Review of “How does the lifetime of detrained cirrus impact the high cloud radiative effect in the tropics?”

By George Horner and Edward Gryspeerdt

Recommendation: Major revisions

This paper addresses two questions about tropical cirrus: (1) how does the cloud radiative effect (CRE) of cirrus from land convection differ from oceanic convection, and (2) how much does changing the lifetime of detrained cirrus impact the overall tropical CRE. Both questions are addressed with a unique cloud tracking algorithm that sorts ISCCP cirrus by their origin (land vs ocean and detrained vs in-situ) and lifetime. Question 2 is additionally addressed by artificially modifying the lifetimes of detrained cirrus, which increases their statistical weight (relative to in-situ cirrus) when computing the average CRE. Answering both of these questions would provide important constraints on how important land-ocean contrasts in convection and lifetime changes in cirrus are for the TOA budget. I think the paper’s methods are sufficient to answer Question 1, but less sufficient to answer Question 2, because it is unclear how representative their idealized calculation of lifetime extension is of a meteorologically- or anthropogenically-driven change in cirrus lifetime. 

For this reason, I recommend major revisions. I provide more major comments below which, if addressed, I would then be happy to recommend this paper for publication. These comments are followed by more minor points below.

Major comments

I believe the paper’s methods are sufficient to answer Question 1, which in and of itself should merit the paper’s eventual publication because it resolves the land/ocean contrast question that another recently published anvil/cirrus tracking study could not (Jones et al, 2024), and because it touches on questions highlighted in the literature, for instance, how much the timing of convection impacts CRE (Gasparini et al, 2022). 

However, I think the paper’s methods are less convincing in answering Question 2, because it is unclear whether increasing the statistical weight of detrained cirrus and calculating the resulting averaged CRE is equivalent to the CRE that would result from a change in cirrus lifetime due to meteorological or anthropogenic factors. For instance, the lifetime could change due to stronger clustering of convective cores within each anvil (Jones et al, 2024); increased updrafts via aerosol invigoration (Abbot and Cronin 2021), or diminished sedimentation in detrained cirrus (Beydoun et al, 2021), and I could imagine that each pathway would impact CRE differently from each other and from the idealized calculation presented in this paper. 

What I think the authors have better constrained is the impact on tropical cirrus CRE that would result from a redistribution of cirrus from in-situ to detrained. The authors could perhaps rephrase their Question 2 to something like “How much does changing the relative abundance of in-situ vs detrained cirrus impact the tropically averaged CRE?”. Or, if they stick to their original phrasing, then they should provide additional analysis, or additional discussion at the very least, of how their method of extending cirrus lifetime and computing CRE is representative of how meteorologically- or anthropogenically-driven changes in lifetime would impact CRE.

Minor comments

• The captions of Figure 4 and Figure 9 should be switched with one another
• I thought the final paragraph of the introduction nicely sets up the rest of the paper. However, the rest of the introduction could be written more succinctly to help propel the reader to the questions that this paper will address. For instance, Lines 31 - 36 could be rewritten as “Cirrus clouds cover approximately 60-80% of the tropics (refs), with about half being formed in-situ and the other half from detrainment (refs).” Lines 75 - 82 could be shortened in a similar way. I encourage the authors to prune the introduction and keep its scope as focused on the two research questions as possible. 
• Line 124: What is the physical reasoning behind choosing a 10% threshold?
• Line 327: You have found that a 50% or 15 hour increase in detrained cloud lifetime results in an increase in the overall high cloud CRE by about 0.6 W/m^2. It would be interesting to know how much cloud lifetime is expected to increase due to, say, the aerosol invigoration hypothesis. If the expected increase in lifetime is much smaller than 50%, then you could say that aerosol invigoration might not matter all that much in terms of its impact on CRE. I think that making these quick assessments with all of the proposed mechanisms that change cirrus lifetime, by connecting to the wider literature, would help make readers care more about your results. And it would illustrate how your result “provides an important constraint on the impact of changes in the lifetime of detrained cirrus in a future climate or in response to aerosol perturbations on the total tropical CRE.”
• This manuscript, either in the introduction or in the conclusion, could mention how it distinguishes itself from other recent papers using cloud tracking of anvil/cirrus systems (e.g. Jones et al, 2024). For instance, the observations used in this manuscript have a longer time record and cover the whole tropics, which allows regional variations such as land/sea contrasts to be addressed. 

