| 05 Nov 2025
New processes to counteract sedimentation of coarse dust particles are required for climate models to agree with aircraft observations
Abstract. Recent observations reveal coarser (d > 2.5 µm) mineral dust particles transported unexpectedly great distances from their Saharan source. Transport models represent this coarse dust transport poorly, which has important feedbacks on the Earth's radiative budget, carbon and hydrological cycles, and human health. In this study, we carry out sensitivity tests on the trans-Atlantic transport and deposition processes governing the long-range evolution of mineral dust size distribution in a climate model (HadGEM3-GA7.1). We test the sensitivity of coarser particle transport to sedimentation, convective and turbulent mixing, impaction scavenging, and dust shortwave absorption by removing or scaling the magnitude of these processes. Reductions in sedimentation of 80 % are shown to have the greatest impact on bringing coarse particle transport in the model into better agreement with in-situ aircraft observations. Other tested processes do not result in the multiple orders of magnitude changes in transported coarser dust mass required for agreement with observations. Since sedimentation is a well-understood physical process, we infer that some additional process/es which are misrepresented or not represented at all in the model must be acting to counteract sedimentation. We find that convective and turbulent mixing in the model have minimal impact on coarse particle long-range transport, but are key in controlling the vertical distribution in the Saharan air layer and marine boundary layer, respectively. This study adds to the growing body of evidence that points to processes involved in coarser mineral dust transport and deposition which are not represented accurately or at all in models, which counteract the sedimentation of coarse particles in the real-world. The work in this paper brings the community one step closer to better understanding and modelling of the full dust size distribution and its impacts on weather and climate.
Ratcliffe et al discuss one of the recent (and yet unexplained) results in atmospheric dust science, "the mysterious long-range transport of giant mineral
dust particles" (van der Does et al, 2018). We know from observations that coarse and giant dust particles are transported long distance, and we know from theory and models that this cannot be fully understood or explained. To address this problem, the authors use the HadGEM-3-GA7.1 model in a free running configuration, and explore how it represents the transatlantic transport of dust. In doing this, they modify the intensity of some processes in the model, such as the sedimentation rate, vertical mixing, scavenging or the absorption of solar radiation, in order to find in which conditions the model can sustain the long-range transport. The only significant modelled process able to modify the dust vertical distribution is sedimentation: a drastic reduction of this process (bringing its magnitude to one fifth of the control magnitude, i.e. a reduction by 80%) can significantly alter the transport of coarse dust and keep it sustained for significant transport distances. More precisely, the required reduction is 50% at the source, 50-80% at the Canaries and Capo Verde, and 80-95% at the Caribbean. Given that sedimentation is a well understood process, the authors say, in fact this means that there must be an additional force keeping the particles aloft, which has not yet been identified. Thus study moreover examines the role of the other processes on the vertical distribution of dust.
The research presented is interesting, significant, and relevant, and the authors have clearly invested considerable effort in their use of the HadGEM model. The study has the potential to merit publication; however, the current manuscript does not yet meet the standards of scientific clarity, structure, and analytical depth required for acceptance. I therefore recommend major revision. In the next round, I expect a substantially rewritten mansucript, not just incremental improvements. I would like the authors to make a major effort to elevate the scientific presentation, organization, and critical reflection on their results. With such a comprehensive revision the paper will be suitable for consideration.
Here are a few comments and suggestions:
1) One of the main results seems to be that a sedimentation reduction of 80% is needed to enable the long-range transport of coarse dust (see abstract, see lines 340, 528, and other places in the paper). However a more articulate result is given at line 301 with different reductions at different locations. I believe that the more articulate message would be more powerful, given that there is no overarching goal of coming with a simple number working everywhere to tune the model, but rather a desire to understand and quantify the model's shortcomings.
2) It is unclear why only the 3xSW model run is compared to REControl whereas all other runs are compared to a Control without radiative transfer turned on. This being a clear (probably strategic) choice in the design of the research, it merits an explanation. It is only when I arrived at line 479 that I may have guessed (without being sure) the reasoning behind such choice: is it because you want to separate the variations in the meteorology from the strict dust processes? If this is a reason, then it should be not only stated in section 2, but an explanation of the pros and cons of this choice should be given.
3) We need to remind that the authors' investigation is about the modelled dust in HadGEM3 and the processes represented in that model. Before results can be generalised to other models, one need to explain how similar or different such other models are. And in any case it would be good to have a language throughout the paper that clearly highlights that results are not for the "real" dust out there in the troposophere, but for the modelled dust which is a possible representation of it. It may be good to mention early on that given that each process investigated has a size-dependency (which is not altered in this study), the results obtained are valid for this specific modelling of each process (emissions, sedimentation, transport, mixing, radiation, etc.) which may be different than the processes in another model or in the natural world.
