the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 06 Sep 2023
Measurement report: Influence of long-range transported dust on cirrus cloud formation over remote ocean: Case studies near Midway Island, Pacific
Abstract. Cirrus clouds play an essential role in regulating the global radiative balance and climate by both reflecting the incoming shortwave solar radiation and reserving the outgoing longwave radiation in the atmosphere. The cirrus-induced net radiative forcing is mainly determined by their microphysical properties, which are strongly associated with the competition between two ice-nucleating mechanisms, i.e., heterogeneous and homogeneous nucleation. However, it is still not well understood whether the long-range transoceanic dust can potentially urge heterogeneous nucleation to the initial ice formation in cirrus clouds even farther over vast remote ocean regions and the response of dominant ice-nucleating mechanism to the concentrations of available ice nucleating particles (INPs). Here we report on the influence of transpacific dust plumes on the ice formation in cirrus clouds via heterogeneous nucleation based on the combined observations of space-borne instruments, i.e., the Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) and Cloud Profiling Radar (CPR). Two cases near Midway Island (28.21° N, 177.38° W), located in the central Pacific, are studied, in which the long-range transported dust plumes originate from intense Asian dust events. For both cases, partial cloud parcels show the typical in-cloud ice crystal number concentrations (ICNC) of <100 L-1 for heterogeneous nucleation with a good agreement (within an order of magnitude) of in-cloud ICNC and nearby dust-related INP concentration (INPC) values, indicating that dust-related heterogeneous nucleation is dominated in ice formation. In addition, for the other parts of clouds without sufficient INP supply, homogeneous nucleation can still be dominated with ICNC values exceeding 300 L-1. Therefore, dust events with sufficient intensity are capable of conducting long-range transport and influencing cirrus formation over remote ocean regions. This study shows that the natural supply of effective INPs to the upper troposphere, such as long-range transported dust aerosols can increase the cloud cover to reflect more solar radiation over oceanic regions and modulate the microphysical properties of cirrus clouds through different ice-nucleating regimes, both of which may further result in a cooling effect on global climate and should be well considered in climate evaluation.
- Preprint
(3437 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2023-1844', Anonymous Referee #1, 03 Oct 2023
General Comments:
This is an interesting study that relates mineral dust plumes from Asian deserts to the number concentration of ice particles in cirrus clouds over Midway Island in the Central Pacific. It does this strictly through satellite remote sensing, using retrieval methods for mineral dust concentration (i.e., ice nucleating particle concentration or INPC) and ice particle number concentration (ICNC). It is well written and organized, but some of the arguments do not appear to be supported by the data, and the results may be overinterpreted. Specifics are given below. I recommend major revisions in rewriting the article, but this may not require a great deal of additional work.
Specific Comments:
1. Figure 4: The high column dust density measurements shown here in the back-trajectory do not ensure that air overlying these Asian deserts at ~ 10 km is having a relatively high concentration of mineral dust, since the column dust magnitude could result almost entirely from dust much below 10 km. This needs to be stated. On the other hand, it is commendable that the authors did this analysis since, although incomplete, it may be using all the available data and does provide important information.
2. Figure 5: There appears to be a problem with color legend. The plotting background is violet, which corresponds to an ice water content of 10-20 mg/m3. Not possible. The same problem occurs with Fig. 9c.
3. Lines 234-235: For what purpose are ICNCs shown for n_ice,25um and n_ice,100um? They provide no closure information with respect to dust INP concentration since most of the ICNC can be associated with D < 25 um (Kramer et al., 2009, ACP). Only n_ice,5um is relevant to the closure being sought with INPC.
4. Lines 236-237: Suggest citing Diao et al. (2015, JGR) to back up this statement.
5. Lines 252-253: Figure 6e shows agreement between ICNC and INPC only near cloud top for Si = 1.15, where ICNC is for n_ice,5um. From the previous page, it states that the layer average INPC is 7 L-1 and 96 L-1 for Si of 1.15 and 1.25, respectively. Since the layer average ICNC is 111 L-1 for Part B, optimal agreement is for Si = 1.25, with ICNC-to-INPC ratio of 1.16. Therefore, Si here should be 1.25, not 1.15.
6. Line 253: It is not clear how the ICNC-to-INPC ratio can be 0.9 for Si = 1.15, based on the previous text. As noted above, this ratio appears to be 1.16 for Si = 1.25, for which closure is optimal.
7. Lines 317-320: Since a large portion of the ICNC resides at D < 25 μm, n_ice,5um should be used rather than n_ice,25um (as assumed in this study). This practice (of using n_ice,5um) was followed for the other May 5th case study. In Fig. 10e, Si = 1.25 agrees best with n_ice,5um, consistent with the previous case study based on Comment 5 above.
8. Lines 308-312: If Si = 1.25 in the 2nd case study as indicated above, then the cirrus cloud of Part A could be produced only by heterogeneous ice nucleation since n_ice,5um = 635 L-1 and the mean dust-related INPC (U17-D) is 417 L-1. This is contrary to what the article states here; that Part A is dominated by homogeneous ice nucleation.
9. Lines 350-352: Alternatively, could this also be explained by variability in cloud updraft velocities?
10. Table 2: As mentioned earlier, for what purpose are ICNCs shown for n_ice,25um and n_ice,100um? They provide no closure information regarding dust INPC.
11. Lines 375-377: RHi (relative humidity with respect to ice) rarely reaches 140-150% in cirrus clouds since het (heterogeneous ice nucleation) always occurs before hom (homogeneous freezing nucleation), and INP and/or pre-existing ice tend to prevent the RHi from reaching the RHi threshold for hom. But if INP concentrations were low enough, the RHi threshold for hom would occur much more often to produce hom cirrus clouds, and their coverage could even exceed the coverage of het cirrus due to the smaller ice crystal sizes having lower fall speeds, as demonstrated in Mitchell et al. (2008, GRL). That is, lower ice sedimentation rates lead to longer cirrus lifetimes and greater cloud coverage. Moreover, the citation of Dekoutsidis et al. is misguided since that paper was showing hom is common in cirrus clouds and occurs mostly near cloud top where RHi is greatest, consistent with the modeling study by Spichtinger and Geirens (2009, ACP). The reference by Cziczo et al. does not support the author’s claim either; rather it argues that most cirrus are het cirrus.
12. Lines 377-380: While changes in UT INPC may alter the microphysical properties of cirrus clouds, this may not result in an increase in cloud cover and associated albedo for the reasons stated above. Moreover, the net radiative effect of cirrus clouds considers the absorption/emission of LW radiation in addition to SW radiation, and whether a net cooling or warming effect occurs may depend primarily on cloud optical thickness, the season, and the latitude.
Citation: https://doi.org/10.5194/egusphere-2023-1844-RC1 - AC1: 'Reply on RC1', Yun He, 08 Dec 2023
-
RC2: 'Comment on egusphere-2023-1844', Anonymous Referee #2, 13 Oct 2023
The manuscript discusses two interesting dust-influenced cirrus events over the Pacific Ocean in the outflow regime of Eastern Asia. The remote-sending-based study makes use of spaceborne lidar and radar observations. The work is original and worthwhile to be published in ACP. The main effort is related to closure studies, i.e., deals with the question to what extent the ice-nucleation particle concentration (INPC, estimated from lidar observations) is in the same range of estimated ice crystal number concentration (ICNC, estimated from combined lidar-radar observations). Based on these ICNC-vs-INPC closure studies the authors also discuss to what extent homogeneous ice nucleation was involved in the (in situ) cirrus evolution.
Many aspects are unclear, a lot of speculative argumentation is given in the main result sections 3 and 4. A more careful discussion is requested. The uncertainty in all the retrieval products needs to be better considered. Therefore, major revisions are required.