References

• Abbot and Cronin, 2021 Aerosol invigoration of atmospheric convection through increases in humidity
• Beydoun et al, 2021 Dissecting Anvil Cloud Response to Sea Surface Warming
• Gasparini et al, 2022 Diurnal differences in tropical maritime anvil cloud evolution
• Jones et al, 2024 A Lagrangian perspective on the lifecycle and cloud radiative effect of deep convective clouds over Africa

Review of “How does the lifetime of detrained cirrus impact the high cloud radiative effect in the tropics?”

General comments:

In “How does the lifetime of detrained cirrus impact the high cloud radiative effect in the tropics?”, Horner and Gryspeerdt present compelling new observational evidence on the importance of detrained cirrus lifetime and the diurnal to the tropical high cloud radiative effect. Understanding the lifecycles of both detrained anvil cirrus and in situ cirrus and how these may respond to climate change is a major focus of present research, and so I consider this work highly relevant for publication. The Lagrangian trajectory method developed for use here allows investigation of the lifetimes of detrained and in situ cirrus to an extent not possible through other observational techniques such as cloud tracking. In addition, the long time span of the dataset (exceeding 30 years) provides a great wealth of observational data to analyse.

I found the manuscript in general to be very well written and presented and interesting to read. However, I found that the “lifetime extension” experiment presented in section 3.5 was much harder to parse. From my understanding, this approach investigates how the net high cloud CRE would change if the proportion of long lived detrained cirrus was to increase. This is not quite the same as uniformly increasing the lifetime for all detrained cirrus, as the present day long-lived cirrus is more likely to result from different conditions such as more intense or organised convection, and may have colder temperatures and occur at different times of day, introducing other factors that affect the net CRE other than the lifetime. This is still a very valid to question to ask, as warming has been hypothesised to lead to an increase in both more intense and more organised convection. I think it would help to rephrase the lifetime extension experiment in these terms.

The main areas I found that were lacking in the manuscript was a discussion of the mechanisms affecting cirrus lifetime, including radiative effects and cloud height. The authors make a good argument for the importance of the diurnal cycle on the net high cloud CRE and the difference between land and ocean origin convection due to differences in the CRE near the start of the trajectories. Figure 7 appears to show an impact of the diurnal cycle on the lifetime of the detrained cirrus as well, which would provide observational evidence for processes previously investigated using models. Further, detailed discussion of the differences seen in the lifetimes of detrained cirrus, including the differences between land and ocean as well as the change in the net CRE along observed trajectories vs the detrained cirrus lifetime would greatly add the to manuscript. I also feel that contributions that the authors make contributions to understanding the behaviour of detrained cirrus are more important and novel than they present them in the paper!

Overall, I recommend that the manuscript undergo major revisions. My suggestion to the authors, should they wish, is that the “lifetime experiment” section of this manuscript be split and expanded into a separate paper with a clearer hypothesis and objective, which could, for example, be published in a letters format. That would also leave more space in this paper for further discussion of the observed differences in detrained cirrus lifetimes and their CRE. However, I fully understand if the authors do not wish to split this manuscript.

I have included full comments below. Many of these suggestions may be more suitable for further work than inclusion in this manuscript, but reading through raised many interesting questions that I think would make valuable future work.

