the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
The impact of Sahara dust aerosols on the three-dimensional structure of precipitation systems of different sizes in spring
Abstract. Saharan dust aerosols interacting with clouds and precipitation in the Atlantic Ocean's intertropical convergence zone can significantly impact storm microphysical and thermodynamic processes. Previous satellite research often focused on individual, km-scale rain pixels, neglecting interconnections among different locations. This study innovatively employs a clustering method to group satellite precipitation radar-observed profiles into organized precipitation systems (PSs) of varying horizontal dimensions. Key features such as the mean storm top height, 85-GHz polarization-corrected microwave brightness temperature, and horizontal area with specific radar reflectivity per layer are analyzed to uncover system-level precipitation characteristics. Observations indicate that dust-laden PSs have higher storm tops, broader upper-level precipitation areas with more large particles, stronger ice scattering signals, and heavier surface rain rates than clean systems. These PSs also exhibit greater convective available potential energy (CAPE) and distinct differences in related dynamic and moisture conditions. Partial correlation and sensitivity analyses revealed that CAPE-induced changes are the primary confounding factor for dust aerosol effects. Notably, even after constraining CAPE and other thermodynamic factors, significant dust-related PS changes persist. This implies that, under comparable thermodynamic conditions, Saharan dust aerosols may enhance mid- and upper-level ice heterogeneous nucleation, thereby increasing the number of ice particles, releasing extra latent heat, and invigorating storms. Overall, this study offers a novel perspective on how dust aerosols affect organized precipitation systems.
- Preprint
(2028 KB) - Metadata XML
-
Supplement
(971 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
CC1: 'Comment on egusphere-2025-2799', Xiong Hu, 27 Aug 2025
-
AC1: 'Reply on CC1', Jing Xi, 29 Sep 2025
Responses to Community’ Comments
We are sincerely grateful to the editors and the community for their valuable time and constructive feedback on our manuscript. The comments are insightful and valuable, and have helped us to clarify and improve our study. We address each point raised in detail below, with our responses provided in bold. We have attached the complete response document (Responses to CC1.pdf) and three images in the compressed file of the supplementary materials.
Community #1
General Comments: The authors utilized a variety of observational data and reanalysis data to study the impact of dust aerosols on the three-dimensional structure of precipitation systems of different sizes. Nevertheless, certain methodological and interpretive aspects warrant further elaboration and refinement.
Reply: We thank the community reviewer for the valuable time and constructive comments, which have helped us to improve our manuscript. All comments have been addressed item by item.
Q1: As a spectral instrument, MODIS cannot directly observe aerosols beneath clouds. Although the authors employed a spatiotemporal interpolation method for aerosol matching, it is worth clarifying whether a cloud fraction threshold was applied during the interpolation process, particularly for PS regions with high cloud coverage.
Reply: We appreciate this insightful question. In our study, no cloud fraction threshold was applied during the interpolation process. To assess the potential influence of cloud coverage on our spatiotemporal interpolation approach, we conducted sensitivity tests using Modern-Era Retrospective Analysis for Research and Applications for version 2 (MERRA-2) data. Specifically, we artificially removed varying proportions of valid data to simulate different cloud cover conditions. For each precipitation system (PS), the averaged MERRA-2 AOD in the PS region was taken as the “true” AOD. Then, these AOD data were removed (white blocks in Fig. 1), and additional values were randomly removed from surrounding areas (gray blocks) to represent different cloud fractions. Our interpolation algorithm was then applied to the AOD data under varying cloud cover conditions, and compared with the true values. Figures 2 and 3 summarize the results.
Across different missing data fractions, the interpolated AOD agrees well with the “true” AOD, with root mean square error (RMSE) remaining low and correlation coefficients exceeding 0.8. Although performance slightly decreases with increasing missing data (e.g., declining correlation and slightly higher RMSE), the overall impact remains minor. This result is likely because the frequent Saharan dust outbreaks in the study region, which persist for several days. Thus, even under high cloud cover condition, valid data from surrounding grids and adjacent days still provide sufficient information to estimate dust aerosol conditions of PSs.