One of the key points for me is: All the retrieval products (INPC, ICNC) can only be obtained within a large uncertainty range of an order of magnitude. ICNC cannot be obtained with an uncertainty of 25% as mentioned in the present study! Impossible! Even if you have well calibrated lidar and well calibrated cloud radar observations, and, in addition, Doppler information about fall speed of ice crystals (and thus shape and size information) the uncertainty can never be lower than expressed by a factor of 3-5 around the retrieval products. This is, e.g., shown in the reference, you provide (Ansmann et al., ACP, 2019). This is also discussed in detail by Buehl et al., AMT, 2019. Now, in the case of the CloudSAT radar we do not have Doppler information. So, the ICNC uncertainty is even higher. All in all, the uncertainty margins are roughly an order of magnitude for INPC and ICNC. That needs to be considered in the closure studies presented here and in all the conclusions drawn from the observations. To my opinion, this will make the full discussion easier and more straight forward.
Details:
Title: This study is not a measurement report. This is an in-depth research approach (cirrus closure study) based on the analysis of spaceborne observations. The term ‘Measurement report’ suggests that just robust observations (measured data) are presented. But the study is widely based on estimated products and interpretation of the findings.
Abstract: The abstract needs to be rewritten after all necessary changes.
Introduction: I would keep the introduction as short as possible. There are so many cirrus papers and all present lengthy paragraphs on the radiative impact. I would suggest that the importance of cirrus in the atmospheric system is just briefly described, followed by gaps in our knowledge regarding cirrus evolution, and then what you are going to present in this study.
Lines 41-42: Note that homogeneous freezing also needs aerosol particles, however pure liquid ones, such as sulfate aerosol (without any insoluble part). In the case of heterogeneous ice nucleation, one needs particles with an insoluble fraction (sites for ice nucleation).
Figure1: Does MERRA also deliver dust profiles? or only column mass values? Would be nice to have model dust profiles up to cirrus level.
Method section:
Line 130: 25% uncertainty is unrealistic as discussed above. To use uncertainty margins of an order of magnitude (a factor of 3 around the retrieved value) makes sense. That is more realistic!
Lines 140-145: I would remove all immersion freezing parameterizations. The two case studies deal with cirrus from 9 to 11.2 km height, and temperatures from -40 to -54°C. Furthermore, your Table 2 considers deposition ice nucleation, only (no immersion freezing INP values are given). Even in virga zones with higher temperatures, there is ice saturation or even ice sub saturation (and especially sub saturation with respect to water). Immersion freezing is impossible in the presence of in situ cirrus with well developed virga structures.
Table 1: Are all the equations needed? A list of parameters and retrieval uncertainties makes sense. Again, please skip all immersion freezing parameterizations.
Result section:
The cirrus shown in Figure 3 is a classical in-situ cirrus, with a well defined cirrus top region, obviously above 10 km height, and a virga zone from 10 km down to 6 km height. The cirrus is clearly dust influenced.
Let me explain how such cirrus layers develop: Ice nucleation starts at cloud top, at the coldest point of the cloud, here the ice nucleation probability is highest. And if dust particles are present, they will trigger ice nucleation. Then diffusional growth of the crystals takes place, collision and aggregation. Sedimentation of ice crystals begins and the growth of the crystals leads to an immediate reduction in relative humidity (throughout the cirrus layer from 6 to 10.5 km height). There is almost no room for homogeneous ice nucleation (in the case of a well-developed cirrus system) because there is practically no potential to create a scenario with sufficiently high relative humidity over ice with values exceeding 150%, required for homogeneous ice nucleation. Homogeneous freezing is only possible at cloud top in the case of rather strong updrafts with large updraft speed so that super saturations levels develop faster than INPs can be activated (to reduce super saturation). All this is described in Kaercher et al. , JGR, 2022. However, such a scenario with rather strong updrafts is not visible in Figure 3. A very harmonic cirrus development exclusively controlled by the activation of dust INPs seems to be realistic. Furthermore, radiative cooling at cloud top of a well-developed cirrus system will cause some amount of downward motion and may contribute to a suppression of such rather strong updrafts. If we keep an uncertainty of about an order of magnitude into account for ICNC and INPC, all observations support that heterogeneous ice nucleation on dust particles dominated. Because of the large uncertainty, I would not further analyze the defined observational periods (part A and part B), i.e., explain the differences in the ICNC values in terms of heterogeneous vs homogeneous ice nucleations. The differences in the results for part A and B just show, to my opinion, the uncertainty in the products.
Back to the study:
Figure 3: The lidar is able to see the small ice crystals after nucleation with sizes of 1-5 micrometers at cloud top, above 10 km. The cloud radar seems to be not able to detect the ice crystals above 10 km height (Figure 5). The radar detects the ice crystals after diffusional growth and collision and aggregation processes, and after the start of sedimentation processes, i.e., several 100 m below cloud top, in the virga zone, as can be seen in Figure 5a and 5c. So, lidar-radar retrievals will not be able to exactly see the ice crystal number concentration of freshly nucleated ice crystals at cirrus top. And when radar comes into plays, aggregation took already place, and the number of crystals already decreased, may be already by a factor of 2, or even a factor of 3-5. Another point: Can we assume a ‘classical’ size distribution (as typically measured with aircraft instrumentation) in the case of freshly formed ice crystals? The size distribution is input in the DARDAR approach, and may be very narrow for the freshly nucleated crystals? All these unknown aspects cause the large uncertainty in the ICNC products (of at least one order of magnitude).
The ICNC values in Figure 5d vary strongly and indicate the uncertainty in all the retrieval products. Therefore I would not introduce part A and part B and ‘believe’ that the rather different findings are caused by different ice nucleation processes. One may formulate hypotheses…, but one needs to consider the large uncertainty in the discussion. Solid conclusions are difficult to draw. And as I mentioned, I am skeptical that homogeneous ice nucleation has a chance to occur in the presence of a well-developed cirrus. The CALIOP lidar indicates dust around the cirrus and no indication that the dust INP reservoir was depleted. To my opinion, a depletion of the the INP reservoir is unlikely during part A and B, and therefore homogeneous ice nucleation is unlikely.
If we keep the uncertainty of one order of magnitude in mind, the ICNC values for part A and B shown in Figure 6e nicely indicate the ICNC uncertainty range. To repeat, to my opinion, only heterogeneous ice nucleation makes sense. In the presence of so many rather favorite dust INP particles (as shown in Fig. 6c, 4000 large dust particles with a diameter > 500 nm per liter were present, and the corresponding dust surface area concentration was high with values up to 40 µm2 cm-3) homogeneous ice nucleation is rather unlikely.
Figure 6 indicates that INPC values for S-ice =1.15-1.25 (Ullrich parameterization, deposition ice nucleation) seem to be very likely (or reasonable) and match roughly the ICNC values (n-ice for crystals with sizes > 5 µm, for the periods of part A and B), when keeping in mind that ICNC is probably underestimated close to the cirrus top (because DARDAR values are not very trustworthy here because of too weak or even missing radar reflectivity values). The best DARDAR values are shown after significant growth of crystals and after aggregation processes (from 9-9.6 km height), but then the ICNC values are already reduced compared to the number of freshly nucleated crystals at cirrus top, probably reduced by a factor of 2-5.
Line 242: As mentioned above, homogeneous freezing in an environment with already existing ice crystals (and thus super saturation values S-ice around 1.0) is very unlikely. When keeping the high dust load and the large uncertainty margins into account, the closure is fine, ICNC and INPC match reasonably well. This is ok! This is a good result.
In summary here, please, do not compare part A and part B, just take the average of both periods and use the averaged values for comparison with the Ullrich results for INPC. Ice super saturation values of 1.15 are close to the values in the paper of Ansmann et al. (2019) for pure (unpolluted) dust scenarios. In the case of aged dust (or polluted dust, case study 2 in this study here) super saturations of 1.35 make sense. Such values are assumed to be realistic for aged, coated dust particles (see Kaercher et al, JGR, 2022). Homogeneous ice nucleation events do not make sense to me at all. However, I leave it open to you to find a proper and careful argumentation for the potential contribution of homogenous ice nucleation.
In the discussions (sections 3 and 4) there are many speculative aspects. Speculations are not justified. A more careful interpretation of the results is needed. And if you have a hypothesis, start with: We hypothesize…. and then the hypothesis must be based on convincing argumentation. And please keep the large uncertainties in mind.
