Review of “Tropical cirrus evolution in a km-scale model with improved ice microphysics” by Gasparini et al. 

This review was produced by a co-review team, as agreed with the editor. 

General comments

The authors study tropical cirrus using an idealised simulation with a convection-permitting model (SAM), multi-campaign aircraft dataset (POSIDEN and ATTREX), satellite data (DARDAR, CCCM and 2C-ICE), and a selection of global-storm resolving models (GSRMs). It analyses these data from a variety of angles, including the use of atmospheric tracers, power spectra, standard deviation of vertical velocity, cloud radiative effect, ice water content (IWC), ice particle concentration and ice particle radius. The tracer analysis in particular focuses on the distinction of cirrus detrained from convective anvil cloud opposed to cirrus formed in-situ. Following this analysis, the authors conclude that in-situ cirrus prevails at colder temperatures and low IWC, while anvil cirrus prevails at warmer temperatures above -60C. The radiative analysis highlights the anvil cirrus as particularly important, but the authors suggest in-situ cirrus is still important in this regard. The authors explore model resolution and show that the updraught power spectra (which they consider important for cirrus formation) only becomes accurate around 250m grid spacing. They achieve better agreement between the model and observations with their proposed methods without the need to increase the computational cost. 

It is great to see multi-campaign aircraft data exploited for improving model performance, and it is great to see some microphysical model development with computational resource in mind. There are many interesting results in this paper and, while they jump around different aspects of cirrus research, we think there is enough coherence for it to make a paper. However, if trying to cover a lot of ground, a paper needs a structure that helps the reader. We don’t think that is quite the case here. The least subjective aspect of this is the presentation of model improvement content after the presentation of scientific model results in relation to the tracers (which must have been based on the model improvements). We cannot understand the logic in this, so we suggest moving section 3 after at least section 4.1 but better would be after the whole of section 4. Other than this, we take issue with a few statements by the authors and believe there are few possible improvements, but once that has been considered we hope to see the manuscript reach publication.

Major comments

L9/L361 – The result suggests that in-situ, thinner clouds are not radiatively important. However, the authors caveat in both the abstract and in the text that such cloud must still be considered when looking at such radiation. The manuscript does not show any results to back this up, and I cannot see any argument made for why that is the case. Why “must” the in-situ CRE be considered? In what way does the thinner cirrus “contribute meaningfully”? 

L270 – “cut the microphysical bias in half ” – I cannot see what this is based upon. In Figure 5, the small particle bias metric goes from 0.92 to 0.65, and the large particle bias metric goes from 0.52 to 0.4. How is this halving? Perhaps you need to give the formula for the metric or explain it more. We don’t fully understand what it is.

L276 – CC applies to the saturation vapour pressure. Whilst I can understand that it plays a role in moisture with height in a stable atmosphere, the connection to anvil cirrus seems much more complicated, as the available moisture at each height is a result of CAPE and low-level humidity. In fact, I might naively assume that because the saturation vapour pressure reduces with height (i.e. temperature), one should expect increasing condensate as moisture is lifted as it cannot remain as vapour. Are you drawing this statement of CC yourself? In which case, please explain the logical steps in greater detail. If this is something you consider to be widely accepted, please point to a reference or two that explains the mechanistic connection between IWC vertical profile and CC.

Fig9 – The GSRM data has not been described. Please add a description to the data section.

Fig 10 and L366 – I strongly object to the illustration on fig10 and the suggestion of any analysis “split into two separate evolution pathways”. Until this point, the paper was doing a brilliant job of both simplifying the problem whilst maintaining necessary complexity (e.g. Figure 3c). However, as far as I can tell, you have not used the tracers in the analysis of CRE. This is unfortunate, and I would strongly encourage to break the CRE analysis down by tracer if possible, as it would be fascinating. However, you have instead subjectively drawn on clouds to Figure 10. These are misleading because there is not such a simple split. Your own Figure 3c shows the large amount of overlap in the central part of the IWP spectrum. Whilst you could adjust the illustration to show that in-situ and anvil cirrus occur in the same place, how will you illustrate the dual-origin component? I often appreciate diagrams, but I think it is necessary to remove this as I don’t think it adequately reflects the value of your results.

L367 – Fig 3c shows similar amounts of anvil cirrus all the way to 0.01 g m-2. This is why I think your illustration is misleading (as described above). It makes one think there is this clear distinction, but there is not. Just because in-situ cirrus dominates <1gm-2 cloud doesn’t mean that a major fraction of anvil cirrus is not spreading to that thinness. 

Minor comments

L49 – In the outline of the paper you first say you demonstrate modification and then look at tracers. I suggest this is a more logical ordering for your sections than you actually have.

Sec 2.1.2 – Possibly it would be good to further subsection this so each modification is clearly labelled as it would make it easier to refer back to the relevant text in fig5 caption.

L167- They use nighttime data only because the lidar signal is noisier during the daytime. Is this representative for the simulation? 

Fig1 – Add description of solid and dashed black lines.

Sec 2.2 – How long is the simulation and how much of that has been used in the results?

Sec 2.3 and 2.4 – Maybe sections 2.3 and 2.4 would be better grouped under a “Data” section. The title for section 2.3 could be shortened to “In-situ observations”. In section 2.4, there’s no need for subsections, it’s clear enough if each of the datasets is presented in separate paragraphs.

Fig 2 – Could a second, later time be shown once convection has subsided? It would be useful to see how such cloud and tracer categorisation progresses after convection finishes but high cloud remains. It would be fine if only panel d was shown for the later time.

L219 – “excludes portions” – what portion is excluded? I don’t recall that being described.

Fig 3 – What’s the length of the time period used to generate this figure?

L225 – dual-origin “dominates” in the middle range is probably worth being explicit about. It looks to be the range should be 1-40 where dual-origin is the dominant category, not 1-10?

Fig 4 – “omit that panel” – but then you include the dashed contour on panel e. I would suggest you should be either including panel j, or omitting the dashed contour from panel e.

L245 – “converging” – It is not clear to me that they are converging. Similar overlap of dashed and solid contours in both 3-4h and 7-8h. What is this statement based on? 

Fig9 – Is the tailing off of both model an observation from the 5/3 slope robust or a result of limited data? If the latter, consider showing uncertainty on the plot or setting a threshold for what parts of the lines are robust and to be plotted.

L413-414 - “We also perform a short sensitivity test…” What does short mean in this context? This explanation would be better suited in the methods.

L415 – (not shown) Can it be added to the appendix? 

Fig5 caption – It would be helpful to reference relevant text for each modification.

Conclusions – The first paragraph says about tracers first but then the following text describes model improvement results first. That’s fine but it highlights that probably the model improvements results should come before tracer results in the main text too?

Technical corrections

L313 – “Of particular significance are the high-frequency fluctuations”. I'd change the order of the sentence: The high-frequency fluctuations are particularly significant...