Comment on <q>Review of Experimental Studies of Secondary Ice
Production</q> by Korolev and Leisner (2020)

Phillips, Vaughan T. J.; Yano, Jun-Ichi; Deshmukh, Akash; Waman, Deepak

doi:https://doi.org/10.5194/acp-2021-123

Preprints

https://doi.org/10.5194/acp-2021-123

Preprints

15 Feb 2021

Review status: this preprint is currently under review for the journal ACP.

Comment on Review of Experimental Studies of Secondary Ice Production by Korolev and Leisner (2020)

Vaughan T. J. Phillips¹, Jun-Ichi Yano², Akash Deshmukh¹, and Deepak Waman¹ Vaughan T. J. Phillips et al.

Show author details

¹Department of Physical Geography, University of Lund, Lund, Sweden
²CNRM, UMR 3589 (CNRS), Meteo-France, 31057 Toulouse Cedex, France

Received: 11 Feb 2021 – Accepted for review: 12 Feb 2021 – Discussion started: 15 Feb 2021

Abstract. This is a comment on the article Review of Experimental Studies of Secondary Ice Production by Korolev and Leisner (2020, hereafter termed KL2020), referring to the discussion about ice fragmentation. We argue that the only two studies characterising fragmentation in ice-ice collisions are not so erroneous as to prevent their use in representing this breakup in numerical models, contrary to the impression given in the review. A scaling analysis suggests that breakup of ice during sublimation can make a significant, albeit lesser, contribution to ice enhancement in clouds.

Vaughan T. J. Phillips et al.

Status: final response (author comments only)

RC1:
'Comment on acp-2021-123', Andrew Heymsfield, 21 Feb 2021

I'm openly sharing my identity. I have tried to address the authors' comment/concern of the Korolev and Leisner article which suggests that ice fragmentation during sublimation is not generally an important process for secondary ice production. To address that point, I drew on the Lagrangian spiral descents during the CRYSTAL-FACE field program, where the in-situ aircraft spiraled in a descending pattern, drifting with the wind and descending at a rate of about 2 m/sw. These were thunderstorm anvils, which began at a temperature of about -15C and ended at temperatures from about 6 to 8C. The relative humidity in the region at temperatures above 0C were significantly below 100% and the ice particles were sublimating, not melting. I have 3 figures for the 3 spirals, showing temperature, relative humidity and total ice concentration. I see no evidence of a concentration increase that would suggest ice fragmentation in the sublimating region is significant. Furthermore, in such regions, sublimating ice particle fragments would each rapidly fully sublimate or be collected by other particles. Andy Heymsfield

Citation: https://doi.org/10.5194/acp-2021-123-RC1
- AC1:
  'Reply on RC1', Vaughan Phillips, 21 Feb 2021
  
  We thank the reviewer for the interesting comment.
  It is unclear if the relative humidity shown is with respect to water or ice. We will assume it is with respect to water.
  The laboratory studies about sublimational breakup that we referred to in our comment show that such breakup only occurs at relative humidities with respect to ice of less than about 70-80%. Following the melting scheme by Phillips et al. (2007, JAS), the onset of melting occurs when the surface temperature of the ice during sublimation reaches 0 degC. We predict that this occurs when the ambient air temperature exceeds about 1.5 degC (codes available from us on request), for the three CRYSTAL-FACE aircraft descents shown by the reviewer. At that ambient temperature for onset of melting (1.6 degC), the relative humidity with respect to ice is 80%.
  For the three CRYSTAL-FACE descents shown by the reviewer, sublimation below the melting level can only occur between 0 and 1.6 degC, an extremely narrow range of heights where a uniformly high relative humidity with respect to ice was observed (> 80%, we infer).
  So, the reviewer's data does not prove the absence of breakup during sublimation generally, since no-one has ever claimed that at such high relative humidities there would be any breakup. In the comment, we actually base our breakup estimate on a scenario with a relative humidity of about 70% or less.
  Equally, CRYSTAL-FACE happened long before the shattering bias was discovered for optical probes. Thus, the probe-tips had not been invented then. So the unfiltered ice concentrations (> 30 microns) shown by the reviewer are likely dominated by artificial fragments from impact of snow on the aircraft probe (e.g. Korolev et al. 2011).
  