Case study 2:
Another nice case with an impact of dust. In this case , CALIOP aerosol typing seem to indicated polluted or coated dust. The ice nucleation efficiency of aged and coated dust particles may be reduced by a factor of 5-10 compared to the ice activity for pure dust. This may or could be considered when computing INPC values with the Ullrich parameterization by multiplying the Ullrich INP values by 0.1-0.2.
Again, I would not compare the results for part A and part B because of the large uncertainties. The results in Figure 10e support my comment here. The results as a whole (part A and B) are in good agreement with the Ullrich INP values for S-ice of 1.35 when considering a factor of 0.1-0.2 less INPC (in the case of polluted or coated, less ice nucleation efficient dust). Kaercher et al., JGR, 2022, used S-ice values around 1.35 for the activation threshold for polluted dust. In case 2, the ICNC values (from the DARDAR approach) increase up to cloud top. Obviously, the radar reflectivity values were strong enough to obtain reasonable ICNC values even close to the nucleation range at cloud top.
To my opinion, there is again no room for homogeneous ice nucleation. There is dust ‘before’ and ‘after’ the cloud region, so a depletion of the dust INP reservoir is not visible. And at these conditions, homogeneous ice nucleation is unlikely.
Please avoid speculations on cirrus type, etc…. in sections 3 and 4. Just mention, what is really available from the observations.
Discussion section:
The first paragraph is not needed to my opinion.
Line 346: Can we have longitude-latitude information (not only latitude). How long (in km) was the cirrus layer? The same for case study 1.
Be careful with ‘dominating homogeneous freezing’ in environments with so much dust. It is simply difficult to produce high ice super saturation in the presence of favorable INPs.
If you follow my suggestions, youcan significantly ‘improve’ Table 2 by reducing the information content. I think ICNC for n-ice (> 5 µm) has to be compared to INPC, however information on n-25, n-100 is useful as well, especially to get a better feeling for the large uncertainties in all ICNC products.
In the case of the Ullrich INPC values, input is dust surface area concentration, S-ice as well as temperature! Temperature needs to be mentioned in Table 2 (Ullrich INP values).
Citation: https://doi.org/10.5194/egusphere-2023-1844-RC2 - AC2: 'Reply on RC2', Yun He, 08 Dec 2023
-
RC3: 'Comment on egusphere-2023-1844', Anonymous Referee #3, 18 Oct 2023
Referee report for the manuscript entitled:
“Measurement report: Influence of long-range transported dust on cirrus cloud formation over remote ocean: Case studies near Midway Island, Pacific”
Authored by Huijia Shen, Zhenping Yin, Yun He, Longlong Want, Yifan Zhan, Dongzhe JingIn this manuscript, remote sensing data has been used to determine whether long-range transported dust from Asia is active as INP in the formation of cirrus clouds. Data are collated from two space-borne instruments to calculate INP concentrations (INPC) and ice crystal number concentrations (ICNC). These values are compared at specific locations within an observed cirrus cloud that relate to where the dust would most likely be entrained. Where the values of INPC and ICNC are comparable, the manuscript concludes that this region is dominated by heterogeneous freezing due to the dust. Where the ICNC is much higher than the INPC, the manuscripts concludes that homogeneous freezing also takes place. Previous studies have shown that long-range transported dust acts as INP (Saharan dust in North America) and Asian dust has been shown to act as INP in China. This manuscript shows that Asian long-range transported dust also acts as INP. I am not an expert of INP, however having experience in adjacent aerosol fields, I am aware that there is still much to understand about INP sources and this manuscript confirms another significant source of dust and its far-reaching impacts.
When considering this manuscript for publication, there are several factors to take into account:
- This manuscript is presented as a measurement report, however the data are retrieved from space-borne instruments rather than field or lab work. Additionally, there is analysis and interpretation of the results which potentially goes beyond the requirements of a measurement report.
- This manuscript does not present novel ideas or methods, however it does establish evidence of long-range transported dust acting as INP in a new area.
- For me, there are some major questions about the analysis of the individual boxes within the cloud. It appears that one of the boxes closest to the dust plume is described as lacking in INP and, as such, homogeneous freezing is dominant. This description requires some justification.
- The methods would be enhanced by some more detailed description.
- Restructuring of some paragraphs and tightening up the language in a few key sentences would greatly aid the reading and understanding of the manuscript.
Based on the above summary and the comments below, I recommend this manuscript is reconsidered after major revisions. I also recommend that this should be submitted as a research article if the following suggestions make the manuscript a more substantial contribution. Note that I do not have expertise in satellite products so am unable to comment on their descriptions here and the suitability of their use.
Major points
On the description of the method, the key aspects from He et al (2022b) are mentioned throughout the manuscript, but it is quite scattered. I would suggest pulling this all together within the methods section so that the method is very clear for a reader who is unfamiliar with it. It is clear that if the ICNC is much higher than INPC then homogeneous freezing must be present, but it is not so clear how one knows that it is only heterogeneous if the ICNC is comparable with INPC. Why could it not still be both? I am not an expert in INP, but I think the key bit of information here is that the ice saturation will not reach the required 140 or 150% if heterogeneous freezing has already begun.
The methods could also contain a description of the two case studies before we dive into the results. In the discussion there is a nice point about the dust being from intense events where the dust is elevated by Mongolian anticyclones. Information like this would be better used in the beginning of the manuscript to set the scene. Is there other relevant meteorological information about the cases? Was there a particular reason for picking these two cases? What times and altitudes were used? Are the cirrus events very rare? A description of the two case studies would follow on nicely from the HYSPLIT model description in Section 2.
The results could be restructured to make the storyline clearer. For each case, I would recommend starting with establishing the case using the HYSPLIT trajectory and the dust mass column density plots. Then the description of the cloud and the dust plume using the CALIPSO cross-sections would follow on nicely. Next, part A and part B need a clear introduction using the DARDAR product. Why were these regions in particular picked? The difference in the ICNC between the two boxes could be described and then finally use the profiles to compare the ICNC values with the INPC. Here, I would suggest stating the ICNC versus the INPC and the ratio for part A and B. Do these values alone indicate homogeneous or heterogeneous freezing? The main method used in the manuscript is this quantification so I would pull it out of the data and emphasise it. Then add the additional, contextual information from other papers about typical values for each mechanism.
Table 2 shows the summary of ICNC and INPC for both cases split into the A and B boxes. Considering that the concentrations are defined as being “comparable” as within an order of magnitude, there does not seem to be a large distinction between the A and B boxes. This is particularly true for the first case where the ICNC values are quite similar for both boxes. For the second case, there does indeed seem to be a distinction between the boxes, however the INPC values still seem much higher than both boxes. These values would benefit from some clarification, and perhaps this could come in the results section if it were restructured as recommended above.
Minor points
The title could state the outcome of the study since an effect has been found by this study. For example, “Long-range transported Asian dust plumes influence cirrus formation over remote ocean.“
The abstract has a nice structure to it, but the sentence beginning “However, it is still not well…” (line 14) is confusing. This is a key sentence and rephrasing it would make the motivation of the research clear in the abstract. The results could also be stated more clearly, for example including the ICNC - INPC ratio.
The comment on line 47 about geoengineering comes across as quite random and a bit of an afterthought. Perhaps it could be rephrased to tie in with the previous sentence about the uncertainty and how understanding of these clouds needs to be improved in order for cloud seeding to be done appropriately.
The introduction has good content to give context to and justify the study. It does very well to lay out why we care about cirrus and this long-range dust transport. However a couple of paragraphs here could be restructured to improve the logic:
- The paragraph beginning at line 50 could be split into first discussing the pristine environment, low AOD, and the related lack of observations of cirrus. It is also a bit unclear whether there are not many cirrus or there are many but it is unexpected because of the pristine environment. Then the question of whether this means cirrus are purely formed by homogeneous could be framed.
