| 12 May 2026
Mineral dust concentration controlling convective cirrus structure and persistence: a large-eddy simulation study with observational constraints
Abstract. Dusty cirrus clouds are optically thick, convectively organized ice clouds that occur during intense mineral dust outbreaks. Previous modeling studies have found a link between mineral dust and convectively generated cirrus, but have been unable to explicitly resolve the associated convective dynamics due to limited spatial resolution. Here, we investigate the evolution and persistence of dusty cirrus using large-eddy simulations constrained by in-situ aircraft observations from the ML-CIRRUS 2014 campaign and complementary remote sensing data.
The simulations indicate that sustained convective cirrus requires mineral dust concentrations exceeding climatological values by approximately one to two orders of magnitude, corresponding to number concentrations on the order of Nd ∼ 1 cm-3. Under these conditions, heterogeneous freezing on mineral dust dominates ice formation and maintains sufficiently high ice crystal number concentrations to sustain strong longwave cooling at cloud top and preserve convective overturning under shortwave radiative warming. In contrast, homogeneous freezing remains largely suppressed across a wide range of simulated conditions. The direct radiative effect of mineral dust is comparatively weak relative to cloud radiative feedbacks and does not significantly influence cirrus evolution.
The simulated cirrus is also highly sensitive to the choice of deposition ice nucleation parameterization, highlighting a major source of uncertainty in representing dust–ice interactions. Overall, the results identify mineral dust availability as the primary control on dusty cirrus persistence and emphasize the need for improved representation of heterogeneous ice nucleation in atmospheric models.
Competing interests: At least one of the (co-)authors is a member of the editorial board of Atmospheric Chemistry and Physics.
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.
General Comments
This study investigates the influence of mineral dust concentrations on cirrus formation through a series of sensitivity simulations for a single case study, evaluated against multiple observational datasets. In doing so, it establishes a link between cloud microphysics, radiation, and convective dynamics.
While this is an ambitious study, the current version is at times difficult to follow and occasionally lacks some essential information.
Uncertainty and comparability of observational and model datasets: This study draws on multiple observational datasets including satellite retrievals (IWC, ICNC, cloud top temperature), in-situ measurements, and cloud radar data alongside model simulations. Each of these carries its own uncertainties and underlying assumptions, yet these are only in a few cases elaborated upon, leaving it unclear to what degree agreement between the different data sources can reasonably be expected. I suggest that a discussion of these uncertainties and limitations be included either in the methods section or the conclusions.
Role of sedimentation: Heterogeneously formed ice crystals in particular are associated with high sedimentation fluxes. This study does not sufficiently discuss whether ice crystals observed at a given altitude were actually formed there, nor the potential role of sedimentation in the vertical redistribution of ice. While the averaging approach and statistical presentation adopted here may largely mitigate the influence of sedimentation on the interpretation of the results, this should be discussed more explicitly.
Readability of the manuscript: The manuscript addresses the complex interplay between cloud microphysics, radiative cooling and heating, and convective dynamics, drawing on data from sensitivity simulations and several distinct observational datasets. For readers who are not experts across all of these areas, the manuscript can be difficult to follow. I therefore recommend investing some effort in restructuring parts of the manuscript, elaborating on key concepts where necessary, and clarifying the connections between them. For example, by more clearly delineating subsections and revisiting the ordering of the large amount of figures. My specific comments below may offer some guidance in this regard.
A list of abbreviations in the appendix would also be helpful.
Specific comments
Line 58-59: Consider providing a reference or including the particle growth equation in the appendix.
Line 63-64: The setup of aerosols, cloud droplets and ice classes in the bin scheme is still unclear to me. Consider expanding the description.
Line 68-70: Is the Koop et al. (2000) rate used for droplet of all sizes and thermodynamic regimes as the specific mention of cloud and raindrops would suggest? The Koop et al. (2000) rate should only be used for nanometer-sized aqueous solution droplets at temperatures below the homogeneous freezing regime of pure water (T < 235K). For micrometer-sized ‘pure’ water droplets the nucleation rate from Koop and Murray (2016) is appropriate.
Line 79-80: Is turbulent mixing of aerosols considered in the model?
Line 86-89: In typical formulations of the vapor deposition growth equation, the capacitance parameter has the dimensions of a length. It is unclear why this is treated as a constant rather than as a function of ice crystal size or mass. As noted above, referencing the growth equation explicitly would help clarify this choice.
Line 90: Provide further details of the assumptions underlying the radiation scheme. For example, does it use bulk quantities such as mean particle size, or is it size-resolved? Is the assumed ice particle shape consistent with those used for growth by vapor deposition (spherical and rosettes)?
Section 2.2.1: Are shattering artifacts, arising from ice crystals breaking up upon entering the cloud probe, a source of uncertainty for ICNC in this dataset?
