the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 18 Dec 2024
Saharan dust linked to European hail events
Abstract. Saharan dust significantly influences hail occurrence in Europe. Using Copernicus Atmosphere Monitoring Service (CAMS) and reanalysis data, crowd-sourced hail reports, lightning data, and radar measurements, we find a strong correlation between elevated dust loading and hail events. Hail coverage exceeding 28 % of 1°×1° grid cells only occurs when dust loading surpasses 2.4 mg m−2, while on hail days the median dust load is 1.82 times higher than on non-hail days (7σ difference). This effect is particularly strong along the Alpine crest, central France, eastern Germany, Austria, and Eastern Europe, where median dust loads more than double on hail days.
By grouping data according to synoptic weather patterns, we confirm that hail days consistently exhibit higher dust concentrations regardless of prevailing synoptic conditions, supporting the robust link between dust and hail. Peak hail activity occurs at 38 mg m−2 or a dust optical depth of 0.033, suggesting enhanced cloud and ice nucleation. Above this range, hail frequency declines, likely due to microphysical or radiative constraints.
Crowd-sourced reports show significantly more hail events on high-dust days, with up to 10 times more reports for hail >20 mm. Statistical hail models, including a logistic regression model (LRM) and a generalized additive model (GAM), rank dust as one of the top three predictors. Its inclusion increases the critical success index (CSI) by 5 % (LRM) and 12 % (GAM), and boosts explained variance in the GAM by 6 %. These findings identify Saharan dust as a key modulator of European hail activity.
Competing interests: The contact author has declared that neither of the authors has any competing interests.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(1170 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(1170 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
- RC1: 'Comment on egusphere-2024-3924', Anonymous Referee #1, 23 Feb 2025
-
AC1: 'Comment on egusphere-2024-3924', Killian Brennan, 26 Feb 2025
After some of the comments from the first reviewer, we checked the preprint which was posted on the 18th of December, and realized that the figure axis labels of most figures are messed up. The figures were intact in the initial submission, however it appears that some fonts were not properly embedded.
The missing figure elements make the figures and, by extension, the manuscript difficult to follow, we apologize for this inconvenience.Following suggestions from the reviewer, we've included rasterized versions of the figures in the manuscript in this comment.
-
RC2: 'Comment on egusphere-2024-3924', Anonymous Referee #2, 02 Jun 2025
This is a very interesting and relevant study investigating the effect of dust concentration on hail using observations. The methodology is sound, although the authors could be a bit more careful in the interpretation of some of the results and discuss the uncertainties more (see specific comments). I like that the manuscript is kept short and precise. However, in some parts a bit more detail might be necessary in both the literature background and the analysis (see specific comments). Most of my comments are minor and I don’t see any reason to stop publication, but I strongly recommend to work on the following aspects.
Specific comments:
- Lines 27-67: Perhaps the authors are more familiar with the topic of aerosol-cloud interactions (focused on hail) than me, but isn’t this topic much less clear than portrayed here? To my knowledge, there are some contradicting results in the literature (see e.g., the sections on aerosol effects in the reviews of Allen et al. 2020 and Raupach et al. 2021) while herein the different physical processes are portrayed as clear picture in just a few short sentences in each paragraph. I’m no expert on this topic but I think some more context might be good on what processes are still uncertain.
Allen, J. T., Giammanco, I. M., Kumjian, M. R., Punge, H. J., Zhang, Q., Groenemeijer, P., et al. (2020). Understanding hail in the Earth system.Reviews of Geophysics, 58, e2019RG000665. https://doi.org/10.1029/2019RG000665
Raupach, T.H., Martius, O., Allen, J.T. et al. The effects of climate change on hailstorms. Nat Rev Earth Environ 2, 213–226 (2021). https://doi.org/10.1038/s43017-020-00133-9 - Most Figures are missing labels. I saw that you added the corrected Figures in your reply to reviewer 1, so this seems resolved?