  Citation: https://doi.org/10.5194/acp-2021-123-AC1
  - AC2: 'Reply on AC1', Vaughan Phillips, 22 Feb 2021
    
    Additionally, everyone agrees that the fragments will disappear during sublimational descent. The point we are making in the comment is that it takes long enough for any given fragment to disappear (about a minute perhaps) that there is a quasi-equilibrium concentration that is enhanced due to the balance between elimination and continual emission in the descending parcel.
    (BTW, I meant to write "below the freezing level" instead of "below the melting level" in the previous response).
    
    Citation: https://doi.org/10.5194/acp-2021-123-AC2
    
    CC1: 'Reply on AC2', Jacob Carlin, 24 Feb 2021
    
    In support of Vaughan’s hypothesis, I wanted to chime in with some evidence we have recently come across of potential sublimational SIP. Specific differential phase (K_DP) has recently been proposed as a potential marker of SIP zones as it is sensitive to large concentrations of anisotropic fragments, but this has so far been limited to SIP during riming (e.g., Grazioli et al. 2015; Sinclair et al. 2016; Kumjian and Lombardo 2017). However, looking at quasi-vertical profiles (QVPs) we have recently found bands of enhanced K_DP situated squarely within the sublimation zone in 9 of 12 cases of strong sublimation, coincident with gradients of Z and RH w.r.t. ice (taken from coincident RAP analyses). We are currently following up and exploring these signatures more quantitatively. However, given the observational laboratory evidence for sublimational SIP (e.g., Oraltay and Hallet 1989; Dong et al. 1994; Bacon et al. 1998), I contextualized these signatures with the existing literature by emulating Fig. 13 of Oraltay and Hallett (1989; reproduced below) and plotting the median K_DP as a function of RHi and ice bulb temperature. Note that Oraltay and Hallett (1989) only included down to 50% RHi and -6C for ice bulb temperatures, so my figure is an extrapolation and it is unclear whether their 70% RHi threshold is maintained at much colder ice bulb temperatures. Regardless, it is apparent that these regions of enhanced K_DP indicative of high concentrations of highly anisotropic particles are concentrated primarily in the region identified by Oraltay and Hallett (1989) as the favorable zone for sublimational breakup. The plot using mean K_DP was similar. These sorts of values are not likely to be due to aggregates which we believe characterize the majority of parent particles in the cases examined, and riming is precluded due to the degree of dry air present. Thus, although the use of K_DP is indirect evidence, under proper conditions (e.g., a continual and sufficiently high flux of parent particles into a sufficiently dry layer and favorable particle habits), I believe Vaughan’s hypothesis of an equilibrium being established between fragment generation and fragment consumption to be plausible.
    Jacob Carlin
    
    Citation: https://doi.org/10.5194/acp-2021-123-CC1
CC2:
'Comment on acp-2021-123', Thomas Leisner, 02 Mar 2021
Here a few important points from the perspective of the authors of the review article:
We do not criticize, and certainly not „negatively criticize“ the work of Takahashi and Vardiman. These pioneering studies are up to now all laboratory experiments we have to rely on regarding ice-ice fragmentation

We criticize, or rather feel sorry, that this work was not taken up and extended to a broader range of atmospherically relevant situations. This follows the general tone of our review article which calls for a new surge in laboratory experiments on secondary ice formation.

We indeed have some objections on how the original experiments were carried out and where they do not reflect what is the predominant situation in clouds where ice-ice collisional fragmentation occurs. This was not the topic of our original article and Phillips et al. cite it here as “personal communication”, without giving any detail. They then refute arguments that are (at least partially, especially when rotations an angular momentum is mentioned) misinterpretations of our communication to them. We strongly feel that the discussion here should stay focused on the content of our review article. Depending on the final content of the comment we will prepare a reply to clarify this.