4) At lines 528-533 the authors discuss how their results for HadGEM3 compare to those of Drakaki and Meng. At line 537 they state that the order of magnitude reductions are remarkably similar. I think it would be important here to (1) name the models used in those papers, (2) explain if those models use similar or different sedimentation schemes (and any other significative difference or similarity), and (3) discuss the significance of the similarity of results in light of the differences between models.
5) Improve the overall quality of the language. Although I am not a native speaker, some expressions come across as overly colloquial. In addition, the current phrasing at times seems to rely on the reader to infer or "guess" the authors’ intended meaning, which should be avoided. E.g. guess some symbols (e.g. d in line 1 of the abstract and line 4 of the introduction: certainly it means diameter, but certainly this has to be stated). Guess some timescales (e.g. "relatively recently" on line 26: can the authors indicate a clear time range?). Guess some symbols as detailed in the comments below regarding the methodology section. Guess where is the land in figure 5 (line 356). There are many things in the paper which are unclear to a reader. Normally, a reviewer should not be required to identify so many instances where clarity is lacking. The expectation is that authors submit a well-written manuscript so that reviewers can focus on evaluating the science rather than the writing. In its current form, a full scientific assessment is not possible; a substantially clearer and better-written paper is required before the science can be properly reviewed.
6) Concerning the figures, I'd suggest to bring figure 9 to section 2.1.3 where CM and TM are explained (with also some extra text to explain those concepts). To bring figure S2 into the main paper given that it is mentioned often in the text: perhaps it could be incorporated into figure 4. Figures 5, 6 and 7 are of a similar type and the authors could consider combining them into one figure with several sub-panels. In figure 8 I am not sure if the difference of differences (panel c) is more at risk of creating a confusion rather than making things clearer: can the description perhaps be made without showing this panel?
7) Also, consider well the figures in the supplement: are they really needed (an in that case would they be better in the main paper?) or are they not so important (and in that case they could be omitted all together? My personal feeling is that S1 could go into the main paper, perhaps included in figure 3; S2 as already said could go into the main paper. S3 and S4 are not so useful in my opinion and could be removed all together. S5 shows that emissions are stable and perhaps this could be simply mentioned in the text (without a figure). S6 is a nice figure highlighting a change in the vertical distribution only at the Caribbean and perhaps it could be kept and also added to figure 4. Finally S7 illustrates the atmospheric warming due to 3xSW, which is a nice point discussed in the paper, therefore it should also be included in the paper. Now, if you follow all my suggestions the paper will have some figures with many figure panels and you may want to go into detail and checking if a selection can be operated among those panels.
8) The Conclusions section currently reads largely as a summary of the results. I would encourage the authors to expand this section with a more critical discussion of the implications of the findings and to place the work more explicitly within the broader context of existing research. In particular, it would be helpful to highlight the novelty of the contribution, discuss any limitations, and outline how the results advance or refine current understanding. Clarifying these aspects would make the conclusions more impactful and informative for the reader.
9) The fact that coarse and giant dust has been missed for a long time is also due to the past use of investigative methods unable to reveal it: instruments unable to reveal them and models unable to represent them. This was due in part to the fact that theory predicted these particles not to be there, therefore not justifying the additional effort. This has to be clearer in the wording of the paper. E.g. the expressions "were not frequently measured" (line 27) and "coarse particles are often disregarded" (line 43) do not convey this message and seem to imply an overall unjustified neglection of the coarse mode, whereas there has been a historical evolution of investigative methods and of our thinking. Please improve the language explaining why they were not measured (line 27) and why they were disregarded (line 43).
10) Line 31: add references to Adebiyi and Kok (2020) and van der Does (2016); moreover, put papers in order of publication (oldest to newest).
11) Lines 45-47: the effect of constraining models to AOD observations is discussed, amongst others, in the conclusions of O'Sullivan et al (2020). I suggest to mention these considerations here.
12) Lines 49 and 58: there is a large difference between a 40% underestimation (line 49) and 11-fold underestimation (line 58). Such a difference in evaluation should be explained to the reader.
13) The tuning parameters indicated at lines 193-198 all affect the dust at the time of emission, and leave the dust transport unaltered. Perhaps this should be clearly stated. Moreover, it would be useful to explain the effect of each tuning parameter on the emitted dust mass and the airborne PSD. Last but not least, to which observation was the tuning tailored? Fennec (line 195) was localised at some of the dust source regions (not all of them), whereas at line 198 tuning has been generalised to a much wider region (the Sahara and Atlantic).
14) Line 219: my understanding from what the authors wrote here is that for 0.05S-0.5S, the fraction 0.05-0.5 indicates a reduction in the Stokes velocity, not of the dust mass sedimented as the abstract and earlier paragraphs had suggested. Please clarify nicely this from the early stages of the paper.