- Line 39: Do you mean here that beside the change in stability, thermal convection is reduced, which inhibits convection initiation? If yes, I suggest writing „thermal convection“ or „boundary layer thermals“ instead of just „convection“. Furthermore, this „negative“ impact of aerosols on CI has also been discussed as a possibly important factor in strong Saharan dust scenarios over Europe (Seifert et al. 2023, Fischer et al. 2025). I think these negative effects could be discussed a bit more in the manuscript since this fits the decline in hail occurence with high dust concentrations in your study (see also comment 14).
Seifert, A., Bachmann, V., Filipitsch, F., Förstner, J., Grams, C. M., Hoshyaripour, G. A., Quinting, J., Rohde, A., Vogel, H., Wagner, A., and Vogel, B.: Aerosol-cloud-radiation interaction during Saharan dust episodes: the dusty cirrus puzzle, Atmospheric Chemistry and Physics, 23, 6409–6430, https://doi.org/10.5194/acp-23-6409-2023, 2023.
Fischer, J., Groenemeijer, P., Holzer, A., Feldmann, M., Schröer, K., Battaglioli, F., Schielicke, L., Púčik, T., Gatzen, C., Antonescu, B., and the TIM Partners: Invited perspectives: Thunderstorm Intensification from Mountains to Plains, EGUsphere, https://doi.org/10.5194/egusphere-2024-2798, 2024 - Line 117: Better write „the latter“ instead of „it“ to make clear that you are referring to EURADHAIL (right?).
- 3: I’m not sure I understand this fig or how you interpret it. There is a peak, but the decrease in hail fraction with lower dust concentrations is very small. So doesn’t this show that hail potential doesn’t change much at lower concentrations?
Perhaps related to this, why is the plot cut off at low concentrations? - To interpret Fig. 3 and more generally the context of dust concentrations I think it would be helpful to add a Figure showing a histogram of dust concentrations underlying your analysis. In other words, how frequent are concentrations e.g., of >20 mg/m2. You only briefly mention some context on the underlying distribution in line 164.
- Section 3.2: I like that you looked into the possible link to weather patterns and I mostly agree with your conclusion. However, I think it is still possible that even within one general synoptic setting higher dust could not be causally linked to more hail but just be correlated with larger-scale processes important for hail formation. For example, it is known that steep lapse-rates are important for hail, which is often found when an elevated mixed layer is advected from the Iberian Peninsula or Africa, areas which are major sources of dust (Schultz et al. 2025). It’s hard to say how well your clustering approach is capturing complex processes like this. Is it possible that even within one of your clusters, there might be days in which the flow supports EML and dust formation and in others it doesn’t? Then dust would only be correlated to hail because of the increased lapse-rates. Considering mentioning this possibility.
Schultz, D. M., M. V. Young, and D. J. Kirshbaum, 2025: The Spanish Plume Elevated Mixed Layer: A Review of Its Use and Misuse within the Scientific Literature. Mon. Wea. Rev., 153, 737–761, https://doi.org/10.1175/MWR-D-24-0139.1. - Line 187: Would it be worth showing the general flow pattern for your clusters? I think this could be interesting because it would show in what synoptic scenarios dust is not important (0.81?).
- Line 189: The coordinate ranges don’t mean much to me. Consider adding a map for target regions (e.g., add to fig 2) or at least say more about why they were chosen.
Also, remove „a“ before „different“. - Line 213: Unclear why „(1)“ is added here.
- Line 239: I don’t follow your argument here. Is the dust concentration so heterogeneous from one grid point to the next that averaging over such a small regions would change the threshold so significantly?
- Section 4: I think the statistical modeling is a good idea to highlight the relevance of dust. However, one may criticize that adding and almost any additional predictor is expected to enhance model performance slightly, so comparing the slight improvements of the model with and without dust could be misleading, no?