What we wanted to touch in the review, and perhaps not stated clearly enough is that we feel that the wide distribution of atmospheric ice with respect to shape, crystallinity, degree of riming etc. has to be assessed in much more detail. We do not believe, that one single parameter can account for all these factors, but this is beyond the scope of our review and could be worked out by us further if necessary.
Citation: https://doi.org/10.5194/acp-2021-123-CC2
- AC3: 'Reply on CC2', Vaughan Phillips, 22 Mar 2021
  
  We thank the authors of the review paper for responding to our commentary.
  I am glad to see that we all agree that the two laboratory studies by Vardiman and Takahashi et al. were important for initiating research into this mechanism of breakup in ice-ice collisions.
  However, regarding the first point from Thomas Leisner, our commentary was about what the authors actually published in their review paper last year (Korolev and Leisner 2020 [KL2020]) and its intellectual basis. Whether the authors now continue to criticise both laboratory studies by Vardiman and Takahashi et al. is not the salient issue.
  Perhaps we can re-visit the text of the corrected review paper by KL2020:
  "A detailed analysis of the Takahashi et al. (1995) laboratory setup indicated that the riming of ice spheres occurred in still air, which resulted in more lumpy and fragile rime compared to that formed in free-falling graupel.The collisional kinetic energy and the surface area of collision of the 2 cm diameter ice spheres also significantly exceed the kinetic energy and collision area of graupel whose typical size is a few millimeters. Altogether, it may result in overestimation of the rate of SIP, compared to graupel formed in natural clouds. …No parameterizations of SIP due to ice–ice collisional fragmentation can be developed at that stage based on two laboratory observations [both studies by Takahashi et al. and Vardiman], whose results are conflicting with each other" (sentence B).
  This text was clearly a negative criticism of the Takahashi and Vardiman lab studies, because both studies aimed to provide a basis for predicting observed concentrations of ice seen in natural clouds. Indeed, Takahashi et al. (1995) wrote in the abstract: "This ice particle production mechanism ... may help to explain the high concentrations of ice crystals observed m mantlme winter cumuli". They go onto make a quantitative prediction of ice concentrations in-cloud resulting from the ice multiplication based on their observations and claim agreement with aircraft observations. Similarly, Vardiman (1978) used his lab/field data of fragmentation as the basis for predicting in-cloud concentrations of ice too, creating a theory with a 0D analytical model.
  The final sentence B of the quote from KL2020 effectively says that no such quantitative predictions can be made from their laboratory observations. These quantitative predictions made in both lab studies constitute "parameterizations" by the experimentalists themselves.
  Regarding the third point, that same sentence B is such a strong statement about the lack of usefulness of both lab studies for representations of natural clouds that it was only reasonable to contact the authors of the review for clarification. When we did so, they elaborated their criticisms with more detail, as related in our commentary.
  Of course, the personal communication citation can be withdrawn, if the KL2020 authors do not agree to our interpretation. But any commentary such as ours need not be restricted only to material of the KL2020 review.
  When such a contentious statement is made, denying the accuracy and usefulness of the only two lab studies for breakup in ice-ice collisions, it is valid to ask what other possible criticisms could be made to justify it, beyond what is merely in the review.
  
  
  
  Citation: https://doi.org/10.5194/acp-2021-123-AC3
CC3:
'Comment on acp-2021-123', Alexei Korolev, 12 Mar 2021

The comment in Phillips et al. (section 3.2, 2021) regarding the SIP mechanism due to fragmentation of sublimating ice particles is redundant. The comment attempts to explain the process of enhanced growth of secondary ice forming from ice sublimational breakup in undersaturated downdrafts with subsequent entrainment in a supersaturated updraft. This essentially repeats the statement from the Korolev and Leisner (2020) review paper (Section 9.2): “Activation of SIP due to the fragmentation of sublimating ice requires spatial proximity of undersaturated and supersaturated cloud regions. In this case, secondary ice particles formed in the undersaturated cloud regions can be rapidly transported into the supersaturated regions prior to their sublimation.”