15) Throughout the manuscript, the authors use the term "coarser dust". In atmospheric science, the standard terminology is "coarse dust" (and, where relevant, "giant dust"). Using the comparative form of the adjective implies an unnecessary contrast (e.g., coarser than fine particles), which is not needed given that the authors already define the category by a diameter threshold (d > 2.5 um). I recommend replacing "coarser" with the neutral and commonly used term "coarse". The same comment applies to the use of "finer". Additionally, please ensure that each dust category is explicitly defined by a clear diameter range, e.g. when you use the word "fine" the first time in the paper indicate that you mean d < 2.5 um.
16) There are many symbols and concepts in the methodology section that require an explanation. What are G and F for example? mass fluxes? why is there a horizontal dust flux that is different than the usual advection term in the atmosphere (concentration x velocity)? and where are the other quantities in equation 1 extracted from to apply the equation? the authors also mention a constant (C) and a tunable parameter (D): give the quantities used. It is also unclear how dust is redeposited to the surface (line 134) as I suppose that this is not yet the atmospheric sedimentation that the authors test, nor what happens to the dust that is emitted with d > 63 um.
17) I have similar questions for equation 2: where is the soil clay fracion taken from? and what is the reasoning behind the scaling factor (sum of G, i=1-9 / sum of G, i=1-6)?
18) Line 143: please give the prescribed values for absorption, scattering and asymmetry parameter. It is unclear how this works (I would assume absorption and scattering to vary with the dust amount and the particle size at least, not to be fixed and prescribed).
19) Line 151: please explain the conceptual difference between CM and TM. It may be beneficial to show figure 9 at this stage of the paper, to aid in explaining it.
20) Line 161: please clarify to the reader that impaction scavenging refers to the effect of cloud drops. Please clarify also if the dry deposition mentioned in the following line is due to TM + sedimentation together. Please clarify if the sedimentation velocity at line 164 is a vertical velocity pointing downward (like I would think).
21) Line 180 sees some symbols which may be confused with previously used symbols. R could be confused with R_A and R_B of equatin 3, and C could be confused with C of equation 1 and C_c of equation 4. Moreover, here C is a dust concentration, therefore how does it relate to M_i in equation 1? I suggest to use consistent and unequivocal symbols throughout your paper, which may mean changing some symbols used in the references that you cite.
22) In table 1, my understanding is that the representative diameter is the geometric mean of the bin edges. I suggest to clarify this in the caption.
23) In section 2.2 the authors mention 8 simulations (line 209) and an additional one (line 210). For better clarity, I suggest to include the control runs in the simulation counts and thus there would be 9 and 2. The whole text in the next few lines would have to be adapted to this change.
24) Line 276: the authors refer to bin numbers but the figure does not show them. Please refer to diameter, or alternatively add the bin numbers in the figure, so as to better guide the reader.
25) Lines 317-318: the terms MCA lands here in an unclear way for two reasons. First, the construction of the sentence make it sound like a property of the Control ("Control with MCA"); please construct the sentence better. Secondly, it is my understanding that the MCA in this article and in the reference cited (that I have not checked) is the mass median altitude (normally I would expect the centroid to be the centre of mass, not the median). Please consider adapting the term.
26) in figure 4a only 4 lines are shown and in 4d there are only 3 (there should be 5 of them). In the caption the sentence about normalisation to bin 6 is unclear, and the sentence on showing only > 0.001 feels unnecessary given that there are no gaps in the data.
27) Line 345: increase in mass concentration: an increase compared to what? is it a comparison between model runs? between the SABL and the SAL of a same model run? please guide the reader into looking at the same part of the figure that you are looking at.
28) Line 357: I cannot find the decrease above the SABL when looking at the figure. Please guide the reader's eyes.
29) Line 376: I suppose the authors meant Westerly (not Easterly) transport.
30) Line 503: "due to the mostly positive values": I think that here you want to say that the positive numbers in figure 8c reveal that there is less loss of particles; however the way you wrote sounds like these positive numbers are the CAUSE (not the method for revealing) this effect. Please make the wording clear.
31) Lines 504-506: 1% and 5% sound negligible to me compared to what we would like to obtain. Are we sure that it is worth mentioning these very small numbers, that are probably smaller that their uncertainty?
32) Lines 513-517: the increase of the air temperatures is really interesting and perhaps could be better explained. At line 513, do you refer to the surface temperature or to the temperature at a specific altitude? At line 514, the term "central SAL" is a bit unclear; as you refer to the 3-4 km altitude why not say "at 3-4 km altitude, i.e. at the centre of the dust vertical distribution in the SAL"? Moreover the sentence at the end of the page about the temperature inversion could be better explained and expanded.