- The ice saturation conditions for both mechanisms is a key part of the background science and this could be written in a separate paragraph. This could then link dust as an INP and give a description of why homogeneous freezing is much less likely to occur if dust is present (high ice saturation is prevented).
- The paragraph beginning at line 67 could also be restructured. From the studies cited here, I have understand that we know Saharan dust travels long distance, we know that Asian dust affects cirrus in China and also travels long distance to North America. So it seems to me that the question is not really whether the dust affects cirrus a long distance away, but rather is this long-distance transport creating a significant effect in an otherwise pristine environment? This paragraph could lead the reader more to this question about what is happening over the ocean.
- The paragraph beginning at line 83 could start with stating that He et al (2022b) determined the freezing mechanism by comparing INPC and ICNC before explaining why this works. Some clarification explaining how one knows that a mechanism is “dominant”, rather than there just being a mixture of both, would help the reader. There is good placement of the descriptions of the products here and nice summary with referencing.
The point about seeder feeder in stratocumulus clouds (line 80) could be made earlier because it is addressing the big question of why do we care about these cirrus? Perhaps it could be stated near the geoengineering comment as wider motivation for understanding cirrus.
On line 82, there is the phrase “it is indispensible” but it is unclear what it is. I am not sure if this line adds anything.
The overview of the paper (line 103) could be clearer and include “section 3”.
The manuscript tends to state what the figure is in the text, e.g. “Figure 5 shows the ice cloud properties, including cloud extinction coefficient, cloud particle effective radius, ice water content, 200 and ice crystal (with size <5 μm)....”. They could consider removing this from the main text since it is already described in the figure caption. This may make the story flow better as the main text can then go straight into what is observed in the figure, e.g. “Figure 5b shows the region of dust-related cirrus... ”.
The introduction to the results sectioned could be cut as there is some repetition that one INP is generating one ice crystal and that secondary ice productions is not being considered to take place in these conditions. But the statement of what is required for “good” agreement and the definition of dust-cirrus interaction event are well placed here.
The authors might consider adding an explanation for why the description of the dust in the main text based on figure 3a and b (line 184) does not fit with where the dust is identified in figure 3c and e.
In the main text about figure 5, the in-cloud averages are stated but there is no interpretation of these. In the introduction it is stated that homogeneous freezing produces more, smaller ice crystals. Do the in-box averages support the allocation of homogeneous/heterogeneous freezing based on the radius and ice water content?
Some clarification on line 236 about “pristine ice crystals” in abundant INP near the top of the cloud could help the reader. Are these pristine ice crystals because they are from homogeneous freezing or because they are formed in pristine environments, where the INP are more numerous at cloud top? Perhaps changing the description of the INP from abundant to more numerous would help.
In the discussions of figure 6e (and for 10e), it might not be clear to the reader how the calculations of INPC at different ice saturation ratios relate to the ICNC values. Does each saturation ratio relate to a different radius? A statement of which INPC and ICNC values are actually being compared for both part A and part B would make it clear to readers like me that are unfamiliar with the POLIPHON method.
For part B starting at line 244, the message would be clearer if the comparison with the INPC came before putting it in the context of Ansmann et al (2019a), Cziczo et al (2013) and the interpretation of the dust acting as INP in the moist region. Additionally if this is the moist region shown by the RH in figure 6d, this could be linked with a “as shown in figure 6d”.
In the paragraph beginning on line 346 with “The overview of …. “, the authors might consider moving the sentences about the dust being uneven and the lack of INP resulting in homogeneous freezing near the start of the paragraph. They could then summarise the findings from part A and B for each case.
Related to the above point, a comment on why part B is shown to have lacking INP but is closest to the region of dust is perhaps warranted.
Line 375 states that cirrus would not form without these natural INP, but this conflicts with the establishment that homogeneous freezing does take appear to take place. Perhaps change to something similar to “would rarely form” or “there would be a much lower frequency of clouds”.
The authors have done well to suggest plenty of future work based on this study.
Technical points
There is some inconsistency between using the terms secondary ice nucleation and secondary ice production in lines 84 and 174.
Blue and purple colours in the profile plots (figures 6e and 10e) are quite hard to distinguish. These could be more contrasting colours.
In figure captions, the authors could consider separating out figure titles from the a), b) descriptions. For example, the figure 2 caption could be “Dust-related conversion…” and then a) and b) to describe each plot. Also in figure 6 and 10.
On line 164, what does a “complete system” mean here?
Some typos:
- The abstract contains some use of the word “dominated” rather than dominant. This might be present in other places too.
- Paragraph starting at line 164 contains several cases of HYSPIT instead of HYSPLIT.
- Line 63, what is the occurrence rate of cirrus measured in?
- Line 73: “Saseen et al 2003” should be Sassen
- Line 75: should “Sassen et al 2001” be 2002?
- Line 119: “MEERA” should be MERRA
- Line 169: “stimulated” → simulated
Citation: https://doi.org/10.5194/egusphere-2023-1844-RC3 - AC3: 'Reply on RC3', Yun He, 08 Dec 2023
-
RC4: 'Comment on egusphere-2023-1844', Anonymous Referee #4, 23 Oct 2023
The manuscript by Shen et al. presents an observational study of cirrus cloud formation near Midway
Island in Pacific Ocean and the ice-nucleation properties of long-range transported desert dust. Authors
have used observational data from Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) and
Cloud Profiling Radar (CPR), DARDAR and MERRA-2 datasets, and POLIPHON and HYSPLIT mod-
els for the data evaluation. As is, the manuscript addresses an important topic (especially for atmospheric
and climate modeling communities) – the impact of long-range transport dust on the formation of cirrus
clouds – but does not present an answer. Before the paper is published, some major concerns would have
to be addressed.My greatest concern with the manuscript is the lack of indepth analysis – additional details (such
as the nucleation mechanism, connection between the initial nucleation conditions and lab experiments)
would improve the manuscript significantly. For this reason, these Major issues in the manuscript should
be addressed:1. The idea and dataset chosen to show the nucleation and cloud formation are great, but the data
analysis seems to be focused on an observation. It would be very interesting to see the atmospheric
conditions data and the discussion of the initial heterogeneous ice nucleation (e.g., the transition
from pure dust to pure cloud at 30.34/0.0 lat/lon on 2010-05-05 trajectory) to see the relation to
the laboratory experiments (such as Koehler et al., Atmos. Chem. Phys., 10, 11955–11968, 2010).2. The manuscript is presented as a Measurement Report, however it is not a new or original measure-
ment, but rather a reanalysis of an old and public dataset. As a result, the manuscript does not
provide substantial insight and conclusions. I would suggest the authors to perform and write-up
an in-depth analysis of the processes based on their expertise and reconsider the manuscript as a
research article.3. Another major shortcoming of the article is the Results section, where most of the section is
focused on spelling out the results seen in figures with little interpretation. It is good to see that
the summary of the results has been provided in Table 2, in the Discussion section, but I would
expect it as part of Results.
For these reasons, in my view, this work is not yet sufficient for publication and I would reconsider
the manuscript after major revisions.Minor issues:
4. In Abstract, lines 24 and 25, the text says ’[...] nucleation is dominated [...]’ and ’[...] nucleation
can still be dominated [...]’ while it should be ’dominant’.5. Line 70, ’concerned’ should be ’considered’ ?
