4) relative to the total L678 area, consistent with previous L810 It wasn’t obvious from the citation that this was a thesis and I’d have no idea
how to find it. L918 O’Neill is the proper capitalisation.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf