Summary

This research focusses on the evolution of simulated ship tracks, examining the sensitivity of the duration of the ship track to precipitation intensity within the marine atmospheric boundary layer clouds.  The simulated ship tracks located within regions of classic stratocumulus cloud decks (closed mesoscale cellular convection) are found to persist for hours to days, in good agreement with observations.  The simulation of ship tracks over the central Pacific, within more broken MABL clouds, potentially open MCC, similarly persist in time, which is not in good agreement with observations.  Here the ship tracks are observed to be relatively short-lived. The hypothesis is that simulations are overly sensitive to the aerosol loading of the ship tracks, shutting down precipitation and allowing the ship tracks to persist for too long.  Effectively the simulations are too sensitive to the 2^nd aerosol indirect effect (i.e., the Albrecht effect.).

The paper is well-motivated and timely, building on the increasing interest in marine cloud brightening.  It falls well within the scope of ACP and will be of interest to the broader community.

Recommendation

This work needs major revisions, as identified below.  While I generally accept their hypothesis, I find their premise to be overly simplistic and their analysis needs to be more rigorous. 

Major Comment 1:

The basic premise is that the only difference between the Eastern Pacific ship tracks (A-C) and the Central Pacific ship tracks (D & E) is the intensity of the precipitation, hence the difference in the longevity of the ship tracks is strictly a function of the precipitation.  Both the satellite observations and simulations suggest that these two regions have vastly different types of MABL clouds with classic closed MCC in the Eastern Pacific and broken cumulus in the Central Pacific, potentially transitioning to trade cumulus or potentially open MCC.  In terms of MABL clouds, you are comparing apples to oranges.  There are differences in cloud fraction, MBL/cloud top height, entrainment from the free troposphere, large scale subsidence, wind speed, estimated inversion strength, and sensible and latent heat fluxes off the ocean.  Do we know if the Central Pacific MABL is decoupled or not?  How do we know that the rate of collision & coalescence removing aerosols within CASIM isn’t sensitive enough to turbulence in the MABL? 

To simply attribute all the differences in the persistence of the ship track to precipitation alone, is overly simplistic. 

While I find this a major concern, it can readily be addressed with a more comprehensive, robust discussion. Be upfront with the limitations of the analysis. 

Major Comment 2:

“Observations of precipitation in the domain are confirmed with overpassed (sic) from the Cloud Profiling Radar (CPR) onboard CloudSat (Stephens et al., 2008)) during the simulation.”

Are you really trying to say that since CloudSat (which CloudSat product, the 2C-column-precip or the 2C-rain-profile?) recorded precipitation somewhere along it’s overpass, that the simulated precipitation is reasonable?

In all sincerity, I was inclined to recommend rejecting the manuscript on that single sentence alone.

This is a major weakness in research, the simulated precipitation has not been evaluated in any way.  Given all the challenges we have in estimating precipitation in shallow convection over the remote ocean and the significant differences commonly found between various precipitation products, both from reanalyses and satellite-based products, it needs to be shown that simulated precipitation has some measure of skill.  The simulated precipitation, after all, underpins the analysis. 

I strongly recommend that a section on the evaluation of the simulated precipitation be added to this manuscript.  Ideally, the evaluation would be made against both ERA5 and GPM-IMERG for the full domain, commenting on the level of skill for both the classic stratocumulus over the Eastern Pacific and the broken cumulus of the Central Pacific.  In addition, an evaluation should be made along the CloudSat overpass, against both the 2C-RP and 2C-CP products.

Major Comment 3:

There appears to be a difference between the simulations and the satellite imagery in the cloud structure/morphology in the Central Pacific.  This difference is evident in Figure 5.  Do ship tracks D & E pass through this cloud field?  If so, please comment on how this may affect the analysis.

Minor comments

Line 32:   “large-scale” is ambiguous to me in this sentence.

Line 32:   The whole sentence confuses me.  Why are you isolating large-scale turbulence here? Are you saying the large-scale turbulence within coarse resolution CGMs are used in activation of CCN in boundary layer clouds? 

Line 34:   “certainty”?  We will never have certainty. But we can have some measure of confidence.

Line 47:   This sentence is out of place to me. The changes to the LWP happen after the suppression of precipitation (Albrecht effect), which is in the next sentence. 

Line 72:  “short timescales” is ambiguous

Line 80:   The simulated ship locations must be prescribed whether from an actual ship or not.

Line 115:   It would be nice to add text stating that the domain covers the Eastern and Central Pacific, off the coast of California. 

Line 120:   Do you need four references here?

Line 175:   Is ‘constrain’ the right word here?  You aren’t assimilating any of the satellite observations, are you?