6. Mistypes and inconsistent labeling of the instrument and datasets: sections 2.1: MEERA-2; 2.5:
HYSPIT.7. Missing satellite track, it should be added to the lon/lat maps. Vertical profile figures should have
double latitude + longitude axis on abscissa (same as provided by CALIPSO).8. The results section is riddled with ’Figure <n> shows [...]’ sentences. I would suggest rephrasing
them to ’As seen in Fig. <n>[...]’, ’Based on data shown in Fig. <n>[...]’, etc.9. Throughout the manuscript there are numerous grammatical errors (similar to the ones pointed out
above) and it should be very carefully revised.10. The authors have a significant number of self-citations that are present in addition to other previous
works, e.g. (He et al., 2021b, 2022b), (Jing et al. 2023). I would suggest removing them where
appropriate.11. The references (He et al., 2021a) and (He et al., 2022c) are not mentioned in the text at all. Please
make sure that all unused references are removed.12. The abstract provides a good summary, but is quite verbose and lengthy. If it is possible to shorten
it without affecting quality, it would be excellent.13. In Figure 8, the dust mass column density is for 2010-05-05, but the HYSPLIT trajectories span
from 2008-04-23 to 2008-04-28. Is it a mistake in the caption, or was wrong dataset used for the
figure?- AC4: 'Reply on RC4', Yun He, 08 Dec 2023
Status: closed
-
RC1: 'Comment on egusphere-2023-1844', Anonymous Referee #1, 03 Oct 2023
General Comments:
This is an interesting study that relates mineral dust plumes from Asian deserts to the number concentration of ice particles in cirrus clouds over Midway Island in the Central Pacific. It does this strictly through satellite remote sensing, using retrieval methods for mineral dust concentration (i.e., ice nucleating particle concentration or INPC) and ice particle number concentration (ICNC). It is well written and organized, but some of the arguments do not appear to be supported by the data, and the results may be overinterpreted. Specifics are given below. I recommend major revisions in rewriting the article, but this may not require a great deal of additional work.
Specific Comments:
1. Figure 4: The high column dust density measurements shown here in the back-trajectory do not ensure that air overlying these Asian deserts at ~ 10 km is having a relatively high concentration of mineral dust, since the column dust magnitude could result almost entirely from dust much below 10 km. This needs to be stated. On the other hand, it is commendable that the authors did this analysis since, although incomplete, it may be using all the available data and does provide important information.
2. Figure 5: There appears to be a problem with color legend. The plotting background is violet, which corresponds to an ice water content of 10-20 mg/m3. Not possible. The same problem occurs with Fig. 9c.
3. Lines 234-235: For what purpose are ICNCs shown for n_ice,25um and n_ice,100um? They provide no closure information with respect to dust INP concentration since most of the ICNC can be associated with D < 25 um (Kramer et al., 2009, ACP). Only n_ice,5um is relevant to the closure being sought with INPC.
4. Lines 236-237: Suggest citing Diao et al. (2015, JGR) to back up this statement.
5. Lines 252-253: Figure 6e shows agreement between ICNC and INPC only near cloud top for Si = 1.15, where ICNC is for n_ice,5um. From the previous page, it states that the layer average INPC is 7 L-1 and 96 L-1 for Si of 1.15 and 1.25, respectively. Since the layer average ICNC is 111 L-1 for Part B, optimal agreement is for Si = 1.25, with ICNC-to-INPC ratio of 1.16. Therefore, Si here should be 1.25, not 1.15.
6. Line 253: It is not clear how the ICNC-to-INPC ratio can be 0.9 for Si = 1.15, based on the previous text. As noted above, this ratio appears to be 1.16 for Si = 1.25, for which closure is optimal.
7. Lines 317-320: Since a large portion of the ICNC resides at D < 25 μm, n_ice,5um should be used rather than n_ice,25um (as assumed in this study). This practice (of using n_ice,5um) was followed for the other May 5th case study. In Fig. 10e, Si = 1.25 agrees best with n_ice,5um, consistent with the previous case study based on Comment 5 above.
8. Lines 308-312: If Si = 1.25 in the 2nd case study as indicated above, then the cirrus cloud of Part A could be produced only by heterogeneous ice nucleation since n_ice,5um = 635 L-1 and the mean dust-related INPC (U17-D) is 417 L-1. This is contrary to what the article states here; that Part A is dominated by homogeneous ice nucleation.
9. Lines 350-352: Alternatively, could this also be explained by variability in cloud updraft velocities?
10. Table 2: As mentioned earlier, for what purpose are ICNCs shown for n_ice,25um and n_ice,100um? They provide no closure information regarding dust INPC.
11. Lines 375-377: RHi (relative humidity with respect to ice) rarely reaches 140-150% in cirrus clouds since het (heterogeneous ice nucleation) always occurs before hom (homogeneous freezing nucleation), and INP and/or pre-existing ice tend to prevent the RHi from reaching the RHi threshold for hom. But if INP concentrations were low enough, the RHi threshold for hom would occur much more often to produce hom cirrus clouds, and their coverage could even exceed the coverage of het cirrus due to the smaller ice crystal sizes having lower fall speeds, as demonstrated in Mitchell et al. (2008, GRL). That is, lower ice sedimentation rates lead to longer cirrus lifetimes and greater cloud coverage. Moreover, the citation of Dekoutsidis et al. is misguided since that paper was showing hom is common in cirrus clouds and occurs mostly near cloud top where RHi is greatest, consistent with the modeling study by Spichtinger and Geirens (2009, ACP). The reference by Cziczo et al. does not support the author’s claim either; rather it argues that most cirrus are het cirrus.
12. Lines 377-380: While changes in UT INPC may alter the microphysical properties of cirrus clouds, this may not result in an increase in cloud cover and associated albedo for the reasons stated above. Moreover, the net radiative effect of cirrus clouds considers the absorption/emission of LW radiation in addition to SW radiation, and whether a net cooling or warming effect occurs may depend primarily on cloud optical thickness, the season, and the latitude.
Citation: https://doi.org/10.5194/egusphere-2023-1844-RC1 - AC1: 'Reply on RC1', Yun He, 08 Dec 2023
-
RC2: 'Comment on egusphere-2023-1844', Anonymous Referee #2, 13 Oct 2023
The manuscript discusses two interesting dust-influenced cirrus events over the Pacific Ocean in the outflow regime of Eastern Asia. The remote-sending-based study makes use of spaceborne lidar and radar observations. The work is original and worthwhile to be published in ACP. The main effort is related to closure studies, i.e., deals with the question to what extent the ice-nucleation particle concentration (INPC, estimated from lidar observations) is in the same range of estimated ice crystal number concentration (ICNC, estimated from combined lidar-radar observations). Based on these ICNC-vs-INPC closure studies the authors also discuss to what extent homogeneous ice nucleation was involved in the (in situ) cirrus evolution.
Many aspects are unclear, a lot of speculative argumentation is given in the main result sections 3 and 4. A more careful discussion is requested. The uncertainty in all the retrieval products needs to be better considered. Therefore, major revisions are required.
One of the key points for me is: All the retrieval products (INPC, ICNC) can only be obtained within a large uncertainty range of an order of magnitude. ICNC cannot be obtained with an uncertainty of 25% as mentioned in the present study! Impossible! Even if you have well calibrated lidar and well calibrated cloud radar observations, and, in addition, Doppler information about fall speed of ice crystals (and thus shape and size information) the uncertainty can never be lower than expressed by a factor of 3-5 around the retrieval products. This is, e.g., shown in the reference, you provide (Ansmann et al., ACP, 2019). This is also discussed in detail by Buehl et al., AMT, 2019. Now, in the case of the CloudSAT radar we do not have Doppler information. So, the ICNC uncertainty is even higher. All in all, the uncertainty margins are roughly an order of magnitude for INPC and ICNC. That needs to be considered in the closure studies presented here and in all the conclusions drawn from the observations. To my opinion, this will make the full discussion easier and more straight forward.
Details:
Title: This study is not a measurement report. This is an in-depth research approach (cirrus closure study) based on the analysis of spaceborne observations. The term ‘Measurement report’ suggests that just robust observations (measured data) are presented. But the study is widely based on estimated products and interpretation of the findings.
Abstract: The abstract needs to be rewritten after all necessary changes.
Introduction: I would keep the introduction as short as possible. There are so many cirrus papers and all present lengthy paragraphs on the radiative impact. I would suggest that the importance of cirrus in the atmospheric system is just briefly described, followed by gaps in our knowledge regarding cirrus evolution, and then what you are going to present in this study.
Lines 41-42: Note that homogeneous freezing also needs aerosol particles, however pure liquid ones, such as sulfate aerosol (without any insoluble part). In the case of heterogeneous ice nucleation, one needs particles with an insoluble fraction (sites for ice nucleation).