Specific comments and suggestions:

Line 4: It wasn’t clear to me that hypothetical changes in anvil lifetime has been discussed in the introduction. While there are many factors that influence the decay rate and hence lifetime of in situ cirrus, I am not aware of a clear hypothesis that explains the mechanisms through which detrained cirrus lifetime would change with global warming

Line 20: Possibly not the best reference for this? I’m not sure that this needs referencing

Line 21: radiative effect, rather than forcing

Line 49: While both detrained cirrus and in situ cirrus are affected by changes in deep convection, I am not sure there is a clear direct link between the two types of cirrus

Line 54: It may be helpful to include that albedo reduces faster than LW emissivity, resulting in the switch from net cooling to net warming as cloud thickness decreases (e.g. Berry and Mace, 2014)

Line 67: Possibly think about combining these two paragraphs + highlight the importance of lifetime along with diurnal cycle. e.g. if the timing of convection remains the same, changes in lifetime may result in more cirrus existing at night or day, which could change the net cirrus CRE without any changes to the optical properties of the cirrus.

Line 75: Changes in anvil cloud height and area are general seen as separate (albeit opposing) feedbacks, as per e.g. Sherwood et al 2020

Line 78: The original proposed precipitation efficiency iris feedback (Lindzen et al., 2001) is generally no longer considered valid. I would focus discussion here on more recent discussion of evidence for and against the stability iris feedback, e.g. hypothesised suppression (e.g. Jeevanjee 2022) and enhancement (e.g. Seeley and Romps, 2015) of convection with warming.

Line 80: This is only true if the advection of anvil cirrus remains the same. However, the stability iris hypothesis is based on a reduction in the large scale overturning circulation, which would mean that for the same (Lagrangian) lifetime of cirrus, the anvil would cover a smaller area

Line 82: It may be good to include discussion of the findings of Raghuraman et al. 2024 on that changes in tropical anvil cirrus properties

Line 86: Average anvil lifetimes are very sensitive to the choice of method, thresholds for cirrus detection and the type of convection studied. Older studies tended to detect less isolated convection and more MCSs so tended to be biased towards larger, longer lived anvils. It may be good to discuss some of the factors driving this, such as the difference in spatial and temporal resolution between the two studies, to put your own results into context

Line 87: Beydoun et al. 2021 similarly found negligible change in anvil lifetime in RCEMIP-style models, which may make a useful second point of reference. However, these models do not necessarily represent processes affecting cirrus lifetime accurately (e.g. lack of diurnal cycle, ocean only) so more research is needed.

Line 90: Herbert and Stier 2023 is a good reference for linking aerosol effects on convection to anvil CRE. However, I would argue that we still don’t have a good understanding of how changes in convective intensity affect anvil properties.

Line 102: One area of discussion missing from the introduction is factors affecting the lifetime and decay of cirrus clouds. Both solar heating (and hence diurnal cycle dependence), such as Gasparini et al. 2022, Sokol and Hartmann 2020, and sublimation/sedimentation of ice clouds e.g. Seeley et al. 2019

Line 106: ISCCP-H data is every three hours, but the ERA5 winds and TSC step is hourly. Are new convective events defined only every three hours, or is some sort of interpolation used to increase the temporal resolution of the ISCCP data?

Line 110: Is there much variance in the wind field in this pressure range? Would this cause much uncertainty in the Lagrangian trajectories?

Line 110: May be clearer to move this refence to immediately after “ERA5 reanalysis” to ensure that it refers to the data source, not the method

Line 114: Suggestion: CERES data has not yet been discussed, so it might be clearer to rephrase along the lines of “…any results involving CRE only cover the period after 2000 when CERES was operational”. Also, does the CERES acronym need to be defined?

Line 117: This is also known as “liquid-origin cirrus” e.g. Luebke et al. 2016, and it may be helpful to make the link here.

Line 119: This last sentence is a little unclear, and could be rephrased to be clearer e.g. “In situ cirrus are those which appear along the trajectory after a period of clear skies”

Line 121: How sensitive is the detrained/in situ divide to the cloud fraction threshold? Other studies have used 20%, would using this threshold have notably changed the results?