- Relatedly, it‘s interesting and supporting your conclusions that dust is more important than other predictors like wind shear, but this opens another question as wind shear is known to be important for hail storms. Could you elaborate on this? Are wind shear and dust highly-correlated so that the model only needs one? Or perhaps wind shear is not so important because the model is trained on hail coverage and not hail size?
- In Fig. 5, it looks like the added value of dust as predictor is mostly from its negative effects (see comment 3) yet the hail-enhancing influence is emphasized a lot more in your text (take for example the manuscript title and abstract). Consider writing about both effects in a more balanced way.
If you disagree, perhaps you could also show somehow that the positive values in Fig. 5 are just as or more important because these dust concentrations occur more frequently? - Your study is mostly based on EURADHAIL and POH as truth for hail. Both heavily rely on radar reflectivity, which can also be high in the presence of strong liquid precipitation. So, could the link you find between dust and hail be at least partially a result of the influence of dust on heavy precip (Zhu et al. 2024)? In other words, even with hail staying equal, an increase in precipitation intensity would result in an increase in reflectivity max and reflectivity area and therefore falsely show an impact of dust in your study. I agree that hail is likely dominating these reflectivity-based parameters, but the question is how big the impact of intensifying precip is. If you agree, this uncertainty should be discussed.
- Related to comment 1, the strong influence of dust on hail you suggest opens the question why other regions of the world which are less directly influenced by major dust sources like the Sahara are still having intense hailstorms hail (e.g., Northern US or South America). Any thoughts?
- I also agree with comment 2 of reviewer 1: How do you know that Saharan dust is dominating? Perhaps you could elaborate a bit on the robustness of the CAMS data and why Saharan dust is most likely?
Citation: https://doi.org/10.5194/egusphere-2024-3924-RC2 - Lines 27-67: Perhaps the authors are more familiar with the topic of aerosol-cloud interactions (focused on hail) than me, but isn’t this topic much less clear than portrayed here? To my knowledge, there are some contradicting results in the literature (see e.g., the sections on aerosol effects in the reviews of Allen et al. 2020 and Raupach et al. 2021) while herein the different physical processes are portrayed as clear picture in just a few short sentences in each paragraph. I’m no expert on this topic but I think some more context might be good on what processes are still uncertain.
- AC2: 'Comment on egusphere-2024-3924', Killian Brennan, 07 Aug 2025
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2024-3924', Anonymous Referee #1, 23 Feb 2025
Dust aerosols have an important influence on cloud formation and development. This manuscript analyzes the influence of Saharan dust on hail in Europe, which has important scientific significance. Nevertheless, the manuscript leaves much to be desired. Here are some specific comments:
1. Lines 43-44: It’s inappropriate to cite unpublished papers.
2. How did the authors determine that all the dust came from the Sahara? Relevant weather pattern analysis is required.
3. Lines 81-84: The author declared that they mainly focus on to investigate the influence of dust aerosol on hail occurrence, but only local days with lightning were included. Can it be understood that hail and lightning occur simultaneously? It should be described in more detail to make it easier for readers to understand.
4. When the availability of the OPERA data is less than 100% in a 1°×1° grid, how is the hail area fraction calculated?
5. Lines 106-109: Do you mean that there are only 140 grid-points are available using EURADHAIL for determine hail events? I found that it conflicts with Figure 2.
6. Lines 114-17: This sentences “POH is an empirical hail detection algorithm estimating ground-level hail probability (0 – 100%) based on the vertical distance between the 45 dBZ echo top height and the freezing level height, following Waldvogel et al. (1979). This approach is more accurate in capturing hail events than EURADHAIL, since it does not include the freezing level.” confused me. Freezing level height is used to judge hail events, why does the author claim that this algorithm is more accurate than EURADHAIL because it does not include freezing level?
7. Many of the labels on the horizontal and vertical axes of the figures are incomplete and need to be carefully modified.
8. Line 145: Thunderstorm day or hail day, which one is right? The same question in the title of Figure 1 and Figure 4.
9. More detailed description about the Q should be added in Figure 1. In addition, how to calculated the fraction of hail days in Figure 3?