Figure1: Does MERRA also deliver dust profiles? or only column mass values? Would be nice to have model dust profiles up to cirrus level.
Method section:
Line 130: 25% uncertainty is unrealistic as discussed above. To use uncertainty margins of an order of magnitude (a factor of 3 around the retrieved value) makes sense. That is more realistic!
Lines 140-145: I would remove all immersion freezing parameterizations. The two case studies deal with cirrus from 9 to 11.2 km height, and temperatures from -40 to -54°C. Furthermore, your Table 2 considers deposition ice nucleation, only (no immersion freezing INP values are given). Even in virga zones with higher temperatures, there is ice saturation or even ice sub saturation (and especially sub saturation with respect to water). Immersion freezing is impossible in the presence of in situ cirrus with well developed virga structures.
Table 1: Are all the equations needed? A list of parameters and retrieval uncertainties makes sense. Again, please skip all immersion freezing parameterizations.
Result section:
The cirrus shown in Figure 3 is a classical in-situ cirrus, with a well defined cirrus top region, obviously above 10 km height, and a virga zone from 10 km down to 6 km height. The cirrus is clearly dust influenced.
Let me explain how such cirrus layers develop: Ice nucleation starts at cloud top, at the coldest point of the cloud, here the ice nucleation probability is highest. And if dust particles are present, they will trigger ice nucleation. Then diffusional growth of the crystals takes place, collision and aggregation. Sedimentation of ice crystals begins and the growth of the crystals leads to an immediate reduction in relative humidity (throughout the cirrus layer from 6 to 10.5 km height). There is almost no room for homogeneous ice nucleation (in the case of a well-developed cirrus system) because there is practically no potential to create a scenario with sufficiently high relative humidity over ice with values exceeding 150%, required for homogeneous ice nucleation. Homogeneous freezing is only possible at cloud top in the case of rather strong updrafts with large updraft speed so that super saturations levels develop faster than INPs can be activated (to reduce super saturation). All this is described in Kaercher et al. , JGR, 2022. However, such a scenario with rather strong updrafts is not visible in Figure 3. A very harmonic cirrus development exclusively controlled by the activation of dust INPs seems to be realistic. Furthermore, radiative cooling at cloud top of a well-developed cirrus system will cause some amount of downward motion and may contribute to a suppression of such rather strong updrafts. If we keep an uncertainty of about an order of magnitude into account for ICNC and INPC, all observations support that heterogeneous ice nucleation on dust particles dominated. Because of the large uncertainty, I would not further analyze the defined observational periods (part A and part B), i.e., explain the differences in the ICNC values in terms of heterogeneous vs homogeneous ice nucleations. The differences in the results for part A and B just show, to my opinion, the uncertainty in the products.
Back to the study:
Figure 3: The lidar is able to see the small ice crystals after nucleation with sizes of 1-5 micrometers at cloud top, above 10 km. The cloud radar seems to be not able to detect the ice crystals above 10 km height (Figure 5). The radar detects the ice crystals after diffusional growth and collision and aggregation processes, and after the start of sedimentation processes, i.e., several 100 m below cloud top, in the virga zone, as can be seen in Figure 5a and 5c. So, lidar-radar retrievals will not be able to exactly see the ice crystal number concentration of freshly nucleated ice crystals at cirrus top. And when radar comes into plays, aggregation took already place, and the number of crystals already decreased, may be already by a factor of 2, or even a factor of 3-5. Another point: Can we assume a ‘classical’ size distribution (as typically measured with aircraft instrumentation) in the case of freshly formed ice crystals? The size distribution is input in the DARDAR approach, and may be very narrow for the freshly nucleated crystals? All these unknown aspects cause the large uncertainty in the ICNC products (of at least one order of magnitude).
The ICNC values in Figure 5d vary strongly and indicate the uncertainty in all the retrieval products. Therefore I would not introduce part A and part B and ‘believe’ that the rather different findings are caused by different ice nucleation processes. One may formulate hypotheses…, but one needs to consider the large uncertainty in the discussion. Solid conclusions are difficult to draw. And as I mentioned, I am skeptical that homogeneous ice nucleation has a chance to occur in the presence of a well-developed cirrus. The CALIOP lidar indicates dust around the cirrus and no indication that the dust INP reservoir was depleted. To my opinion, a depletion of the the INP reservoir is unlikely during part A and B, and therefore homogeneous ice nucleation is unlikely.
If we keep the uncertainty of one order of magnitude in mind, the ICNC values for part A and B shown in Figure 6e nicely indicate the ICNC uncertainty range. To repeat, to my opinion, only heterogeneous ice nucleation makes sense. In the presence of so many rather favorite dust INP particles (as shown in Fig. 6c, 4000 large dust particles with a diameter > 500 nm per liter were present, and the corresponding dust surface area concentration was high with values up to 40 µm2 cm-3) homogeneous ice nucleation is rather unlikely.
Figure 6 indicates that INPC values for S-ice =1.15-1.25 (Ullrich parameterization, deposition ice nucleation) seem to be very likely (or reasonable) and match roughly the ICNC values (n-ice for crystals with sizes > 5 µm, for the periods of part A and B), when keeping in mind that ICNC is probably underestimated close to the cirrus top (because DARDAR values are not very trustworthy here because of too weak or even missing radar reflectivity values). The best DARDAR values are shown after significant growth of crystals and after aggregation processes (from 9-9.6 km height), but then the ICNC values are already reduced compared to the number of freshly nucleated crystals at cirrus top, probably reduced by a factor of 2-5.
Line 242: As mentioned above, homogeneous freezing in an environment with already existing ice crystals (and thus super saturation values S-ice around 1.0) is very unlikely. When keeping the high dust load and the large uncertainty margins into account, the closure is fine, ICNC and INPC match reasonably well. This is ok! This is a good result.
In summary here, please, do not compare part A and part B, just take the average of both periods and use the averaged values for comparison with the Ullrich results for INPC. Ice super saturation values of 1.15 are close to the values in the paper of Ansmann et al. (2019) for pure (unpolluted) dust scenarios. In the case of aged dust (or polluted dust, case study 2 in this study here) super saturations of 1.35 make sense. Such values are assumed to be realistic for aged, coated dust particles (see Kaercher et al, JGR, 2022). Homogeneous ice nucleation events do not make sense to me at all. However, I leave it open to you to find a proper and careful argumentation for the potential contribution of homogenous ice nucleation.
In the discussions (sections 3 and 4) there are many speculative aspects. Speculations are not justified. A more careful interpretation of the results is needed. And if you have a hypothesis, start with: We hypothesize…. and then the hypothesis must be based on convincing argumentation. And please keep the large uncertainties in mind.
Case study 2:
Another nice case with an impact of dust. In this case , CALIOP aerosol typing seem to indicated polluted or coated dust. The ice nucleation efficiency of aged and coated dust particles may be reduced by a factor of 5-10 compared to the ice activity for pure dust. This may or could be considered when computing INPC values with the Ullrich parameterization by multiplying the Ullrich INP values by 0.1-0.2.
Again, I would not compare the results for part A and part B because of the large uncertainties. The results in Figure 10e support my comment here. The results as a whole (part A and B) are in good agreement with the Ullrich INP values for S-ice of 1.35 when considering a factor of 0.1-0.2 less INPC (in the case of polluted or coated, less ice nucleation efficient dust). Kaercher et al., JGR, 2022, used S-ice values around 1.35 for the activation threshold for polluted dust. In case 2, the ICNC values (from the DARDAR approach) increase up to cloud top. Obviously, the radar reflectivity values were strong enough to obtain reasonable ICNC values even close to the nucleation range at cloud top.
To my opinion, there is again no room for homogeneous ice nucleation. There is dust ‘before’ and ‘after’ the cloud region, so a depletion of the dust INP reservoir is not visible. And at these conditions, homogeneous ice nucleation is unlikely.
Please avoid speculations on cirrus type, etc…. in sections 3 and 4. Just mention, what is really available from the observations.
Discussion section:
The first paragraph is not needed to my opinion.
Line 346: Can we have longitude-latitude information (not only latitude). How long (in km) was the cirrus layer? The same for case study 1.
Be careful with ‘dominating homogeneous freezing’ in environments with so much dust. It is simply difficult to produce high ice super saturation in the presence of favorable INPs.
If you follow my suggestions, youcan significantly ‘improve’ Table 2 by reducing the information content. I think ICNC for n-ice (> 5 µm) has to be compared to INPC, however information on n-25, n-100 is useful as well, especially to get a better feeling for the large uncertainties in all ICNC products.
In the case of the Ullrich INPC values, input is dust surface area concentration, S-ice as well as temperature! Temperature needs to be mentioned in Table 2 (Ullrich INP values).
Citation: https://doi.org/10.5194/egusphere-2023-1844-RC2 - AC2: 'Reply on RC2', Yun He, 08 Dec 2023
-
RC3: 'Comment on egusphere-2023-1844', Anonymous Referee #3, 18 Oct 2023
Referee report for the manuscript entitled:
“Measurement report: Influence of long-range transported dust on cirrus cloud formation over remote ocean: Case studies near Midway Island, Pacific”
Authored by Huijia Shen, Zhenping Yin, Yun He, Longlong Want, Yifan Zhan, Dongzhe JingIn this manuscript, remote sensing data has been used to determine whether long-range transported dust from Asia is active as INP in the formation of cirrus clouds. Data are collated from two space-borne instruments to calculate INP concentrations (INPC) and ice crystal number concentrations (ICNC). These values are compared at specific locations within an observed cirrus cloud that relate to where the dust would most likely be entrained. Where the values of INPC and ICNC are comparable, the manuscript concludes that this region is dominated by heterogeneous freezing due to the dust. Where the ICNC is much higher than the INPC, the manuscripts concludes that homogeneous freezing also takes place. Previous studies have shown that long-range transported dust acts as INP (Saharan dust in North America) and Asian dust has been shown to act as INP in China. This manuscript shows that Asian long-range transported dust also acts as INP. I am not an expert of INP, however having experience in adjacent aerosol fields, I am aware that there is still much to understand about INP sources and this manuscript confirms another significant source of dust and its far-reaching impacts.
When considering this manuscript for publication, there are several factors to take into account:
- This manuscript is presented as a measurement report, however the data are retrieved from space-borne instruments rather than field or lab work. Additionally, there is analysis and interpretation of the results which potentially goes beyond the requirements of a measurement report.
- This manuscript does not present novel ideas or methods, however it does establish evidence of long-range transported dust acting as INP in a new area.
- For me, there are some major questions about the analysis of the individual boxes within the cloud. It appears that one of the boxes closest to the dust plume is described as lacking in INP and, as such, homogeneous freezing is dominant. This description requires some justification.
- The methods would be enhanced by some more detailed description.
- Restructuring of some paragraphs and tightening up the language in a few key sentences would greatly aid the reading and understanding of the manuscript.
Based on the above summary and the comments below, I recommend this manuscript is reconsidered after major revisions. I also recommend that this should be submitted as a research article if the following suggestions make the manuscript a more substantial contribution. Note that I do not have expertise in satellite products so am unable to comment on their descriptions here and the suitability of their use.
Major points
On the description of the method, the key aspects from He et al (2022b) are mentioned throughout the manuscript, but it is quite scattered. I would suggest pulling this all together within the methods section so that the method is very clear for a reader who is unfamiliar with it. It is clear that if the ICNC is much higher than INPC then homogeneous freezing must be present, but it is not so clear how one knows that it is only heterogeneous if the ICNC is comparable with INPC. Why could it not still be both? I am not an expert in INP, but I think the key bit of information here is that the ice saturation will not reach the required 140 or 150% if heterogeneous freezing has already begun.
The methods could also contain a description of the two case studies before we dive into the results. In the discussion there is a nice point about the dust being from intense events where the dust is elevated by Mongolian anticyclones. Information like this would be better used in the beginning of the manuscript to set the scene. Is there other relevant meteorological information about the cases? Was there a particular reason for picking these two cases? What times and altitudes were used? Are the cirrus events very rare? A description of the two case studies would follow on nicely from the HYSPLIT model description in Section 2.
The results could be restructured to make the storyline clearer. For each case, I would recommend starting with establishing the case using the HYSPLIT trajectory and the dust mass column density plots. Then the description of the cloud and the dust plume using the CALIPSO cross-sections would follow on nicely. Next, part A and part B need a clear introduction using the DARDAR product. Why were these regions in particular picked? The difference in the ICNC between the two boxes could be described and then finally use the profiles to compare the ICNC values with the INPC. Here, I would suggest stating the ICNC versus the INPC and the ratio for part A and B. Do these values alone indicate homogeneous or heterogeneous freezing? The main method used in the manuscript is this quantification so I would pull it out of the data and emphasise it. Then add the additional, contextual information from other papers about typical values for each mechanism.
Table 2 shows the summary of ICNC and INPC for both cases split into the A and B boxes. Considering that the concentrations are defined as being “comparable” as within an order of magnitude, there does not seem to be a large distinction between the A and B boxes. This is particularly true for the first case where the ICNC values are quite similar for both boxes. For the second case, there does indeed seem to be a distinction between the boxes, however the INPC values still seem much higher than both boxes. These values would benefit from some clarification, and perhaps this could come in the results section if it were restructured as recommended above.
Minor points
The title could state the outcome of the study since an effect has been found by this study. For example, “Long-range transported Asian dust plumes influence cirrus formation over remote ocean.“
The abstract has a nice structure to it, but the sentence beginning “However, it is still not well…” (line 14) is confusing. This is a key sentence and rephrasing it would make the motivation of the research clear in the abstract. The results could also be stated more clearly, for example including the ICNC - INPC ratio.
The comment on line 47 about geoengineering comes across as quite random and a bit of an afterthought. Perhaps it could be rephrased to tie in with the previous sentence about the uncertainty and how understanding of these clouds needs to be improved in order for cloud seeding to be done appropriately.
The introduction has good content to give context to and justify the study. It does very well to lay out why we care about cirrus and this long-range dust transport. However a couple of paragraphs here could be restructured to improve the logic:
- The paragraph beginning at line 50 could be split into first discussing the pristine environment, low AOD, and the related lack of observations of cirrus. It is also a bit unclear whether there are not many cirrus or there are many but it is unexpected because of the pristine environment. Then the question of whether this means cirrus are purely formed by homogeneous could be framed.
- The ice saturation conditions for both mechanisms is a key part of the background science and this could be written in a separate paragraph. This could then link dust as an INP and give a description of why homogeneous freezing is much less likely to occur if dust is present (high ice saturation is prevented).
- The paragraph beginning at line 67 could also be restructured. From the studies cited here, I have understand that we know Saharan dust travels long distance, we know that Asian dust affects cirrus in China and also travels long distance to North America. So it seems to me that the question is not really whether the dust affects cirrus a long distance away, but rather is this long-distance transport creating a significant effect in an otherwise pristine environment? This paragraph could lead the reader more to this question about what is happening over the ocean.
- The paragraph beginning at line 83 could start with stating that He et al (2022b) determined the freezing mechanism by comparing INPC and ICNC before explaining why this works. Some clarification explaining how one knows that a mechanism is “dominant”, rather than there just being a mixture of both, would help the reader. There is good placement of the descriptions of the products here and nice summary with referencing.
The point about seeder feeder in stratocumulus clouds (line 80) could be made earlier because it is addressing the big question of why do we care about these cirrus? Perhaps it could be stated near the geoengineering comment as wider motivation for understanding cirrus.
On line 82, there is the phrase “it is indispensible” but it is unclear what it is. I am not sure if this line adds anything.
The overview of the paper (line 103) could be clearer and include “section 3”.
The manuscript tends to state what the figure is in the text, e.g. “Figure 5 shows the ice cloud properties, including cloud extinction coefficient, cloud particle effective radius, ice water content, 200 and ice crystal (with size <5 μm)....”. They could consider removing this from the main text since it is already described in the figure caption. This may make the story flow better as the main text can then go straight into what is observed in the figure, e.g. “Figure 5b shows the region of dust-related cirrus... ”.