Figure 1: Is the annual mean for 2008?

Line 125: It would be of interest, for future studies, to include a third “cloud free” flag in the algorithm, which might allow more investigation into the formation and persistence of in situ cirrus that forms in airmasses diverged from convection

Line 154: ISCCP cloud type histograms have previously been shown to over-attribute toa cooling to low clouds (see e.g. Figure 5 of Stephens et al. 2018). This may lead to a positive (waming) bias to your results, which would be good to discuss.

Line 155: How is the monthly mean calculated? As the high cloud fraction varies with the diurnal cycle, in particular over land, simply averaging all points with low high cloud fraction could bias the sampling to certain times of day which would in turn affect the SW mean. Binning observations by local time, calculating the mean for each time bin over the month then averaging all could reduce bias.

Figure 2: Am I correct in thinking that fig. 2 a and b show ToA SW flux, not CRE?

Line 169: In what way do they disappear in the annual mean? Would they not have a positive bias to the measured CRE? What happens if you set all positive high cloud SW CRE values to 0?

Line 170: While I agree that the low cloud LW CRE will be small, so is the net high cloud CRE you find later in the paper, and so this bias could be significant. How large is the difference in the background flux if you include low cloud CRE as with the SW fluxes?

Line 191: What is the standard deviation of the detrained cirrus lifetime? It would be interesting to compare the variance in lifetime vs that seen by Luo and Rossow, 2004

Figure 3: There appear to be a lot of trajectories that are very short (only one time step?). Do these occur in regions with very frequent convection, so that existing trajectories get replaced by newly initialised ones very rapidly?

Figure 3: What causes the scallop-like pattern along the lines? Is it linked to the 3-hourly resolution of the ISCCP data?

Line 194: The description of the weighted average is a little unclear, it may be better to rephrase along the lines of “…CRE for detrained, in situ and the net of all air parcels”

Line 194: Are these averages weighted by the area of each pixel?

Line 199: While the number of in situ parcels early on is low, there does seem to be a clear trend in their CRE. Perhaps many of these cases are caused by mid-tropospheric clouds (e.g. congestus) which don’t classify as high cloud by the ISCCP definitions and so have low high-cloud fraction, have lower LW CRE but still have larger SW CRE

Line 211: This should be referred to as a radiative effect, not a forcing.

Line 211: It would be helpful to provide separate values for the detrained and in situ cirrus lifetime CRE to provide perspective for the lifetime experiments of section 3.5

Figure 4: It might be useful to combine figs. 3 & 4 as panels within one figure, to show that the early net CRE is mainly due to detrained cirrus, and the later CRE more due to in situ cirrus

Figure 4/5: It would be interesting to show variance of the CRE in figs. 4 and 5, possibly using the interannual variability as done in fig. 6

Section 3.3: This subsection could possibly be combined into 3.1

Line 214 typo: figure 1(d) -> 1(c)

Line 221: Does it hide details? I would rephrase it as showing different things, fig. 1f shows where land origin air parcels are most commonly seen over oceans (and vice versa) with the SE Atlantic off the coast of Africa being the most common, but fig. 1c highlights that these instances occur everywhere and can travel a long way from the coast

Line 226: It would be useful to provide a value for the proportion of oceanic/land origin parcels

Line 228: Is the difference in surface albedo between land and ocean may also be a cause of these differences. It would be interesting to separate land-origin parcels that are seen over the ocean (and vice versa) to try and isolate some of these effects

Line 231: I would also mention that while oceanic convection does tend to be more uniform, it does have a peak in the early morning which would add further weight to your contrasting diurnal cycle argument

Figure 6: How is the DCC count normalised? The values here are a little confusing. It may be clearer to plot the y axis as a fractional frequency/proportion

Line 240: The land-origin parcels also have larger SW cooling over their lifetime, which could indicate that they are optically thicker than ocean-origin

Section 3.5: It was quite difficult to understand the process throughout this section. It may be clearer to lay it out as a full experiment, with some background as to the causes of changes in cirrus lifetime in a future climate, a hypothesis for how you expect it to affect high cloud CRE, etc. although then it is becoming more of a paper on its own.