10. In figure 1, the mass of dust concentration is only divided into 2 groups. If the dust mass concentration is divided into three groups, does the maximum hail area fraction change with the dust mass concentration group as described in the manuscript？
11. Lines 156-158: Such analysis does not make sense, since aerosols of different scales co-exist in hail days.
12. Lines 180-184 and Lines 190-192: How does the author determine the optimal number of clustering centers?
13. Lines 225-226: “availability” should be “variable”. Why are different moisture variable used in LRM and GAM models.
14. Lines 226-229: This sentence confused me. Which variable is most important for hail event prediction, dust loading or CAPE？Citation: https://doi.org/10.5194/egusphere-2024-3924-RC1 -
AC1: 'Comment on egusphere-2024-3924', Killian Brennan, 26 Feb 2025
After some of the comments from the first reviewer, we checked the preprint which was posted on the 18th of December, and realized that the figure axis labels of most figures are messed up. The figures were intact in the initial submission, however it appears that some fonts were not properly embedded.
The missing figure elements make the figures and, by extension, the manuscript difficult to follow, we apologize for this inconvenience.Following suggestions from the reviewer, we've included rasterized versions of the figures in the manuscript in this comment.
-
RC2: 'Comment on egusphere-2024-3924', Anonymous Referee #2, 02 Jun 2025
This is a very interesting and relevant study investigating the effect of dust concentration on hail using observations. The methodology is sound, although the authors could be a bit more careful in the interpretation of some of the results and discuss the uncertainties more (see specific comments). I like that the manuscript is kept short and precise. However, in some parts a bit more detail might be necessary in both the literature background and the analysis (see specific comments). Most of my comments are minor and I don’t see any reason to stop publication, but I strongly recommend to work on the following aspects.
Specific comments:
- Lines 27-67: Perhaps the authors are more familiar with the topic of aerosol-cloud interactions (focused on hail) than me, but isn’t this topic much less clear than portrayed here? To my knowledge, there are some contradicting results in the literature (see e.g., the sections on aerosol effects in the reviews of Allen et al. 2020 and Raupach et al. 2021) while herein the different physical processes are portrayed as clear picture in just a few short sentences in each paragraph. I’m no expert on this topic but I think some more context might be good on what processes are still uncertain.
Allen, J. T., Giammanco, I. M., Kumjian, M. R., Punge, H. J., Zhang, Q., Groenemeijer, P., et al. (2020). Understanding hail in the Earth system.Reviews of Geophysics, 58, e2019RG000665. https://doi.org/10.1029/2019RG000665
Raupach, T.H., Martius, O., Allen, J.T. et al. The effects of climate change on hailstorms. Nat Rev Earth Environ 2, 213–226 (2021). https://doi.org/10.1038/s43017-020-00133-9 - Most Figures are missing labels. I saw that you added the corrected Figures in your reply to reviewer 1, so this seems resolved?
- Line 39: Do you mean here that beside the change in stability, thermal convection is reduced, which inhibits convection initiation? If yes, I suggest writing „thermal convection“ or „boundary layer thermals“ instead of just „convection“. Furthermore, this „negative“ impact of aerosols on CI has also been discussed as a possibly important factor in strong Saharan dust scenarios over Europe (Seifert et al. 2023, Fischer et al. 2025). I think these negative effects could be discussed a bit more in the manuscript since this fits the decline in hail occurence with high dust concentrations in your study (see also comment 14).
Seifert, A., Bachmann, V., Filipitsch, F., Förstner, J., Grams, C. M., Hoshyaripour, G. A., Quinting, J., Rohde, A., Vogel, H., Wagner, A., and Vogel, B.: Aerosol-cloud-radiation interaction during Saharan dust episodes: the dusty cirrus puzzle, Atmospheric Chemistry and Physics, 23, 6409–6430, https://doi.org/10.5194/acp-23-6409-2023, 2023.