Q2: The study categorizes PSs into small (<2000 km²), medium (2000-10000 km²), and large (>10000 km²) classes based on their horizontal area. Could the authors please specify if these area thresholds were defined with reference to the climatological characteristics of PSs commonly found in the tropical Atlantic ITCZ region?
Reply: Thank you for raising this point. The area thresholds used in this study were determined based on previous studies on the characteristics of PSs. For instance, Liu et al. (2019) classified PSs with areas >2000 km² as mesoscale convective systems (MCSs) in their analysis of the intensity, height, and size variations of PSs under El Niño–Southern Oscillation conditions in the tropics and subtropics. They also found that PSs exceeding 10,000 km² contributed significantly to the annual mean rainfall (Fig. 7e in their paper). Similar thresholds have also been widely adopted in other studies (Zipser et al., 2008; Liu et al., 2017).
Liu, C., Chen, B., and Mapes, B. E.: Relationships between Large Precipitating Systems and Atmospheric Factors at a Grid Scale, Journal of the Atmospheric Sciences, 74, 531-552, 10.1175/jas-d-16-0049.1, 2017.
Liu, N., Liu, C., and Lavigne, T.: The Variation of the Intensity, Height, and Size of Precipitation Systems with El Niño–Southern Oscillation in the Tropics and Subtropics, Journal of Climate, 32, 4281-4297, 10.1175/jcli-d-18-0766.1, 2019.
Zipser, E. J., Liu, C., Cecil, D. J., Nesbitt, S. W., and Sherwood, S.: A Cloud and Precipitation Feature Database from Nine Years of TRMM Observations, Journal of Applied Meteorology and Climatology, 47, 2712-2728, 10.1175/2008jamc1890.1, 2008.
Q3: The paper primarily focuses on the aerosol-cloud interaction process involving dust acting as ice nuclei. Could the authors elaborate on whether a more quantitative investigation was conducted regarding the associated water-phase processes? Furthermore, while the dust's radiative effect is not discussed in detail within the text, it is depicted in the Fig. 10. Could this aspect be explained more thoroughly?
Reply: Thank you for this valuable comment. In our study, we primarily focused on PSs with vertical development exceeding 6 km, and did not conduct a quantitative investigation of the associated water-phase processes. The microphysical processes within PSs are complex, and changes in the ice-phase processes due to the ice nuclei (IN) effect can also influence the liquid-phase processes below the freezing level. However, conducting a quantitative analysis of these processes is challenging when relying solely on observational data. In future work, we can perform statistical analysis on PSs that develop in warm clouds, which will allow for a more in-depth investigation of the liquid-phase processes and their interactions with aerosols.
For the second question, it is generally recognized that dust radiative effects stabilize the atmosphere and suppress convection. Therefore, the observed enhancement of PS development in our study is more likely driven by microphysical processes, including the IN effect and the CCN effect. However, a few modeling studies (e.g., Cheng et al., 2019) have shown that dust radiative effects can delay convection initiation, allowing for energy accumulation, which may ultimately lead to more intense convective development once triggered. This highlights the complexity of dust radiative effects, thus it is difficult to quantify their impact using observational data. Future modeling studies will be needed to conduct sensitivity experiments to disentangle the contributions of dust radiative and microphysical effects.
Cheng, C.-T., Chen, J.-P., Tsai, I. C., Lee, H.-H., Matsui, T., Earl, K., Lin, Y.-C., Chen, S.-H., and Huang, C.-C.: Impacts of Dust–Radiation versus Dust–Cloud Interactions on the Development of a Modeled Mesoscale Convective System over North Africa, Monthly Weather Review, 147, 3301-3326, 10.1175/mwr-d-18-0459.1, 2019.
Q4: Figure 5 shows a reduction in the 20 dBZ area below the freezing level for stratiform precipitation in small- and medium-sized PSs under dusty conditions, which the authors attribute to the evaporation effect associated with the Saharan Air Layer. Is there more direct evidence supporting this proposed mechanism? For instance, was a significant variety in low-level humidity co-observed?
