the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 18 Sep 2025
Can rime splintering explain the ice production in Arctic mixed-phase clouds?
Abstract. Secondary ice production (SIP) can increase ice crystal number concentration (ICNC) by several orders of magnitude, particularly in clean clouds with low concentrations of ice-nucleating particles (INPs). The most common SIP process in models is rime splintering (RS) also called as the Hallett-Mossop process. The generally adopted RS-formulation gives 350 splinters per milligram of rimed ice at the temperature of 268 K. We used large-eddy simulations to examine if rime splintering could explain the high ICNC observed during the ACLOUD (Arctic CLoud Observations Using airborne measurements during polar Day) campaign where cloud temperatures close to 268 K are favourable for rime splintering. With the default model setup, the splinter production rate had to be multiplied by a factor ten to close the gap between modelled and observed ICNCs. Similar changes have been made in other modelling studies. The factor of ten multiplier helped to trigger SIP so that it became a self-sustaining process, fully independent of the primary freezing initiated by INPs. Our simulations reached realistic steady-state ICNCs and maintained stable mixed-phase clouds through the 24-hour simulation time. Additional sensitivity tests showed that the efficiency of SIP depends strongly on model parametrizations and air temperature, so that simulations with a modified setup were able to reproduce the observed ICNCs without the factor of ten multiplier.
-
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
(1487 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(1487 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-4470', Anonymous Referee #1, 22 Oct 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-4470/egusphere-2025-4470-RC1-supplement.pdfCitation: https://doi.org/
10.5194/egusphere-2025-4470-RC1 -
RC2: 'Comment on egusphere-2025-4470', Anonymous Referee #2, 13 Nov 2025
This paper addresses the origin of the high ice particle concentrations observed in an Arctic mixed-phase cloud during the ACLOUD campaign, which is a topic that still raises several open questions. The main results show that the rime splintering (RS) mechanism alone cannot explain the observed ice crystal number concentrations (ICNC). The authors demonstrate that either increasing the RS efficiency by a factor of ten or modifying the model setup and parameterizations of microphysics can lead to more realistic ICNC values.
Overall, this is a good quality study. The combined use of both bin and bulk microphysics approaches is quite rare and strengthen the conclusions of the study. It is also very interesting to see that modifying the parameterizations of microphysics have a such large effect on SIP efficiency.
I see two main issues with the present study. First I don't understand why other SIP mechanisms are not considered, given that were studied in one previous study: Calderón et al. (2025) (https://doi.org/10.5194/egusphere-2025-2730). Even if the temperature range studied here is not optimal for a high efficiency of collisional breakup or fragmentation of freezing drops (more active around –15 °C), these processes could still influence the ICNC. Secondly, while the sensitivity experiment are valuable, their design and presentation could be improved for greater clarity.
Despite these points, the study is of high quality and offers valuable insights into SIP. I recommend that the authors address the following points to further strengthen the manuscript before publication.
Major comments:
Although significant uncertainties remain regarding SIP, artificially increasing the rime splintering (RS) efficiency in the model seems a bit artificial, as the literature does not support such strong RS. This point should be more clearly emphasized in the study.
Collisional breakup the breakup process has been shown to be effective in Arctic mixed-phase clouds (e.g., Karalis et al., 2022: https://doi.org/10.1016/j.atmosres.2022.106302 and Sotiropoulou et al., 2021: https://doi.org/10.5194/acp-21-9741-2021). Therefore, it would also be valuable to test whether collisional breakup or DS could explain the observed ICNC without artificially boosting the RS rate. If I am correct, these SIP mechanisms can be activated in SALSA, as done in Calderón et al. (2025) ?
As ice crystal concentration is typically expressed in L⁻¹, I recommend using this unit instead of kg⁻¹ for consistency with the INP units and the literature.
Lines 69-70: Although no large drops were observed, it would be useful to check whether the model generates any. The presence of such drops could potentially trigger DS.
Lines 73-79: The presence of millimeter sized ice particles suggest that collisional breakup might occur.
Lines 137-141: It is is good point to highlight and discuss that.
Section 2.2.2: The HM process in SALSA and the number of bins used are not described, unlike for the SB scheme. Adding this information would improve clarity for readers.
Lines 238-239: This statement seems too strong. I would suggest saying that RS is insignificant, rather than SIP in general, since other SIP mechanisms were not considered.
Figure 2: In the factor 10 multiplier experiment, the LWP decreases to approximately 40 g m⁻², which is a bit lower than the observed values of 48–82 g m⁻² mentioned in Line 65. The factor 5 experiment appears to be more consistent with the observed range.