Fischer, J., Groenemeijer, P., Holzer, A., Feldmann, M., Schröer, K., Battaglioli, F., Schielicke, L., Púčik, T., Gatzen, C., Antonescu, B., and the TIM Partners: Invited perspectives: Thunderstorm Intensification from Mountains to Plains, EGUsphere, https://doi.org/10.5194/egusphere-2024-2798, 2024 - Line 117: Better write „the latter“ instead of „it“ to make clear that you are referring to EURADHAIL (right?).
- 3: I’m not sure I understand this fig or how you interpret it. There is a peak, but the decrease in hail fraction with lower dust concentrations is very small. So doesn’t this show that hail potential doesn’t change much at lower concentrations?
Perhaps related to this, why is the plot cut off at low concentrations? - To interpret Fig. 3 and more generally the context of dust concentrations I think it would be helpful to add a Figure showing a histogram of dust concentrations underlying your analysis. In other words, how frequent are concentrations e.g., of >20 mg/m2. You only briefly mention some context on the underlying distribution in line 164.
- Section 3.2: I like that you looked into the possible link to weather patterns and I mostly agree with your conclusion. However, I think it is still possible that even within one general synoptic setting higher dust could not be causally linked to more hail but just be correlated with larger-scale processes important for hail formation. For example, it is known that steep lapse-rates are important for hail, which is often found when an elevated mixed layer is advected from the Iberian Peninsula or Africa, areas which are major sources of dust (Schultz et al. 2025). It’s hard to say how well your clustering approach is capturing complex processes like this. Is it possible that even within one of your clusters, there might be days in which the flow supports EML and dust formation and in others it doesn’t? Then dust would only be correlated to hail because of the increased lapse-rates. Considering mentioning this possibility.
Schultz, D. M., M. V. Young, and D. J. Kirshbaum, 2025: The Spanish Plume Elevated Mixed Layer: A Review of Its Use and Misuse within the Scientific Literature. Mon. Wea. Rev., 153, 737–761, https://doi.org/10.1175/MWR-D-24-0139.1. - Line 187: Would it be worth showing the general flow pattern for your clusters? I think this could be interesting because it would show in what synoptic scenarios dust is not important (0.81?).
- Line 189: The coordinate ranges don’t mean much to me. Consider adding a map for target regions (e.g., add to fig 2) or at least say more about why they were chosen.
Also, remove „a“ before „different“. - Line 213: Unclear why „(1)“ is added here.
- Line 239: I don’t follow your argument here. Is the dust concentration so heterogeneous from one grid point to the next that averaging over such a small regions would change the threshold so significantly?
- Section 4: I think the statistical modeling is a good idea to highlight the relevance of dust. However, one may criticize that adding and almost any additional predictor is expected to enhance model performance slightly, so comparing the slight improvements of the model with and without dust could be misleading, no?
- Relatedly, it‘s interesting and supporting your conclusions that dust is more important than other predictors like wind shear, but this opens another question as wind shear is known to be important for hail storms. Could you elaborate on this? Are wind shear and dust highly-correlated so that the model only needs one? Or perhaps wind shear is not so important because the model is trained on hail coverage and not hail size?
- In Fig. 5, it looks like the added value of dust as predictor is mostly from its negative effects (see comment 3) yet the hail-enhancing influence is emphasized a lot more in your text (take for example the manuscript title and abstract). Consider writing about both effects in a more balanced way.
If you disagree, perhaps you could also show somehow that the positive values in Fig. 5 are just as or more important because these dust concentrations occur more frequently? - Your study is mostly based on EURADHAIL and POH as truth for hail. Both heavily rely on radar reflectivity, which can also be high in the presence of strong liquid precipitation. So, could the link you find between dust and hail be at least partially a result of the influence of dust on heavy precip (Zhu et al. 2024)? In other words, even with hail staying equal, an increase in precipitation intensity would result in an increase in reflectivity max and reflectivity area and therefore falsely show an impact of dust in your study. I agree that hail is likely dominating these reflectivity-based parameters, but the question is how big the impact of intensifying precip is. If you agree, this uncertainty should be discussed.
