the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 06 Jan 2026
Radiative Influence of Dust Aerosols on the Evolution of Tropical Storm Hermine
Abstract. This study investigates the impact of dust aerosols on the evolution of Tropical Storm Hermine (2022) using the Weather Research and Forecasting model coupled with Chemistry (WRF-Chem) and observational data from the NASA Convective Processes Experiment – Cabo Verde (CPEX-CV). The objective is to evaluate how varying initial dust aerosol conditions influence storm development and to uncover the mechanisms behind these effects. Three WRF-Chem simulations were conducted with different initial aerosol concentrations: one with no aerosols, one with realistic dust concentrations, and one with intermediate aerosol levels. The simulations were compared against observational data from CPEX-CV and the best track data from the United States' National Hurricane Centre, focusing on parameters such as wind, pressure, aerosol optical depth, and radar reflectivity. The results indicate that the radiative effect of dust aerosols led to a weaker and more disorganized storm system compared to simulations without the inclusion of dust, highlighting the critical role of dustradiation interactions in modifying storm intensity. Furthermore, the study found that the ECMWF's Atmospheric Composition Reanalysis 4 (CAMS) underestimated atmospheric dust concentrations, in comparisons to observations, underlining the necessity for accurate observational data to validate aerosol-related processes and improve model predictions. These findings emphasize the complexity of dust aerosol-storm interactions and the importance of improving aerosol representations in simulations of tropical cyclones.
- Preprint
(6781 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 19 Apr 2026)
- RC1: 'Comment on egusphere-2025-6413', Anonymous Referee #1, 21 Mar 2026 reply
-
RC2: 'Comment on egusphere-2025-6413', Anonymous Referee #2, 08 Apr 2026
reply
Review of “Radiative Influence of Dust Aerosols on the Evolution of Tropical Storm Hermine” by Garber, Pu, and Hallar.
General comments:
The manuscript presents interesting simulation results on the effects of Saharan dust on the evolution and intensity of Tropical Storm Hermine, which was observed during the NASA Convective Processes Experiment–Cabo Verde (CPEX-CV). The model is able to generate a reasonable simulation of Hermine when dust aerosols are included at a sufficient amount that they best match the observations. This simulation is compared to a simulation in which the default model set up is used, which underpredicted aerosols, and to a clean case with no aerosols.
The paper needs major revisions before it can be considered acceptable for publication. The effort needed may require more time than the journal typically allows for revisions. Some key points are detailed in the major comments section below. At a higher level, three key points need to be made.
- The analysis is rather superficial, focused mainly on differences between simulations, typically at only one or two times. It can be very difficult to interpret difference fields without also seeing the original fields. Furthermore, difference fields at only a single time provide insufficient evidence of the processes or impacts that the authors claim to find. If the authors expand their analysis to include more information on system evolution and net effects (e.g., domain averages, a time series of domain averaged profiles, Hovmoller diagrams, etc.), I think that may allow for much greater insight into processes than can be determined from the existing evidence. In addition, the paper often consists of a succession of figures for which only 2-3 sentences are written to describe the results, hence, the sense that the discussion is superficial. A deeper understanding of the simulation results should be reflected in the writing.
- The simulation uses a single moment microphysical scheme that appears to be unable to account for microphysical effects of the aerosols. If a double-moment aerosol-aware microphysical scheme is available in WRF, then that parameterization should really be used rather than the current one. The authors discuss the limitations of the microphysics scheme and yet reach conclusions about microphysical impacts that are not represented in the model. Furthermore, even if the authors were to use a more advanced microphysics scheme, the 3-km model grid spacing (minimum resolvable scale of 12 km) may not allow for sufficiently strong updrafts to generate adequate supersaturations for cloud droplet initiation and may therefore underrepresent the microphysical effects of aerosols. In fact, it could result in one believing that they are representing microphysical impacts of aerosols when instead the scheme is defaulting to some prescribed number of CCN. If the authors decide to focus solely on radiative impacts, then they should avoid drawing any conclusions about microphysical impacts of dust that are not represented in the simulations.