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.
Phillips, T.J., Yano, J.-I., Deshmukh, A., and Waman, D.: Comment on “Review of Experimental Studies of Secondary Ice Production” by Korolev and Leisner (2020), Atmos. Chem. Phys. Discussion, https://doi.org/10.5194/acp-2021-123, 2021

Citation: https://doi.org/10.5194/acp-2021-123-CC3
- CC4: 'Reply on CC3', Akash Deshmukh, 20 Apr 2021
  
  I thank Korolev for this comment.
  We disagree that our remarks about sublimation breakup were redundant in the commentary (Phillips et al. 2021, Sec. 3.2).
  Originally, the original review (Korolev and Leisner, 2020 ["KL2020"]) concluded that:
  "Since small ice fragments have lower terminal fall velocity, their residence time in the undersaturated environment may be long enough to result in their complete evaporation before they can re-enter a supersaturated environment. This appears to be a significant limitation of the SIP mechanism due to sublimation breakup. This mechanism is also unlikely to explain explosive concentrations of small ice crystals frequently observed in convective and stratiform frontal clouds (e.g., Lawson et al., 2017; Korolev et al., 2020) ... Activation of SIP due to the fragmentation of sublimating ice requires spatial proximity of undersaturated and supersaturated cloud regions. In this case, secondary ice particles formed in the undersaturated cloud regions can be rapidly transported into the supersaturated regions prior their sublimation. Such conditions may occur in cloud regions affected by entrainment and mixing with out-of-cloud dry air."
  Firstly, the mechanism stated here by KL2020 is about entrainment mixing between environment and cloud edges. There is no mention of convective downdrafts. In fact, KL2020 does not even mention descent anywhere as a cause of the sublimation breakup. By contrast, our focus was chiefly on convective downdrafts (Phillips et al. 2021, Sec. 3.2) and we provided the scale analysis to estimate the ice enhancement in such descent. Of course, we mentioned entrainment mixing as a possible cause of breakup but our emphasis was on deep convective descent.
  Secondly, the concept depicted by KL2020 involves fragments being emitted at a particular instant of time and disappearing at some later time by total sublimation with the possibility of transfer into an adjacent supersaturated region. Whether that mixing happens too early or too late determines if the fragments are actually transferred.
  Again, our picture is different. In Sec. 3.2 of our commentary (Phillips et al. 2021), we write that there is continual emission of fragments during deep descent in a convective downdraft (e.g. 2 m/s) that is perhaps 1 or 2 km deep. Irrespective of the rate of mixing from the downdraft to any adjacent ascent, the ice concentration reaches a quasi-steady state due to the balance between generation and depletion of fragments by total sublimation. This is what we mean by quasi-equilibrium.
  According to our conceptual picture, there is no need for spatial proximity between subsaturated and supersaturated regions because the same quasi-equilibrium ice concentration will be transferred regardless of the rate of the mixing or timing of re-circulation from downdraft to updraft. Similarly enhanced ice concentrations from sublimation breakup will be in the recirculated air whether entering the updraft sooner or later. In the commentary we concluded:
  "The pivotal point here is that such a quasi-equilibrium concentration is maintained throughout the entirety of the subsequent
  
  [convective] descent after being reached. Thus, any recirculation of downdraft air into the surrounding convective ascent would transfer
  
  air enriched in fragments for their subsequent vapour growth and survival."
  We hope that in the rest of this debate here the uniqueness of our ideas can be at least acknowledged.
  
  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.
  Phillips, T.J., Yano, J.-I., Deshmukh, A., and Waman, D.: Comment on “Review of Experimental Studies of Secondary Ice Production” by Korolev and Leisner (2020), Atmos. Chem. Phys. Discussion, https://doi.org/10.5194/acp-2021-123, 2021
  
  Citation: https://doi.org/10.5194/acp-2021-123-CC4
RC2: 'Comment on acp-2021-123', Anonymous Referee #2, 25 Apr 2021