Reply: Table 4 of the manuscript presents statistical characteristics of meteorological variables for PSs under clean and dusty conditions. RHlow, defined as the mean relative humidity between 1000 and 850 hPa, is consistently lower under dusty conditions across all PS size categories. The difference of small-sized PSs is statistically significant at the 95% confidence level.
Q5: In Table 5, the sample sizes across different CAPE bins are notably imbalanced (e.g., for large PSs in the CAPE5 bin: clean n=5, dusty n=14). The statistical reliability of results derived from such small sample sizes is a concern.
Reply: Thanks for pointing this out. Large-sized PSs are much less frequent than smaller ones, and sampling is further constrained by the precipitation radar (PR) swath, as only PSs untruncated by the edge of orbit were selected in this study. Although significance tests were applied, the limited sample sizes may reduce result robustness. We will explicitly note this limitation in the revised manuscript when presenting Table 5.
-
AC2: 'Reply on CC1', Jing Xi, 29 Sep 2025
Publisher’s note: this comment is a copy of AC1 and its content was therefore removed on 1 October 2025.
Citation: https://doi.org/10.5194/egusphere-2025-2799-AC2
-
AC1: 'Reply on CC1', Jing Xi, 29 Sep 2025
-
RC1: 'Comment on egusphere-2025-2799', Anonymous Referee #1, 08 Sep 2025
General Comments: The aim of this manuscript is to statistically understand the effect of dust aerosols on the three-dimensional structure of precipitation systems of different sizes. This is an interesting and valuable attempt, as it is not common to study aerosol effects based on a large number of observational samples with the whole precipitation system as the research unit. Also, the authors have carefully considered the influence of meteorological conditions and employed multiple approaches (e.g., partial correlation analysis, and CAPE constraint) for investigation. So I think this work is well-constructed and scientifically meaningful, hence can be accepted for publication after the minor issues are addressed.
Major Comments:
- L177-180: The authors categorized PSs into three types: small-sized (< 2000 km²), medium-sized (between 2000 km² and 10000 km²), and large-sized (> 10000 km²). Please clarify the reason for selecting these specific thresholds.
- Figure 3: Why do other characteristics of PSs significantly increase under dusty conditions, while the differences in PS areas between clean and dusty conditions are not apparent?
- Section 4: In the analysis of physical mechanisms, this manuscript mentioned both the CCN effect and the IN effect of dust, but failed to clearly distinguish between these two effects. A more explicit elaboration would be necessary and beneficial.
Minor Comments:
- L14: The term ‘dimensions’ in ‘varying horizontal dimensions’ may be misleading, as it refers to size rather than dimension here.
- L21: I would recommend rephrasing the sentence ‘significant dust-related Ps changes persist’ to ‘significant dust-induced changes in PS properties persist’ for greater clarity and precision.
- Figure 1: PS2 and PS4 have noticeably smaller areas compared with the other PSs. As stated in L109-110, only PSs larger than 80 km2 were selected. Therefore, these two PSs should be excluded from this part of the analysis to avoid potential misinterpretation.
- Figure 6: Why are the significance levels of the differences between maximum radar reflectivity profiles of PSs under clean and dusty conditions not marked, as was done in the other profile figures?
- Figure 10: Numerous shape markers are used to represent different hydrometeors in the cloud, but these markers are not clearly labeled. A legend explaining this should be added.
Citation: https://doi.org/10.5194/egusphere-2025-2799-RC1 -
RC2: 'Comment on egusphere-2025-2799', Anonymous Referee #2, 09 Sep 2025
This manuscript investigatesthe impact of dust aerosols on the three-dimensional structure of precipitation systems of different sizes using a variety of observational data and reanalysis data. The authors employed a clustering method to group satellite precipitation radar profiles into organized precipitation systems, which is a novel and valuable approach.Studying precipitation from the system perspective provides deeper insights into aerosol–precipitation interactions and their coupling with environmental conditions.However, there remain a few minor issues that need further clarification and refinement, as outlined below.