- Given the above issues, significant questions remain about the results such that it is hard to identify advancements beyond, or new knowledge of, dust-TC interactions compared to prior studies (see the list of prior studies in the fourth comment under minor comments).
Specific comments
Major comments:
Line 126: Is wet deposition included in the simulations? If not, it should be. The authors mention dry deposition but not wet deposition. If the latter is not included, then you will be overestimating aerosol impacts on the storm by keeping aerosols around longer than they should be, especially in the core of the storm where you see some significant differences.
Lines 175-177: Since you do not represent aerosol-cloud interactions in the microphysics scheme, this comparison is faulty. You are getting a similar looking pattern but obviously for different reasons. These plots show a mix of positive and negative values. It might be worthwhile having a side panel that shows the domain average difference as a function of height. Even better would be a time series of vertical profiles to capture system evolution.
Lines 186-187: This overlap is not evident in the figure since no dust information is included. Perhaps add some dust contours. Given the non-monotonic impacts on storm intensity (Fig. 12), one might wonder whether some of these differences are due to random perturbations generated by radiation effects. Is there a way to demonstrate the non-randomness of the results? Ensembles would be ideal but probably not within the scope of the paper. Given that microphysical effects of dust are not included, the impacts on relative humidity are either due to temperature changes or circulation (vertical transport) changes. To evaluate moisture changes, it would be better to look at specific humidity so that the change is independent of temperature changes. In terms of circulation changes, are the moisture changes consistent with the vertical motion differences? Are radiative effects possibly increasing moisture detrainment from low-level clouds compared to less dusty cases? The authors often rely on conjecture to explain differences in the simulations without digging deeper to demonstrate the actual underlying processes.
Figure 6 and later figures: Showing differences without showing the original fields is rather useless, particularly for fields that rapidly evolve. For all we know, many of these differences may reflect changes in positions of cloud and precipitation fields, i.e., shifts in the cold OLR corresponding to convective systems, that yield an unclear net effect. It certainly isn't clear that the differences indicate direct absorption and emissions by aerosols.
Lines 199-201: Again, these are really noisy difference fields and may largely reflect positional differences in convective cloud systems (spurred by radiation-induced random perturbations) rather than direct absorption/emission of radiation. In the top panel of Fig. 7 in particular, the largest differences appear to reflect differences in the positions and maybe characteristics of convective outflows (cold pools). The Clean and Extra figures should be provided in addition to the differences, but I think some method is needed to separate out the effects of cloud-system-generated differences (i.e., similar cloud systems at different locations in the domain) and radiative differences, or to say another way, random variations versus systematic differences. As mentioned previously, perhaps this can be done by looking at the vertical profile of the domain-averaged values and, even better, its variation in time.
Lines 217-220: Both cases have the main vorticity structures in the northeastern quadrant. The Extra case shows an extension of the cyclonic circulation to the southwest, but the actual center of circulation appears to be very close the southwestern-most vorticity maximum (see the northwesterly wind just SW of this vorticity maximum). Vorticity plots at a single time are insufficient evidence for the case that you are making. A better case can be made by looking at vorticity in a storm-centered reference frame and looking at a Hovmoller diagram of the low-level vorticity as a function of radius and time.
Lines 222-225: This is just a single time. How did the reflectivities vary in time? For all the reader knows, the Extra case may have had higher reflectivity earlier or later than this snapshot given that reflectivity can vary on relatively short time scales. The Extra case appears to have stronger convection well to the ENE that may have provided cold inflow into the forming eyewall, thereby weakening the inner convection and the storm intensity. As with vorticity, a Hovmoller diagram could prove useful here. Similar comments apply to vertical air motion (vertical air motion should not be given in knots but in meters per second). In fact, a figure showing Hovmoller diagrams of vorticity, vertical velocity and reflectivity side by side would be very informative. Tangential and radial winds might also be of interest.