Section 2.5: The setup and general design of the UCLALES simulation would benefit from further clarification. In particular: is the model best understood as an air parcel simulation, but discretized over a three-dimensional domain rather than the zero-dimensional framework of a conventional box model? Is the initial thermodynamic state horizontally homogeneous? Given that the domain already extends from the surface to 14 km, is the domain itself lifted by the vertical wind profile, is adiabatic forcing applied as in a typical box model simulation? Please also specify the boundary conditions. Addressing these points should make the section more accessible to readers unfamiliar with this modeling approach.
Line 189: Which threshold is being referenced? Koop et al. (2000) does not provide a (singular) ice supersaturation threshold for nucleation.
Line 231-232: Briefly summarize the mechanisms responsible for the mixing described in the referenced theory. If the mixing arises from cirrus cloud processes themselves, it is worth considering whether the BASE-MIXED profile is an appropriate initial condition for simulations prior to cirrus formation.
Figure 6: Use different markers for CONV and SEDI layer thresholds.
Line 256-257: Specify the exact region (in time, and horizontal and vertical extent) to which this statement refers. It is not apparent from the figure where the enhanced and reduced IWC values are located — between 9 and 11 km, IWC appears to be close to 10 mg m−3 throughout, with no clear correspondence to the regions of elevated updraft or downdraft.
Figure 7: Clarify how the plotted heating rates are calculated. Are they horizontal averages? Additionally, explain the distinction between the shaded regions of high and low transparency.
Line 271: The reference of Fig. 7 seems incorrect as this figure does not show vertical velocities. Is the reference supposed to be Fig. 6?
Line 274: Is the reference supposed to be Fig. 6? If not, indicate the CONV layer in Fig. 7 as well.
Figure 8: Are these horizontally averaged vertical velocities?
Figure 9: Clarify what the panel labels (e-h) indicate in the sub-caption since the first time this figure is referenced in Section 3.1 this is not clear. Where in the manuscript are panels (c-h) discussed? Consider moving panels (c-h) to the supplement and including the MODIS image from Fig. 1d here for an easier comparison.
Figure 10: The lines representing DARDAR, MIRA, OPC, Model Base, and NIXE should be assigned unique style markers in addition to color, ensuring the figure remains legible both in greyscale and for readers with color vision deficiencies. Consult the ACP figure submission guidelines in this regard. Clarify whether the lines represent the mean, median or singular values. Which of those lines represent spatially and / or temporally averaged properties? Consider adding the CONV and SEDI lines here as well.
Figure 11: Assign unique style markers to the lines as hey are hard to read with green-red and more severe color vision deficiencies. Add the choice of ND to the panel labels.
Figure 12: Clarify whether the lines mean, median or singular values and if they are averaged. As mentioned for other figures, consider assigning the lines unique style markers for readability in greyscale.
Lines 368–374: Could sufficiently high supersaturation for homogeneous freezing be achieved between model output steps, above the observed cloud top (with subsequent sedimentation of the homogeneously formed ice crystals), or as a result of spatial inhomogeneities in the RHi field?
Line 378: Clarify or remind the reader on why these range of ND was chosen.
Line 395: Does the model explicitly distinguish between ice formed by heterogeneous and homogeneous nucleation, or is this inferred from the simulation output?
Section 3.2.1: The first three paragraphs of this section are difficult to follow, as they move rapidly between figures and concepts. The figures in particular have not been sufficiently introduced. Consider restructuring this section to improve clarity and readability.
Figure 13: Again it is unclear whether these values are averaged. Include the ND values in the panel labels.
Figure 14: Clarify if the statistics represent the mean model output and how they are averaged.
Line 409-410: The phrasing of this sentence is confusing.
Line 404-410: Is the equivalent potential temperature gradient sufficient to explain the occurrence of new convective cells or are they other dynamical properties that should be considered here?
Line 416-419: What is the output interval and microphysical timestep of the simulations? Is there are chance that higher RHi values are reached but not recorded as the newly nucleated ice crystals depletes RHi before the next output step?
Section 3.2.2: This section covers several distinct topics such as sensitivity analysis, the influence of the diurnal cycle, and the representativeness of the case study for regional dusty cirrus. This restuls in rapid transitions between figures and concepts that make the section difficult to follow. Consider splitting this into clearly delineated subsections.
Line 467-468: Which property is described as stabilizing under SW daytime heating? RHi and ICNC in Figure 16 appear stable in the nighttime scenario and unstable during the day, which seems contrary to the statement made here. Or is this referencing the vertical distribution of net. rad. heating/cooling that promotes some dynamical instability? In that case also elaborate.
Line 469-470: Where exactly does the net radiative cooling intensify between 14 and 16 hours? I do not see that in Figure 16.
Line 475: Figure 18 is referenced before Figure 17. Consider reordering them.
Figure 19: Again, consider assigning the lines unique markers.
Technical corrections
Equation A2: Explanation of the coefficients and parameters is missing, e.g., wLS.
Figure B1: The second panel has the wrong label.