- Related to comment 1, the strong influence of dust on hail you suggest opens the question why other regions of the world which are less directly influenced by major dust sources like the Sahara are still having intense hailstorms hail (e.g., Northern US or South America). Any thoughts?
- I also agree with comment 2 of reviewer 1: How do you know that Saharan dust is dominating? Perhaps you could elaborate a bit on the robustness of the CAMS data and why Saharan dust is most likely?
Citation: https://doi.org/10.5194/egusphere-2024-3924-RC2 - Lines 27-67: Perhaps the authors are more familiar with the topic of aerosol-cloud interactions (focused on hail) than me, but isn’t this topic much less clear than portrayed here? To my knowledge, there are some contradicting results in the literature (see e.g., the sections on aerosol effects in the reviews of Allen et al. 2020 and Raupach et al. 2021) while herein the different physical processes are portrayed as clear picture in just a few short sentences in each paragraph. I’m no expert on this topic but I think some more context might be good on what processes are still uncertain.
- AC2: 'Comment on egusphere-2024-3924', Killian Brennan, 07 Aug 2025
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 756 | 116 | 24 | 896 | 28 | 41 |
- HTML: 756
- PDF: 116
- XML: 24
- Total: 896
- BibTeX: 28
- EndNote: 41
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Cited
1 citations as recorded by crossref.
Killian P. Brennan
Lena Wilhelm
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(1170 KB) - Metadata XML
Dust aerosols have an important influence on cloud formation and development. This manuscript analyzes the influence of Saharan dust on hail in Europe, which has important scientific significance. Nevertheless, the manuscript leaves much to be desired. Here are some specific comments:
1. Lines 43-44: It’s inappropriate to cite unpublished papers.
2. How did the authors determine that all the dust came from the Sahara? Relevant weather pattern analysis is required.
3. Lines 81-84: The author declared that they mainly focus on to investigate the influence of dust aerosol on hail occurrence, but only local days with lightning were included. Can it be understood that hail and lightning occur simultaneously? It should be described in more detail to make it easier for readers to understand.
4. When the availability of the OPERA data is less than 100% in a 1°×1° grid, how is the hail area fraction calculated?
5. Lines 106-109: Do you mean that there are only 140 grid-points are available using EURADHAIL for determine hail events? I found that it conflicts with Figure 2.
6. Lines 114-17: This sentences “POH is an empirical hail detection algorithm estimating ground-level hail probability (0 – 100%) based on the vertical distance between the 45 dBZ echo top height and the freezing level height, following Waldvogel et al. (1979). This approach is more accurate in capturing hail events than EURADHAIL, since it does not include the freezing level.” confused me. Freezing level height is used to judge hail events, why does the author claim that this algorithm is more accurate than EURADHAIL because it does not include freezing level?
7. Many of the labels on the horizontal and vertical axes of the figures are incomplete and need to be carefully modified.
8. Line 145: Thunderstorm day or hail day, which one is right? The same question in the title of Figure 1 and Figure 4.
9. More detailed description about the Q should be added in Figure 1. In addition, how to calculated the fraction of hail days in Figure 3?
10. In figure 1, the mass of dust concentration is only divided into 2 groups. If the dust mass concentration is divided into three groups, does the maximum hail area fraction change with the dust mass concentration group as described in the manuscript？
11. Lines 156-158: Such analysis does not make sense, since aerosols of different scales co-exist in hail days.
12. Lines 180-184 and Lines 190-192: How does the author determine the optimal number of clustering centers?
13. Lines 225-226: “availability” should be “variable”. Why are different moisture variable used in LRM and GAM models.
14. Lines 226-229: This sentence confused me. Which variable is most important for hail event prediction, dust loading or CAPE？