Line 247: The discussion here of the detrained lifetime is very short, and I think some important findings of this paper have been rushed over. The difference between land and ocean lifetime shown in fig. 7 is very interesting, and could do with more discussion either here or in section 3.1

Line 248: Are only parcels which are observed to transition from detrained to in situ included in the analysis here? Do these parcels have a different net CRE to those which don’t undergo a transition to in situ cirrus before being replaced by new convection?

Line 250: This could be rephrased to make it clearer that this lifetime modification technique is the original development of the authors.

Figure 7: Mean is possibly not the best statistic to use here, given the skewed distributions. It could be informative both the provide the mean and the peak lifetimes for both

Figure 7: Both the land and ocean distributions have the same shape, with an interesting oscillation in the rate of decay, but with different phase. Is this further evidence of a diurnal cycle effect on the lifetime of cirrus, with daytime lofting/nightime decay? This would provide further evidence for the impact of land vs ocean origin for detrained cirrus impacts. Would be interesting to explore further in future

Figure 7: There are a number of very short trajectories which look to be anomalous. Would removing these have a noticeable impact on the results?

Line 256: The wording of this is slightly unclear, as the radiative properties along each individual trajectory are not modified. The lifetime extension method appears to me to be asking “what if we saw an increase in the occurrence of long-lived detrained cirrus and a corresponding decrease in in situ cirrus”. It might be clearer to rephrase the purpose of the lifetime experiment in these terms.

Line 260: I don’t think that this approach is entirely analogous. The proposed lifetime extension approach increases the weighting given to longer-lived detrained cirrus at the expense of in-situ cirrus. As these longer lived cirrus are likely to belong to a distinct population (e.g. more organised and more intense convection, and may occur at different times of day…) this may have different results to stretching the observed properties included CRE along the TSC axis. I would expect that longer lived detrained cirrus tends to occur at higher altitudes and therefore has a more warming LW CRE than shorter lived cirrus at a lower altitude

Line 263: It would be interesting to look deeper into how the lifetime of the observed cirrus trajectories relates to their properties. E.g. do shorter lived detrained cirrus have different average CRE to longer lived cirrus? Either over just the detrained cirrus lifetime or the entire duration of the trajectory.

Figure 9: The change in CRE doesn’t appear to be linear with the lifetime extension, it would be good to discuss this in the text

Line 274: How sensitive are these values to changes in the previously calculated lifetime net CRE for detrained and in situ cirrus? I expect that uncertainty in these values may cause a larger uncertainty in the lifetime extension CRE than the interannual variability

Figure 10: Would it be clearer to show 50% instead of 25%?

Line 304: This is a little difficult to understand. Do you mean to say that this situation occurs in regions where the average low cloud cover is high (e.g. oceanic cold pools), and hence the average background albedo is also high. If a trajectory is over this location with an optically thin high cloud and no/little low cloud then the observed albedo would be lower than the average background, hence resulting in a “negative” high cloud SW CRE?

Line 305: This assumes that there is no correlation between high and low cloud cover. I am not sure whether or not this is the case, however.

Line 314: It would be interesting to see how changing the threshold for detrained vs in situ changes the CRE for each, possibly as supplementary materials

Line 334: Is this such a simple statement to make? From a traditional view, if the average CRE of anvils is 0, then the lifetime should not matter… However, if the lifetime does not change proportionally (e.g. the cirrus decays more slowly at lower optical thicknesses than at higher), or there is a coupling with the variability of convection across the diurnal cycle (which you have shown is the case over land in this paper) then it may be important. I think you could make a much stronger argument from the results which you have presented here in favour of the anvil lifetime having an important role in the net anvil CRE, and that the impact of changes in anvil lifetime may be missed in previous studies such as those using RCEMIP style models with no diurnal cycle, or studies focusing on the tropical oceans where the diurnal cycle of convection is much weaker.