Lines 255-257: The Sotiropoulou et al. (2020) study also mention that “Only the combination of both rime splintering (RS) and collisional break-up (BR) can explain the observed ICNCs, since both of these mechanisms are weak when activated alone.” This suggests that combining RS with other processes could give realistic ICNCs, rather than artificially increasing RS.
This suggests that combining RS with other processes could give realistic ICNCs, rather than artificially increasing RS.
Lines 267-269: It is good to highlight the computational cost of the two microphysics schemes.
Lines 271-274: Very interesting to see that after a while INPs are not needed
Section 3.3: for both Fig 5 and 6:
- The comparison between the 100 kg⁻¹ RS×10 line and the 1000 kg⁻¹ RS×0 line makes it difficult to determine whether the variations are due to turning off RS or changing the INP number. Adding an intermediate 100 kg⁻¹ RS×0 line would allow isolating the effect of RS deactivation without altering the INP number.
- Suggestion: showing only one time per profile could make the figure more readable, as these lines are similar and not central to the discussion.
Section 3.4: Examining the effect of the model parameters on SIP production is very interesting. Including a table in the appendix summarizing all the tests would help readers, or at least more clearly mentioning the experiment names in the text.
Fig 7: Presenting sensitivity results first with a fixed RS multiplier would help, as varying both RS and model parameters complicates comparisons. I suggest separating the experiments into (1) varying model parameters with RS fixed, and (2) varying the RS multiplier to clarify their effects.
Lines 340-342: Yes, this is an excellent point to highlight.
Lines 343-349: This is very interesting. Why not test other temperature curves here as well?
Conclusion: It might be important to mention that the RS multiplier is somewhat artificial and does not directly reflect observations of this process and is primarily used to achieve the observed ice particle concentrations. This point is particularly relevant regarding the study of Seidel et al. (2024), which questions the RS efficiency.
Lines 393–394: Yes, other SIP processes may occur at colder temperatures, but it is important to keep in mind that collisional breakup can also happen between -3 and -8°C like RS or even between 0 and –3 °C, unlike RS.
Minor comments:
Lines 87-90: The first sentence indicates that DS may occur, whereas the second one suggests it does not. I understand the message but this is a bit is confusing and could be clarified.
Lines 196-200: could be moved to the previous section about SALSA microphysics.
Lines 383-386: The Lawson et al. (2015) parametrization has no temperature dependency, it is only parameterized as as function of drop size. But yes the DS process is fore sure dependent on temperature as showed for example recent experiments of Keinert et al. (2020). (https://doi.org/10.1175/JAS-D-20-0081.1)
Lines 390-391: In my opinion, it is more due to the fact that no parametrization was available for DS until the paper of Lawson et al. (2015) and that the RS parametrization was used and available since decades.
Citation: https://doi.org/10.5194/egusphere-2025-4470-RC2 - AC1: 'Replies to referees', Tomi Raatikainen, 16 Jan 2026
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-4470', Anonymous Referee #1, 22 Oct 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-4470/egusphere-2025-4470-RC1-supplement.pdf
-
RC2: 'Comment on egusphere-2025-4470', Anonymous Referee #2, 13 Nov 2025
This paper addresses the origin of the high ice particle concentrations observed in an Arctic mixed-phase cloud during the ACLOUD campaign, which is a topic that still raises several open questions. The main results show that the rime splintering (RS) mechanism alone cannot explain the observed ice crystal number concentrations (ICNC). The authors demonstrate that either increasing the RS efficiency by a factor of ten or modifying the model setup and parameterizations of microphysics can lead to more realistic ICNC values.
Overall, this is a good quality study. The combined use of both bin and bulk microphysics approaches is quite rare and strengthen the conclusions of the study. It is also very interesting to see that modifying the parameterizations of microphysics have a such large effect on SIP efficiency.
I see two main issues with the present study. First I don't understand why other SIP mechanisms are not considered, given that were studied in one previous study: Calderón et al. (2025) (https://doi.org/10.5194/egusphere-2025-2730). Even if the temperature range studied here is not optimal for a high efficiency of collisional breakup or fragmentation of freezing drops (more active around –15 °C), these processes could still influence the ICNC. Secondly, while the sensitivity experiment are valuable, their design and presentation could be improved for greater clarity.
Despite these points, the study is of high quality and offers valuable insights into SIP. I recommend that the authors address the following points to further strengthen the manuscript before publication.
Major comments:
Although significant uncertainties remain regarding SIP, artificially increasing the rime splintering (RS) efficiency in the model seems a bit artificial, as the literature does not support such strong RS. This point should be more clearly emphasized in the study.