Line 182:   Which CloudSat product?

Line 183:   I skimmed through Kato et al. (2010) but could find no mention of a precipitation product being produced, nor LWP or Nd.  Is this reference correct? If so please add the detail necessary to from the referenced material to the products used in this paper.   

Line 190:   Ship plumes are advected in wind fields, not positions.

Line 254:   Should you also define e_L here?

Figure 1:   This may work better if you present the control (left column) and the difference (right column.

Figure 9b:  Is this for the full four-year time period, as stated in line 186?  If so, why is it meaningful to compare a four-year average, that contains winter and summer, against a simulation of a single day during July?  Please address this issue.

Final comment:  This research focusses the impact of aerosol loading (the ship track) on the clouds and precipitation. It doesn’t explore the processes where clouds and precipitation impact the aerosol loading, i.e., collision-coalescence and wet-deposition.  It may strengthen your findings to consider this in the discussion.

Kang, L., Marchand, R., Wood, R. & McCoy, I. L. Coalescence scavenging drives droplet number concentration in Southern Ocean low clouds. Geophys. Res. Lett. 49, e2022GL097819 (2022).

Terai, C., Bretherton, C., Wood, R. & Painter, G. Aircraft observations of aerosol, cloud, precipitation, and boundary layer properties in pockets of open cells over the southeast Pacific. Atmos. Chem. Phys. 14, 8071–8088 (2014).

Wood, R., Leon, D., Lebsock, M., Snider, J. & Clarke, A. D. Precipitation driving of droplet concentration variability in marine low clouds. J. Geophys. Res.: Atmos. 117, D19210 (2012).

Alinejadtabrizi, T., Lang, F., Huang, Y. et al. Wet deposition in shallow convection over the Southern Ocean. npj Clim Atmos Sci 7, 76 (2024). https://doi.org/10.1038/s41612-024-00625-1

Review of Tippet et al., 2025: Evaluating simulations of ship tracks in a high-resolution model, submitted to ACP

The study by Tippet et al uses a ‘high-resolution’ atmospheric model with representations of aerosol and cloud microphysics to simulate the formation of cloud ship tracks. The author’s goal is to test the model using these ship track simulations to evaluate whether the model is suitable for application to marine cloud brightening scenarios.

The authors find that the model in question is able to simulate aerosol-cloud perturbations in some situations, but falls short in others.

I have significant methodological concerns, including about the apparently untested setup of the aerosol scheme. The model description is very thin, making it impossible to evaluate whether the model configuration is suitable. There is very little discussion and insight offered after presentation of the results, which limits the value of the manuscript for those who are not using the model in question. I also have comments about the structure of the paper. I believe significant revisions are required before acceptance can be considered.

Major comments and points requiring clarification

1. Overall, I found the manuscript to be quite disjointed. For example, Section 2.3 is a Methods section, however it contains results, before going back into Methods in Section 2.4. Then in the Results section, Equation 2 is outlining a method. A lot of Section 4.1 could also be moved to Methods.
2. Title and use of ‘high-resolution’: The phrase ‘high-resolution’ might mean different things to different readers. Perhaps consider using a more specific phrase, like ‘convection-permitting’ or ‘convection-resolving’ through the manuscript.
3. The manuscript’s language and sentence structure could be tightened up and polished. Doing so would make the manuscript considerably easier to read. I have tried to highlight various places in Other comments
4. The model description needs to be extended to give the reader basic and more accurate information about the model configuration. Specifically:
   1. The CASIM description (currently two sentences) doesn’t detail what cloud microphysics processes are represented.
   2. The nomenclature used for UKCA and GLOMAP is incorrect. UKCA is the framework used in the UM for handling atmospheric composition, whether aerosol or gas-phase chemistry. GLOMAP, specifically GLOMAP-mode, is the microphysical aerosol scheme available in the UM.
   3. There is no information about what non-ship aerosol emissions are included
5. The simplified GLOMAP-mode aerosol assumption (accumulation mode only) is a non-standard configuration. Has it been described and tested elsewhere? While it may be computationally cheaper, I have significant concerns about whether it is actually scientifically valid. The simulated aerosol in this simplified configuration is not representative of real world aerosol, which calls into question the simulated aerosol-cloud interactions. For example:
   1. Other components of marine aerosol are completely neglected, e.g. sea salt, which has a different hygroscopicity to sulfate, which will affect cloud droplet formation
   2. The other aerosol size modes can also contribute to cloud droplet formation, and possibly also impact the initiation of precipitation.
6. It is stated that aerosol number concentrations are initialised to 200 cm-3. Does the model reinitialise these concentrations every time it cycles? What happens at the boundaries? Are the simulated aerosol number concentrations allowed to increase or deplete in response to aerosol microphysical processes? Or are they tightly constrained?
7. Has any evaluation of the revised aerosol configuration been undertaken? In terms of aerosol number concentrations, or impacts on the cloud fields of modifying the aerosol?
8. With the above three points in mind, there is no discussion on what the consequences of the revised aerosol configuration might have on the results, say in comparison to the standard full aerosol configuration.
9. With the limits of ARG acknowledged, was any consideration given to implementing the Nenes et al., 2014 activation scheme?
10. The results in Section 3.2.2 (Timescales of response) are important, but don’t seem to be discussed further. Why is the response different between model and observations?