Major Comments:
1.In section 2.1, since MODIS cannot detect aerosols below clouds, the authors used a spatiotemporal interpolationand extrapolation method to estimate the dustconcentration. It is noted that the spatial extent of the extrapolation varies with the size of PSs, and the size of PSs can reflect cloud coverageto some extent. However, this manuscript does not evaluate the impact of cloud coverage.I believe sensitivity experiments should be added.
2. In section 3.2,theauthors analyzed the influence of meteorological conditions on dust effects and found that CAPE plays a significant role. The author did not explain why CAPE emerges as a more prominent factor compared to otherdynamic andmoisture conditions(e.g., vertical wind shear, relative humidity).Such an explanation is critical for understanding of the complex interactions between precipitation, aerosols, and meteorology.
3.In section 4, although the radiative effect of dust can causewarming of the midtroposphere and cooling of the near surface, thereby suppressing convection, the influence of dust radiative effects cannot be ruled out. This manuscriptprovides a rather limited introduction to dust radiative effects, which requires further elaboration.
4. The PSs selected for this study primarily consist of deep convective cloud systems that exceed 6 km in height. This selection criterion may affect the statistical results. Therefore, this should be highlighted in the conclusion section, so readers are aware of its potential influence on the findings.
Minor Comments:
1.In line 38, 'with a temperature of between -5 and +2 ℃' should be corrected to 'with temperaturesbetween -5 and +2 ℃'.
2. In Table 5, the sample sizes of large-sized PSs across different CAPE bins appear too small, which raises concerns about the statistical reliability of the results.
3. In Figure 3, it would be more accurate to replace 'area' with 'near-surface precipitation area', since it is defined as the number of pixels with near-surface precipitation rates greater than 0 mm/h multiplied by the pixel area.
4. The expression 'meteorology conditions'should bereplacedwith 'meteorological conditions'throughout the manuscriptfor grammatical accuracy and consistency.
Citation: https://doi.org/10.5194/egusphere-2025-2799-RC2
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
1,708 | 44 | 18 | 1,770 | 29 | 19 | 17 |
- HTML: 1,708
- PDF: 44
- XML: 18
- Total: 1,770
- Supplement: 29
- BibTeX: 19
- EndNote: 17
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1
The authors utilized a variety of observational data and reanalysis data to study the impact of dust aerosols on the three-dimensional structure of precipitation systems of different sizes. Nevertheless, certain methodological and interpretive aspects warrant further elaboration and refinement.
Q1: As a spectral instrument, MODIS cannot directly observe aerosols beneath clouds. Although the authors employed a spatiotemporal interpolation method for aerosol matching, it is worth clarifying whether a cloud fraction threshold was applied during the interpolation process, particularly for PS regions with high cloud coverage.
Q2: The study categorizes PSs into small (<2000 km²), medium (2000-10000 km²), and large (>10000 km²) classes based on their horizontal area. Could the authors please specify if these area thresholds were defined with reference to the climatological characteristics of PSs commonly found in the tropical Atlantic ITCZ region?
Q3: The paper primarily focuses on the aerosol-cloud interaction process involving dust acting as ice nuclei. Could the authors elaborate on whether a more quantitative investigation was conducted regarding the associated water-phase processes? Furthermore, while the dust's radiative effect is not discussed in detail within the text, it is depicted in the Fig. 10. Could this aspect be explained more thoroughly?
Q4: Figure 5 shows a reduction in the 20 dBZ area below the freezing level for stratiform precipitation in small- and medium-sized PSs under dusty conditions, which the authors attribute to the evaporation effect associated with the Saharan Air Layer. Is there more direct evidence supporting this proposed mechanism? For instance, was a significant variety in low-level humidity co-observed?
Q5: In Table 5, the sample sizes across different CAPE bins are notably imbalanced (e.g., for large PSs in the CAPE5 bin: clean n=5, dusty n=14). The statistical reliability of results derived from such small sample sizes is a concern.