Collisional breakup the breakup process has been shown to be effective in Arctic mixed-phase clouds (e.g., Karalis et al., 2022: https://doi.org/10.1016/j.atmosres.2022.106302 and Sotiropoulou et al., 2021: https://doi.org/10.5194/acp-21-9741-2021). Therefore, it would also be valuable to test whether collisional breakup or DS could explain the observed ICNC without artificially boosting the RS rate. If I am correct, these SIP mechanisms can be activated in SALSA, as done in Calderón et al. (2025) ?
As ice crystal concentration is typically expressed in L⁻¹, I recommend using this unit instead of kg⁻¹ for consistency with the INP units and the literature.
Lines 69-70: Although no large drops were observed, it would be useful to check whether the model generates any. The presence of such drops could potentially trigger DS.
Lines 73-79: The presence of millimeter sized ice particles suggest that collisional breakup might occur.
Lines 137-141: It is is good point to highlight and discuss that.
Section 2.2.2: The HM process in SALSA and the number of bins used are not described, unlike for the SB scheme. Adding this information would improve clarity for readers.
Lines 238-239: This statement seems too strong. I would suggest saying that RS is insignificant, rather than SIP in general, since other SIP mechanisms were not considered.
Figure 2: In the factor 10 multiplier experiment, the LWP decreases to approximately 40 g m⁻², which is a bit lower than the observed values of 48–82 g m⁻² mentioned in Line 65. The factor 5 experiment appears to be more consistent with the observed range.
Lines 255-257: The Sotiropoulou et al. (2020) study also mention that “Only the combination of both rime splintering (RS) and collisional break-up (BR) can explain the observed ICNCs, since both of these mechanisms are weak when activated alone.” This suggests that combining RS with other processes could give realistic ICNCs, rather than artificially increasing RS.
This suggests that combining RS with other processes could give realistic ICNCs, rather than artificially increasing RS.
Lines 267-269: It is good to highlight the computational cost of the two microphysics schemes.
Lines 271-274: Very interesting to see that after a while INPs are not needed
Section 3.3: for both Fig 5 and 6:
- The comparison between the 100 kg⁻¹ RS×10 line and the 1000 kg⁻¹ RS×0 line makes it difficult to determine whether the variations are due to turning off RS or changing the INP number. Adding an intermediate 100 kg⁻¹ RS×0 line would allow isolating the effect of RS deactivation without altering the INP number.
- Suggestion: showing only one time per profile could make the figure more readable, as these lines are similar and not central to the discussion.
Section 3.4: Examining the effect of the model parameters on SIP production is very interesting. Including a table in the appendix summarizing all the tests would help readers, or at least more clearly mentioning the experiment names in the text.
Fig 7: Presenting sensitivity results first with a fixed RS multiplier would help, as varying both RS and model parameters complicates comparisons. I suggest separating the experiments into (1) varying model parameters with RS fixed, and (2) varying the RS multiplier to clarify their effects.
Lines 340-342: Yes, this is an excellent point to highlight.
Lines 343-349: This is very interesting. Why not test other temperature curves here as well?
Conclusion: It might be important to mention that the RS multiplier is somewhat artificial and does not directly reflect observations of this process and is primarily used to achieve the observed ice particle concentrations. This point is particularly relevant regarding the study of Seidel et al. (2024), which questions the RS efficiency.
Lines 393–394: Yes, other SIP processes may occur at colder temperatures, but it is important to keep in mind that collisional breakup can also happen between -3 and -8°C like RS or even between 0 and –3 °C, unlike RS.
Minor comments:
Lines 87-90: The first sentence indicates that DS may occur, whereas the second one suggests it does not. I understand the message but this is a bit is confusing and could be clarified.
Lines 196-200: could be moved to the previous section about SALSA microphysics.
Lines 383-386: The Lawson et al. (2015) parametrization has no temperature dependency, it is only parameterized as as function of drop size. But yes the DS process is fore sure dependent on temperature as showed for example recent experiments of Keinert et al. (2020). (https://doi.org/10.1175/JAS-D-20-0081.1)
Lines 390-391: In my opinion, it is more due to the fact that no parametrization was available for DS until the paper of Lawson et al. (2015) and that the RS parametrization was used and available since decades.
Citation: https://doi.org/10.5194/egusphere-2025-4470-RC2 - AC1: 'Replies to referees', Tomi Raatikainen, 16 Jan 2026
Peer review completion
Journal article(s) based on this preprint
Data sets
Brief description of the simulations, source code of UCLALES-SALSA, and the simulation data used in this publication T. Raatikainen https://a3s.fi/12001823-acloud-pub/index.html
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 2,299 | 182 | 44 | 2,525 | 34 | 37 |
- HTML: 2,299
- PDF: 182
- XML: 44
- Total: 2,525
- BibTeX: 34
- EndNote: 37
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Silvia Calderón
Emma Järvinen
Marje Prank
Sami Romakkaniemi
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(1487 KB) - Metadata XML