Section 3.5: Why does the focus suddenly switch from differences between Clean and Extra to Clean and Intermediate? You should stick with the same cases (Clean and Extra). Intermediate comparisons can be an extra set of figures, but the switch shouldn't be made without justification. Another key question is why Intermediate produces a stronger impact on intensity than Extra. It suggests a non-monotonic relationship and makes one wonder how random convective events might alter the evolution of the storm. Do time series of maximum wind show similar results?
Lines 250-251: Dust is prevalent throughout the storm based on Fig. 2, so you can't really argue that only the outer bands were invigorated. In fact, if that were true, one would think that the Extra case should be even weaker. In general, there is insufficient analysis to explain the results. Analyses of time- and radially varying fields like vertical air motion, reflectivity, vorticity, and boundary layer potential or equivalent potential temperature might shed more light on the processes involved.
Minor comments:
Line 28: There are a lot of papers besides Braun (2010) that discuss the SAL. Include some of the early papers by Carlson and Prospero (1972, J. Applied Meteor.), Carlson (1979, Mon. Wea. Rev.), Karyampudi and Carlson (1988, J. Atmos. Sci.), Karyampudi and Pierce (2022, MWR), Dunion and Velden (2004, BAMS, DOI: 10.1175/BAMS-85-3-353 ), etc.
Lines 35-44: Cotton et al. (2012, Tropical Cyclogenesis Research and Review, DOI: 10.6057/2012TCRR03.05), Herbener et al. (2014, JAS, DOI: 10.1175/JAS-D-13-0202.1) and Lynn et al. (2016, JAS, DOI: 10.1175/JAS-D-14-0150.1) found that aerosols getting into the eyewall can intensify storms, but such periods may be followed by aerosols getting into outer convection at later times.
Lines 45-49: Karyampudi and Carlson (1988, JAS) and Karyampudi and Pierce (2002, MWR) found evidence of both enhancement and weakening of TCs due to the SAL. They should be mentioned as well.
There is a fairly large body of work examining radiative impacts of aerosols on TCs that are not discussed at any significant length or at all, but should be, including:
- Chen et al. (2010, J. Geophys. Res., doi:10.1029/2010JD014158)
- Reale et al. (2014, Geophys. Res. Letters, doi:10.1002/2014GL059918)
- Wang et al. (2014, Nature Clim. Change, DOI: 10.1038/NCLIMATE2144)
- Chen et al. (2015, Q. J. Royal Met. Soc., DOI:10.1002/qj.2542)
- Bretl et al. (2015, JGR-Atmos., doi:10.1002/2014JD022441)
- Chen et al. (2017, Atmos. Chem. Physics, doi.org/10.5194/acp-17-7917-2017)
- Pan et al. (2018, J. Climate, DOI: 10.1175/JCLI-D-16-0776.1)
- Strong et al. (2018, JGR-Atmos., doi.org/10.1029/2017JD027808)
- Shi et al. (2021, MWR, DOI: 10.1175/MWR-D-20-0344.1)
- Pan et al. (2024, JGR-Atmos, doi.org/10.1029/2023JD039245)
Lines 59-60: You highlight here the importance of using the observational data with models, but the only data that you use is the lidar data in one figure. You don't show dropsonde data or any other data, nor do you show MODIS observations of clouds and dust that would provide very useful context. The paper could benefit from better validation.
Line 71: Best track information and your Fig. 15 have it forming northeast of Cabo Verde, not southwest.
Line 74: It moved briefly to the NNW (during the period of your study) but then moved predominantly northward and northeastward. So you need to be more precise in your description.
Lines 75-77: Caution must be taken in generalizing this case. It was nearly completely embedded from the start within the dust layer rather than south of the dust/SAL boundary as is more typical of storms in the eastern Atlantic (Karyampudi and Carlson 1988, Braun 2010). You address the issue of generality in the conclusions, but perhaps a mention early on would be good as well.
Lines 104 and 106: By dust coefficients, do you mean coefficients used in the emission of dust from the surface?
Line 111: Change cloud to convection. The grid likely supports clouds at larger scales.
