REVIEW of the article “Multi-thermals and high concentrations of secondary ice: A modelling study of convective clouds during the ICE-D campaign”

by Cui et al.

This study explores the role of various parameters such as freezing onset temperature of INPs, mineral dust as INP, multi-thermals on secondary ice production in convective clouds observed during the ICE-D campaign. They carried out multiple simulations to investigate these aspects. Based on model simulations they conclude that the higher concentration of INPs with warmer freezing onset temperature may affect the secondary ice formation in these clouds. They also found that multiple thermal plays an important role in determining the secondary ice production in tropical clouds which is an important finding.

Overall, the research work done in the paper will be a valuable contribution in understanding the role of INPs with warmer onset temperatures and multiple thermal on secondary ice processes. However, with the current version authors need to work on a few sections.

 I would recommend the manuscript be published after some specific points and minor corrections have been addressed.

Specific comments:                                    

1. To increase the readability of the manuscript, I suggest renaming the simulations and use some meaningful names. For example: ‘control’, ‘double-HM’, ‘early onset1’, ‘cooper10x’, ‘early onset1 & cooper10x’, ‘early onset1 & 100xINP’, ‘early onset2 & 100xINP’, ‘Demott’, ‘early Demott ‘, ‘Demott 10xINP, ‘multi-thermals’ etc. It's hard to follow the results and conclusions with current names.
2. One of the major objectives of the paper is to investigate whether multiple thermals in the clouds could explain the observed ice concentration. Therefore, some discussion about previous studies on the conditions favorable for secondary ice production is expected (e.g. graupel fall velocity, updraft speed, broader drop size distribution, etc). (Reference:  Cloud Conditions Favoring Secondary Ice Particle Production in Tropical Maritime Convection, Andrew Heymsfield1, and Paul Willis). What are the limitations of the current understanding of these processes? How your study is going to explore these uncertainties.
3. The default parameterization for primary ice nucleation used in the Morrison scheme is based on Cooper, 1986. The authors need to discuss the ice nucleation modes, supersaturation ranges w.r.t. water and ice for this scheme. It should be clear that which ice nucleation modes are active with the primary ice nucleation. Whether you have changed the onset temperature for only immersion-freezing mode?
4. It will be good to have a separate section for the ICE-D Observation (Section 2). The text after line number 91 to 106 from the introduction and current section 2 can be merged in the new section. Include the brief discussion on quality control of ice number concentration from 2DC-data. What is the size range of ice particles from observations mentioned in the paper? Whether the particles smaller than 200 µm were considered? Show that the conditions were favorable in the observed clouds for secondary ice formation. e.g. presence of drop larger than 24 µm size, presence of graupel particles in HM zone, etc in the same section. Figures 12 and 13 and relevant results can be added to this section. It should be clear that the comparison between model-simulated ice particles and observations is for particles of a similar size range.
5. Line 88: Authors mentioned previous work by Blyth and Latham (1997) on the role of multi-thermal in HM process. Mention in which aspects your simulations are different than theirs. What are the novel approaches of your study? Does your study agree/disagree with their simulations? How realistic are the simulated thermal in the model? The authors need to add some discussion on this.
6. In the conclusions, the authors mentioned that the multiple thermal still cannot reproduce the observed highest concentration of ice particles. There should be more discussion about it. Other secondary ice mechanisms may not be fully absent at temperatures warmer than -10^o Observations indicated the presence of fragments of frozen drops. The parametrization of secondary ice particles from frozen drops by Phillips et al. 2017 indicates that at those warmer temperatures this process might have a considerable contribution to the secondary ice formation.

Minor/technical comments:

Line 76: Cite recent review articles on secondary ice formation e.g. Review of experimental studies of secondary ice production: A Korolev, T Leisner Atmospheric Chemistry and Physics 20 (20), 11767-11797; A new look at the environmental conditions favorable to secondary ice production: A Korolev, I Heckman, M Wolde, AS Ackerman, AM Fridlind, LA Ladino, Atmospheric Chemistry and Physics 20 (3), 1391-1429.

Line 91:  Define ICE-D in the text also even if it is defined in the abstract.

Line 94: Rephrase/remove the sentence ‘However….

Line 101: Is FAAM defined earlier?

Line 119: Full stop is missing after … and aerosol

Line 134:  Correct the figure number. It is Figure 1a. Also add the axes titles (Lat, Long) on the plot. Correct the figure numbers in the rest part of the paragraph e.g. line number 135, 138, 141.

Figure 3:  Check the legends for 2DC data. Add the axes title in X-axes (Ice particle number concentration). Mention the size range of ice particles considered here.

Figure 4: Add axes titles.

Line 150: Mention the lowest diameter of ice particles considered in estimating their concentrations. Are you considering particles smaller than 200 um from the observations?

Line 151: Change -6,8 ^oC to -6.8 ^oC.

Line 163: Add reference for Hallet Mossop process

Line 209:  change 14C to -14⁰C

Line 213:  Change the graupe to the graupel

Line 221: Check an empty bracket after velocity

Line 254: Do you mean ‘starting freezing temperature’.

Line 255: How do you know the ice particles observed during aircraft observations are only originating from secondary ice processes.