Technical corrections:

Line 169: typo disappear -> disappear

Line 195: acronym; time since convection -> TSC

Figure 4/9 captions swapped

Line 240: do -> due

Figure 10: 0 value is not aligned between the two axes

Line 329: This sentence is duplicated from line 327

Line 362: Extraneous “9” at the end of the sentence

Suggested references:

Beydoun, H., Caldwell, P. M., Hannah, W. M., & Donahue, A. S. (2021). Dissecting Anvil Cloud Response to Sea Surface Warming. Geophysical Research Letters, 48(15), e2021GL094049. https://doi.org/10.1029/2021GL094049

Gasparini, B., Sokol, A. B., Wall, C. J., Hartmann, D. L., & Blossey, P. N. (2022). Diurnal Differences in Tropical Maritime Anvil Cloud Evolution. Journal of Climate, 35(5), 1655–1677. https://doi.org/10.1175/JCLI-D-21-0211.1

Herbert, R., & Stier, P. (2023). Satellite observations of smoke–cloud–radiation interactions over the Amazon rainforest. Atmospheric Chemistry and Physics, 23(7), 4595–4616. https://doi.org/10.5194/acp-23-4595-2023

Jeevanjee, N. (2022). Three Rules for the Decrease of Tropical Convection With Global Warming. Journal of Advances in Modeling Earth Systems, 14(11), e2022MS003285. https://doi.org/10.1029/2022MS003285

Luebke, A. E., Afchine, A., Costa, A., Grooß, J.-U., Meyer, J., Rolf, C., Spelten, N., Avallone, L. M., Baumgardner, D., & Krämer, M. (2016). The origin of midlatitude ice clouds and the resulting influence on their microphysical properties. Atmospheric Chemistry and Physics, 16(9), 5793–5809. https://doi.org/10.5194/acp-16-5793-2016

Raghuraman, S. P., Medeiros, B., & Gettelman, A. (2024). Observational Quantification of Tropical High Cloud Changes and Feedbacks. Journal of Geophysical Research: Atmospheres, 129(7), e2023JD039364. https://doi.org/10.1029/2023JD039364

Seeley, J. T., & Romps, D. M. (2015). Why does tropical convective available potential energy (CAPE) increase with warming? Geophysical Research Letters, 42(23), 10,429-10,437. https://doi.org/10.1002/2015GL066199

Seeley, J. T., Jeevanjee, N., Langhans, W., & Romps, D. M. (2019). Formation of Tropical Anvil Clouds by Slow Evaporation. Geophysical Research Letters, 46(1), 492–501. https://doi.org/10.1029/2018GL080747

Sherwood, S. C., Webb, M. J., Annan, J. D., Armour, K. C., Forster, P. M., Hargreaves, J. C., Hegerl, G., Klein, S. A., Marvel, K. D., Rohling, E. J., Watanabe, M., Andrews, T., Braconnot, P., Bretherton, C. S., Foster, G. L., Hausfather, Z., von der Heydt, A. S., Knutti, R., Mauritsen, T., … Zelinka, M. D. (2020). An Assessment of Earth’s Climate Sensitivity Using Multiple Lines of Evidence. Reviews of Geophysics, 58(4), e2019RG000678. https://doi.org/10.1029/2019RG000678

Sokol, A. B., & Hartmann, D. L. (2020). Tropical Anvil Clouds: Radiative Driving Toward a Preferred State. Journal of Geophysical Research: Atmospheres, 125(21), e2020JD033107. https://doi.org/10.1029/2020JD033107