Line 113: Three km is marginal but often considered acceptable for simulating convection, i.e., convection permitting, but one needs to get to sub-km grid spacing to actually resolve convection. Your minimum resolvable scale is 4*DX or 12 km. WRF is particularly diffusive so the true resolution is likely larger than that, so not really capable of resolving convection.
Lines 120-121: That is not quite right. While aerosols can’t change the radiative properties of the clouds in these simulations, the aerosol-radiative coupling may change cloud system structures in a way that then changes cloud-radiative feedbacks on system evolution. Does the microphysics scheme (or the aerosol physics module) account for wet removal of aerosols? If not, then radiative effects within the storm core will be exaggerated and several of your findings may not be valid.
Lines 146-150: You should probably state earlier that you are simulating AOD at 550 nm. You can probably join these two paragraphs as well since the second paragraph seems to relate directly to the first.
Line 146: What is meant by mapping coefficients? Do you mean emission related coefficients?
Line 159: How were extinction coefficients calculated?
Line 161: Do you mean in the HALO data or in the simulations? I assume the former, but just checking.
Lines 166-171: First, the x-axis of the observations should be converted to distance using the aircraft speed so that the scales and locations of the features can be compared more directly. Second, what causes the white-out of the lidar features? It looks like the signal is completely extinguished in these shallow features, suggesting low-level liquid clouds (there can be no ice at these levels). If not liquid clouds, then the aerosol extinction must significantly exceed the 0.4 limit on the color scale. Assuming the white-out regions are liquid clouds, it may aid comparisons if you either include cloud extinction in the simulated plots or at least show contours of cloud water content.
Figures 3 and 5: The locations of these cross sections should be shown somewhere, preferably in the context of a horizontal plane view. Figure 2 might work but the times do not match.
Lines 188-190: It isn't clear why you show this figure here. It appears to be a cross section taken totally out of context since we don't know where it is located relative to the storm. Also note that such a dry air mass may be readily displaced by the envelope of air, i.e., the marsupial pouch of Dunkerton et al. (2009), surrounding the storm, so the implication that the storm experiences this dry air directly is unfounded. The last sentence seems unnecessary given that all of the cases have the same environment and there is no expectation for Clean to be any different.
Lines 202-203: These things should be verifiable from the data and can likely provide a clearer assessment than looking at the differences. As with other difference fields, it is hard to determine if there is a systematic change. Looking at the time series of the domain-average values may help.
Lines 225-226: That is one possible cause of asymmetry. Another is vertical wind shear (see image), which on larger scales was westerly and closer to the storm was southwesterly. Convection is often downshear to downshear-left, which would put the convection generally in the same area.
Lines 296-297: This statement ignores the fact that the intensity of the Extra case was in between Clean and Intermediate.
Line 298: Just the initial dust loading or also subsequent emissions of dust?
Lines 303-306: The increase in relative humidity in the Extra case appears to be throughout the core, so it does not support the idea of the outer convection being favored at the expense of the inner core.
Lines 318-320: This statement is rather nonsensical. By definition, if you remove radiative and microphysical impacts, then there is no mechanism for impact. Furthermore, there is no microphysical impact because the microphysics scheme does not appear to be aware of aerosols or has no coupling between aerosols and microphysical processes.
Line 321: Zhang et al. (2007) did not address mid-level moisture and all of their cases were well organized, so this statement is incorrect.
Technical corrections
Line 34: There appears to be an error here. Probably should be "optical and other properties" or something like that. Does the vertical distribution of aerosols actually affect the properties of the dust? Probably not, but it does impact the radiative and thermodynamic effects.
Line 54: CPEX-CV used the NASA DC-8 aircraft flying the HALO lidar. There is no HALO aircraft.
Line 55: Since you later refer to RF 09 and 10, define the acronym here or remove the acronym use.
Line 58. A period is needed at the end of the sentence.
Line 88: How does the CAMS acronym flow from Atmospheric Composition Reanalysis?