Section 4.3: Be specific about simulations involving changes in freezing efficiency. Mention clearly what are the changes made for this.

Line 266: Correct ‘TLXTEN’  

Line 394: I think the Morrison microphysics scheme was applied for all the runs. It was modified for other sensitivity tests. Make this point clear in the text.

Line 420: Add some of the challenges in measuring full spectra of INP and CCN.

Figure 2: Check the X-axes title. What is HI? Is it defined in the text? Also, mention the Temperature on the color bar in the box. Also, change the title of X axes to Altitude (above MSL??) (m). Check its position w.r.t plot.

Figure 7: Check the spelling of ‘ratio’ in the figure title.

Table 1: RLX3X100 is the experiment where you relax the onset temperature in the ice nuclei number concentration and not the ice concentration. Similar to RLX2X100.

The authors discussed observations of ice crystal concentrations measured in-situ during the ICE-D field campaign in Cape Verde during August 2015. It was suggested that, based on INP measurements, primary nucleation alone could not explain the high ice crystal concentrations and so secondary ice processes, e.g. rime splintering, occurred in the convective cloud.

Simulations with one or multiple thermals were performed with several configurations of the heterogeneous ice nucleation scheme. The maximum ice crystal concentrations were than compared with the ones measured during aircraft passes through the cloud.

Major Comments

The vertical and horizontal extent as well as the lifetime of the modeled cloud are not compared to the observed cloud, for example by utilizing weather radar data. The microphysics of the model are also very sensitive to the initial conditions, especially the vertical profile of water vapor. The authors used initial profiles combined from dropsonde measurements and NCEP/NCAR reanalysis data, but did not show these profile or discuss which parts of the simulated clouds are located in the region approximated with reanalysis data. The authors should put more emphasis on linking observations with model results and discussing uncertainties of initial conditions.

A major part of the work focuses on evaluating the contribution of secondary ice from rime splintering. The authors identified secondary ice by observing the temperature zone where the Hallett-Mossop (HM) processes is active, e.g. between -3 and -8 °C., and attributed ice crystals in that temperature range to rime splintering. However, transport processes, e.g. advection and sedimentation of ice crystals from the colder part of the cloud into the HM zone, are not discussed.

Furthermore a majority of the sensitivity experiments increased the onset freezing temperature from -8 °C to – 3 °C or higher, hence heterogeneous nucleation is now possible inside the HM zone. This makes it very difficult to distinguish between ice from rime splintering and primary nucleation in that temperature range. The authors should build a stronger argument that increased ice crystal concentrations in the HM zone did indeed originate from enhanced rime splintering. For example sensitivity simulations without rime splintering could be performed and included in the comparisons.

Minor & technical comments

Line 63: clarify that homogeneous formation refers to homogeneous freezing of cloud/rain or solution droplets. Whereby the latter is often just referred to as homogeneous nucleation.

Line 64: with temperature

Line 76: there are also newer review papers relevant for secondary ice processes e.g. Korolvel et al. (2020)

Line 93: missing full stop after Field et Al

Line 119: missing full stop after aerosol

Line 134-143: there are a few mismatches in the figure references

Line 151-153: Fig. 3 does not provide information about temperature. The statement that ice appeared only at T < -10 °C is supported with Fig. 2.

Line 153-154: this section has to be expanded to support the statement that measured ice concentrations were lower than what primary nucleation would suggest since Fig. 2 and 3 do not show INP concentrations

Line 164: other relevant microphysical processes, which do not require interaction of cloud particles, like sedimentation and particle growth by deposition of water vapor should be mentioned. Also the Hallett-Mossop temperature zone is mentioned multiple times in this work and the temperature range should be explicitly stated in this section

Line 191-193: include information about model time steps and intervals of writing output. The latter is important for how clearly you can identify nucleation events in your data.

Line 209: -14 °C instead of -14C

Line 213: misspelled graupel

Line 216: check grammar of last sentence in this line

Line 221: parenthesis without content

Line 266: this statement should be better connected to the figure, that actually shows the increase in ice crystal concentrations

Line 288: misspelled increase

Line 332: remove ‘also’ before ‘approximately’

Line 334: remove space between m and s^-1

Line 340: check grammar in second part of the sentence

Line 367: panel d in Fig. 12 does not exist

Line 367-377: there are no ice particles shown in Fig. 13

Fig 4.: labels of isotherms are barely visible, but important for interpreting the plot. Purple is misspelled in the plot label.

Fig. 5: color bar should be changed so that the colors below the dark blue are more continuous. Also a logarithmic plot would be good since ice crystal concentrations below 1 /L are hidden, but important in how ice in the top of the cloud and the HM zone connect.

Fig. 7: RLX10 is called RLXTEN in the text

Fig. 8: the color bar is very hard to read. A continuous color bar should be used instead. Also the plot label misses the information that maximum ice crystals concentrations are shown. RLX10 is called RLXTEN in the text

Fig. 9: color bar should be changed so that the colors below the dark blue are more continuous.

References

Korolev, A. and Leisner, T.: Review of experimental studies of secondary ice production, Atmos. Chem. Phys., 20, 11767–11797, https://doi.org/10.5194/acp-20-11767-2020, 2020.