The introduction to the results sectioned could be cut as there is some repetition that one INP is generating one ice crystal and that secondary ice productions is not being considered to take place in these conditions. But the statement of what is required for “good” agreement and the definition of dust-cirrus interaction event are well placed here.
The authors might consider adding an explanation for why the description of the dust in the main text based on figure 3a and b (line 184) does not fit with where the dust is identified in figure 3c and e.
In the main text about figure 5, the in-cloud averages are stated but there is no interpretation of these. In the introduction it is stated that homogeneous freezing produces more, smaller ice crystals. Do the in-box averages support the allocation of homogeneous/heterogeneous freezing based on the radius and ice water content?
Some clarification on line 236 about “pristine ice crystals” in abundant INP near the top of the cloud could help the reader. Are these pristine ice crystals because they are from homogeneous freezing or because they are formed in pristine environments, where the INP are more numerous at cloud top? Perhaps changing the description of the INP from abundant to more numerous would help.
In the discussions of figure 6e (and for 10e), it might not be clear to the reader how the calculations of INPC at different ice saturation ratios relate to the ICNC values. Does each saturation ratio relate to a different radius? A statement of which INPC and ICNC values are actually being compared for both part A and part B would make it clear to readers like me that are unfamiliar with the POLIPHON method.
For part B starting at line 244, the message would be clearer if the comparison with the INPC came before putting it in the context of Ansmann et al (2019a), Cziczo et al (2013) and the interpretation of the dust acting as INP in the moist region. Additionally if this is the moist region shown by the RH in figure 6d, this could be linked with a “as shown in figure 6d”.
In the paragraph beginning on line 346 with “The overview of …. “, the authors might consider moving the sentences about the dust being uneven and the lack of INP resulting in homogeneous freezing near the start of the paragraph. They could then summarise the findings from part A and B for each case.
Related to the above point, a comment on why part B is shown to have lacking INP but is closest to the region of dust is perhaps warranted.
Line 375 states that cirrus would not form without these natural INP, but this conflicts with the establishment that homogeneous freezing does take appear to take place. Perhaps change to something similar to “would rarely form” or “there would be a much lower frequency of clouds”.
The authors have done well to suggest plenty of future work based on this study.
Technical points
There is some inconsistency between using the terms secondary ice nucleation and secondary ice production in lines 84 and 174.
Blue and purple colours in the profile plots (figures 6e and 10e) are quite hard to distinguish. These could be more contrasting colours.
In figure captions, the authors could consider separating out figure titles from the a), b) descriptions. For example, the figure 2 caption could be “Dust-related conversion…” and then a) and b) to describe each plot. Also in figure 6 and 10.
On line 164, what does a “complete system” mean here?
Some typos:
- The abstract contains some use of the word “dominated” rather than dominant. This might be present in other places too.
- Paragraph starting at line 164 contains several cases of HYSPIT instead of HYSPLIT.
- Line 63, what is the occurrence rate of cirrus measured in?
- Line 73: “Saseen et al 2003” should be Sassen
- Line 75: should “Sassen et al 2001” be 2002?
- Line 119: “MEERA” should be MERRA
- Line 169: “stimulated” → simulated
Citation: https://doi.org/10.5194/egusphere-2023-1844-RC3 - AC3: 'Reply on RC3', Yun He, 08 Dec 2023
-
RC4: 'Comment on egusphere-2023-1844', Anonymous Referee #4, 23 Oct 2023
The manuscript by Shen et al. presents an observational study of cirrus cloud formation near Midway
Island in Pacific Ocean and the ice-nucleation properties of long-range transported desert dust. Authors
have used observational data from Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) and
Cloud Profiling Radar (CPR), DARDAR and MERRA-2 datasets, and POLIPHON and HYSPLIT mod-
els for the data evaluation. As is, the manuscript addresses an important topic (especially for atmospheric
and climate modeling communities) – the impact of long-range transport dust on the formation of cirrus
clouds – but does not present an answer. Before the paper is published, some major concerns would have
to be addressed.My greatest concern with the manuscript is the lack of indepth analysis – additional details (such
as the nucleation mechanism, connection between the initial nucleation conditions and lab experiments)
would improve the manuscript significantly. For this reason, these Major issues in the manuscript should
be addressed:1. The idea and dataset chosen to show the nucleation and cloud formation are great, but the data
analysis seems to be focused on an observation. It would be very interesting to see the atmospheric
conditions data and the discussion of the initial heterogeneous ice nucleation (e.g., the transition
from pure dust to pure cloud at 30.34/0.0 lat/lon on 2010-05-05 trajectory) to see the relation to
the laboratory experiments (such as Koehler et al., Atmos. Chem. Phys., 10, 11955–11968, 2010).2. The manuscript is presented as a Measurement Report, however it is not a new or original measure-
ment, but rather a reanalysis of an old and public dataset. As a result, the manuscript does not
provide substantial insight and conclusions. I would suggest the authors to perform and write-up
an in-depth analysis of the processes based on their expertise and reconsider the manuscript as a
research article.3. Another major shortcoming of the article is the Results section, where most of the section is
focused on spelling out the results seen in figures with little interpretation. It is good to see that
the summary of the results has been provided in Table 2, in the Discussion section, but I would
expect it as part of Results.
For these reasons, in my view, this work is not yet sufficient for publication and I would reconsider
the manuscript after major revisions.Minor issues:
4. In Abstract, lines 24 and 25, the text says ’[...] nucleation is dominated [...]’ and ’[...] nucleation
can still be dominated [...]’ while it should be ’dominant’.5. Line 70, ’concerned’ should be ’considered’ ?
6. Mistypes and inconsistent labeling of the instrument and datasets: sections 2.1: MEERA-2; 2.5:
HYSPIT.7. Missing satellite track, it should be added to the lon/lat maps. Vertical profile figures should have
double latitude + longitude axis on abscissa (same as provided by CALIPSO).8. The results section is riddled with ’Figure <n> shows [...]’ sentences. I would suggest rephrasing
them to ’As seen in Fig. <n>[...]’, ’Based on data shown in Fig. <n>[...]’, etc.9. Throughout the manuscript there are numerous grammatical errors (similar to the ones pointed out
above) and it should be very carefully revised.10. The authors have a significant number of self-citations that are present in addition to other previous
works, e.g. (He et al., 2021b, 2022b), (Jing et al. 2023). I would suggest removing them where
appropriate.11. The references (He et al., 2021a) and (He et al., 2022c) are not mentioned in the text at all. Please
make sure that all unused references are removed.12. The abstract provides a good summary, but is quite verbose and lengthy. If it is possible to shorten
it without affecting quality, it would be excellent.13. In Figure 8, the dust mass column density is for 2010-05-05, but the HYSPLIT trajectories span
from 2008-04-23 to 2008-04-28. Is it a mistake in the caption, or was wrong dataset used for the
figure?- AC4: 'Reply on RC4', Yun He, 08 Dec 2023
Data sets
Cloud–Aerosol Lidar and Infrared Pathfinder Satellite Observation data base CALIPSO https://subset.larc.nasa.gov
Cloud properties combining the CloudSat radar and the CALIPSO lidar measurment from raDAR/liDAR data base DARDAR https://www.icare.univ-lille.fr
MERRA-2 inst3_3d_aer_Nv: 3d,3-Hourly, Instantaneous, Model-Level, 490 Assimilation, Aerosol Mixing Ratio V5.12.4 Global Modeling and Assimilation Office (GMAO) (2015) https://doi.org/10.5067/LTVB4GPCOTK2
AERONET Aerosol Inversion (V3) database, Aerosol Robotic Network [data set] AERONET https://aeronet.gsfc.nasa.gov/new_web/data.html
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 683 | 208 | 48 | 939 | 26 | 26 |
- HTML: 683
- PDF: 208
- XML: 48
- Total: 939
- BibTeX: 26
- EndNote: 26
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1