This reviewer does not find the ideas and criticisms of KL2020 in manuscript particularly well organized or helpful with regard to evaluating the importance of ice-ice collisions to SIP. The discussion is largely qualitative and a repeat of what has been published previously by the authors. The language used is in several places exaggerated discussing the validation of both the Takahashi1995 experiments and the author’s modeling results (presented in Yano2011, Yano2016, Yano2016b, Phillips2017, Philips2017b and Phillips 2018, hereafter termed "Y&P"). The Y&P work makes an important contribution to SIP research and several papers are cited in KL2020 but not described in detail as the focus of the review is experimental studies of SIP. In the early papers Y&P use a single parameter, the number of fragments produced per collision, extracted (and scaled) from the Takahashi1995 laboratory experiments to characterize the fragmentation. The parameter in Yano2011 was not temperature, humidity, LWC or particle size/character dependent. A more complex formulation is discussed in Phillips2017,2018 but this is not discussed here. At issue in this manuscript is the validity of the Takahashi1995 measurements and the simple scaling used in Yano2011,2016 to describe the in-cloud SIP process. Does the fragmentation observed by Takahashi1995 using 2 cm ice balls held on rods accurately simulate a SIP process in clouds? For several reasons I think this remains an open question.

I have the following comments on the manuscript:

1) Line 11: Shouldn’t the word be "untested" rather than "erroneous" in describing the current situation? It seems KL2020 is not suggesting Takahashi1995 is erroneous. We have no basis to conclude the validity one way or another. Instead, there is concern these results do not simulate actual in-cloud collision process. Confirming experiments with free-falling and proper-size ice particles have not been done yet.

There is a hint in the manuscript (see lines 145, 281, 288, 301) that the authors believe the comparison of their model results with field-study measurements (like in Phillips2017b) is in sufficient agreement as to validate a fit-derived fragmentation parameter and are not sensitive to the value extracted from Takahashi1995. If this is the claim, then this point is important and needs to be discussed in its own section. Phillips2017b shows good simulation-observation agreement but it is unclear to what extent the fragmentation parameter is the only free-parameter. More discussion of how one can determine a fragmentation parameter from a SIP cloud model-field data comparison would be interesting.

2) Line 22: "Impossible" seems an exaggeration. The question is whether the laboratory result and scaling used in Yano2011 is a realistic description of in-cloud processes.

3) Line 23: "unreliable, which is not the case," - this statement needs to be explained. On what basis do the authors claim to know the Takahashi1995 data set is a reliable representation of in-cloud collisions? I have a similar comment to the text on Line 41 and several other places in the manuscript. Are the authors claiming a Phillips2017b-like analysis of other field data sets also shows good model-data agreement? Wasn’t the breakup parameterization used here more complex than simply the size, velocity, and KE scaling of Takahashi1995? Perhaps provide more discussion of the later model refinements and the evidence for the Takahashi1995 extracted parameter used in Yano2011?

4) Lines 52-110: This material repeats much of what is stated in various places in Y&P.   It is so scattered and un-quantitative as to provide few new insights. A more organized presentation would be appreciated.

5) I comment on the Takahashi1995 characterization: This reviewer doesn’t find the Takahashi1995 result particularly compelling as a simulation of what occurs between free ice particles in clouds. Not to say it is erroneous, just on it’s face not compelling. The reviewer’s opinion isn’t particularly important here but again a more organized quantitative analysis of the Takahashi method rather than the collection of scattered hand-waving arguments would be appreciated if this is to be one focus of the manuscript.

The SIP mechanism is unknown at this point. There have been several suggested ideas. The authors have published a model based on ice-ice fragmentation during collisions and claim their model can describe the process. The Takahashi1995 study collided two 1.8 cm diameter ice spheres, counted the crystals on a collector plate covering a fraction of the chamber bottom, and then multiplied that number by 4 as an estimate for the number of crystals ejected from their colliding spheres. Are these crude experimental finding for 2 cm sphere collisions an accurate representation of the ejection rate for ice crystals in clouds?

The scaling relation used in Yano2011 (to apply Takahashi1995 result to realistic cloud particles) considers only differences in particle diameter and fall velocity. Later in Y&P kinetic energy, growth time, vapor density, collision dynamics/type and stochastic considerations are all mixed into this scaling. But the scaling has not been confirmed by experiments. Korolev2010 mentions several questions in applying Takahashi1995 to actual cloud particle collision processes. I won’t describe these considerations but will discuss several other considerations.