Lines 142-143: I recommend changing this sentence to "However, comparison of CAMS results to CPEX-CV observations suggests that the initial dust plume location was reasonably well simulated, but its magnitude was underestimated."
Figures: For most figures, the figure panel titles and axes text are too small to read.
Lines 247-248: There isn't much analysis here. They are just plots of winds. So change to something like "Wind fields (Fig. 14) ..."
Citation: https://doi.org/10.5194/egusphere-2025-6413-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 241 | 76 | 21 | 338 | 22 | 22 |
- HTML: 241
- PDF: 76
- XML: 21
- Total: 338
- BibTeX: 22
- EndNote: 22
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This study uses WRF-Chem to investigate the radiative impact of Saharan dust on Tropical Storm Hermine (2022), utilizing NASA CPEX-CV observations for model initiation and validation. While the application of a fully-coupled model to this event is appreciated, the manuscript needs major revisions before it can be published.
The whole story is not clear:
1) The title is “radiative influence”, but the paper does not effectively distinguish between radiation and microphysical effects. When using a model like WRF-Chem with MOSAIC, these effects often occur simultaneously without a methodological framework to distinguish between the two. And in some parts of results and discussion, the authors explicitly explain through the “dust’s role as effective cloud condensation and ice nuclei”.
2) It is not clear how the study advances over previous work. The Introduction mentions that “Despite decades of study, the net effect of Saharan dust on TC development is not fully resolved”; yet the title of this study is only on “radiative influence”. Also, Results and Discussion currently reconfirm existing studies without offering new insights into dust-TC interactions. The Introduction feels like a literature review that doesn't link to the specific gaps this study tries to fill. Similarly, the abstract focuses more on the summary of results than on the contribution and novelty.
3) The way that the observation-like dust aerosols are simulated is not very convincing. Why does doubling the initial conditions and mapping coefficients result in a consistent doubling of concentration? As the authors didn’t explicitly compare the reanalysis CAMS and observation-based CPEX-CV, the “50% underestimation in dust loading” is vague, without knowing which variable (AOD, extinction, mass) or which height/time was compared. This undermines the credibility of the setup for the Extra simulation as well. Moreover, is the “50% underestimation in CAMS” one of the results or just background information? It is written in the supposed result part in the abstract, but then in the Introduction without further elaboration in the Results.
4) The Results section is organized in a confusing way. Why put dust concentrations as the first section, as the simulations were set up according to different levels of concentration? Also, the third section is titled "Radiative and thermodynamic effects," but since the whole paper is about this, that title should be automatically a top-level theme, not a small sub-section.
5) The section on “track differences” is too short and not well explained. The authors are conflating Experimental Design with Physical Results. “The tracks remained geographically close throughout, minimizing environmental variability.” This is like a justification for the experimental design. But here, it is supposed to be the results. As a reader, we want to know what this means to the effect of dust aerosol on the storm. Is it because Hermine was a weak system dominated by strong background flow? Or because the radiative forcing of the dust was too weak to alter the storm's vertical tilt?
6) The Discussion is largely repeating the Results, without providing any new insights into the knowledge level. It’s hard to see how this paper advances beyond previous knowledge.
7) The figures need a lot of improvement. Current figures look like raw model output rather than scientific illustrations. 15 figures are too many, and they don't tell a coherent story. For example, in Fig. 3, the x-axis "Distance Along Flight Path" has no meteorological meaning. This is a human-defined trajectory, not a natural coordinate; how does it actually relate to the distance from the storm center, or anything meaningful? Besides, many figures have very small fonts, titles that overflow, and repetitive wording.
8) For the overall manuscript, especially Results and Discussion, the topic sentences do not stand out, and the paragraphs are not well organized, making it hard to capture the core messages.
Specific comments:
L187: “driven by radiation changes”. How and what kind of changes? Not clearly stated.
L268: confused. Why “track comparisons” can ensure “similar environmental conditions”? It is usually the other way around.
Fig.3: What does the white space mean in the top panel? If there is no data, it should be stated. I am not convinced that the "Extra" run correctly captures the vertical plume based on this plot.