There are several ideas for how fragments occur during collisions. Apparently for some temperature, LWC, convection and humidity/riming conditions, cloud processes produce irregular or “fuzzy” ice spheres with fragile irregularities protruding from their surface. The idea presumes the protuberances grow with time and their fragility may (or may not) also increase. When a collision occurs involving at least one of these fuzzy particles some protuberances break off as fragments. This potential secondary ice production mechanism requires the fragments somehow find themselves a region with sufficient humidity to survive and grow thereby increasing the ice particle number density. Andy H’s comment describes cases where the fragments likely will not survive. Phillips2017b describes a case where the fragments apparently do survive. Others will have different or more sophisticated ideas for the microphysics. But in this SIP mode the fragments originate at the surface of one or more of the colliding particles. The surface of the particles involve roughening or new crystallite nucleation such that the protuberances grow via riming or vapor deposition. These processes are temperature, RH, LWC and particle-size dependent. The surface roughness of the spheres in the Takahashi experiments were not characterized. At the surface during lumping/roughening or protuberance formation there will be epitaxial effects from the underlying crystallinity that likely will depend on the initial formation and growth process of the underly crystal. The 2 cm spheres in Takahashi1995 began as frozen liquid water inside a metal sphere. This freezing process is quite different from the variety of ice particle formation processes that occur in clouds undergoing SIP. Potential differences in the ice surface properties alone might call into question the relevance of the experiment to actual cloud particle-particle collisions.

Second, the collision itself is likely different from what occurs between cloud particles. In the experiment the 2 cm spheres move via a rod frozen into the center of the sphere. The rigidity of the rod is important to the amount of energy exchanged in the inelastic collision. A springy rod and axel will cause the balls to react much different than a ridged rod. Also during the collision, the strain along the axis of the rod will be different than in other directions causing perhaps larger amounts of gouging into the ice surface than would occur with free particles. An apparatus holding the ice on rods adds complications to evaluating the forces in each inelastic collision. One suspects the energy transfer and the potential for gouging into the ice surface are different from what occurs when free-particles collide. There are also aerodynamics considerations and charging effects for both the particles and the fragments created in colliding smaller crystals. Perhaps all these potential effects wash out and the simple kinetic energy considerations are good enough to describe what is occurring. But it does seem fair that some experimental work using um scale and larger particles is needed before one can be confident the simple scaling idea, like that applied in the Yano2011 analysis, is valid.

This reviewer suggests the manuscript needs considerable re-work and clarification.

Citation: https://doi.org/10.5194/acp-2021-123-RC2

Vaughan T. J. Phillips et al.

Viewed

Total article views: 642 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
434	188	20	642	3	2

HTML: 434
PDF: 188
XML: 20
Total: 642
BibTeX: 3
EndNote: 2

Views and downloads (calculated since 15 Feb 2021)

Month	HTML	PDF	XML	Total
Feb 2021	252	117	11	380
Mar 2021	115	47	5	167
Apr 2021	64	23	4	91
May 2021	3	1	0	4

Cumulative views and downloads (calculated since 15 Feb 2021)

Month	HTML	PDF	XML	Total
Feb 2021	252	117	11	380
Mar 2021	115	47	5	167
Apr 2021	64	23	4	91
May 2021	3	1	0	4

Viewed (geographical distribution)

Total article views: 653 (including HTML, PDF, and XML) Thereof 653 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 05 May 2021

Short summary

For decades now, high concentrations of ice observed in precipitating mixed-phase clouds have created an enigma. Such concentrations are higher than can be explained by the action of aerosols or by the spontaneous freezing of most cloud-droplets. Partly, the controversy has persisted due to the lack of laboratory experimentation in ice microphysics, especially regarding fragmentation of ice, a topic reviewed by a recent paper. Our comment attempts to clarify some issues about that review.


Total:	0
HTML:	0
PDF:	0
XML:	0