Other comments

Abstract, line 3-4: ‘A good natural analogue for investigating MCB is analysis of ship tracks’. Ship tracks are not natural, and describing their suitability as ‘good’ is over-stated. Perhaps ‘possible’?

Abstract, line 13: ‘unrealistically easy suppression’. By ‘easy’, perhaps ‘rapid’ or ‘early’ is meant?

Line 37: Delete ‘and’ after the second comma?

Line 44: ‘…and can be useful they can be…’?

Line 60: ‘…an cloud…’?

Line 67: ‘best estimate’, perhaps better referred to as a ‘proxy’?

Line 68: ‘This makes them’, please be more specific about what is being referred to.

Line 73: ‘logistical’? Perhaps meant ‘logical’?

Line 80: ‘… yet the simulated ship locations are prescribed and not from the actual ships that caused the observed ship tracks’. Can you expand on what is meant here? I don’t follow, surely in all simulations ship locations are prescribed? The model is not explicitly simulating the movement of the ship.

Line 84: ‘infer’, mean ‘simulate’?

Section 2.1: the title ‘Evaluating ship tracks’ isn’t very descriptive. I suggest something like ’Criteria for evaluating ship tricks’. Additionally, some commentary on how the criteria are applied, i.e. what is considered ‘correct’?

Line 115: in describing the domain, it would be helpful to state where the domain is alongside the co-ordinates, e.g. eastern North Pacific Ocean. A map would also be helpful for orientation; Figure 2 would be a good candidate if it was provided earlier on in the manuscript.

Line 139: what year is simulated?

Line 148: could you refer to Fig. 1c here to show the cellular convection?

Line 149: reference to ‘a drizzling scene’ corresponds to Fig. 1g?

Line 149: using the word ‘scene’ to describe a geographic domain is confusing. That terminology is useful when considering satellite images, but not appropriate for modelling.

Line 153: Reference to Fig. 5 suggests figures need to be re-ordered

Figure 1: the ‘ships’ column would be better presented as a difference plot for contrast to the ‘control’ column. For example, differences in (g) and (h) are very hard to see.

Figure 1: at what level are Na and Nd shown? Surface?

Figure 1 caption: do the dotted lines in (b) show the ship location histories rather than just their locations. If so, please state this in the caption.

Line 156: is there any significance attached to them being container ships? Is this an indirect proxy for size?

Line 163: is there any logic for emissions being added at 10 m?

Line 169: are the authors aware of the difference between condensation nuclei and cloud condensation nuclei? It’s not clear why a CN range is quoted, without further explanation of why 5x10^14 was then chosen.

Line 170: ‘issues with activation schemes’, is there a reference to support this?

Line 172: ‘so that all aerosols within the simulation are consistent’, can this statement be clarified? Does consistent mean homogeneous? In space and/or time?

Line 174: I don’t understand how this approach helps to ‘isolate causal aerosol effects’

Line 226: reference to Fig. S3 should be S2?

Line 228: I don’t follow the logic that ‘relatively clean’ equates to low background variability, can you expand?

Line 234: ‘cm-3’ is this a flux? i.e. per unit area per time?

Line 240: ‘implementing simple’ should be ‘implementing a simple’?

Line 247: ‘has be documented’ should be ‘has been documented’?

Figure 4 caption: please state what the axis labels are in unabbreviated form

Figure 6: I think it would also be valuable to see the simulated enhancements calculated using the lateral offset method

Line 299: are conditions matched between precipitating and non-precipitating conditions in the model? Or selected based on observed conditions? If the latter, are the simulated conditions representative of the observed conditions?

Line 305: Fig. S4 should be S3?

Line 309: ‘On short timescales’, meaning ‘Shortly after the ship passes’ or similar?

Figure 8 caption: Second sentence should begin with something like ‘The figure shows…’?

Line 314: ‘therefore there must be some process within the model…’. This is a bit unfulfilling. Can you review the output diagnostics? Or offer some supporting commentary linked to the processes included in the model?

Line 341: I don’t think it is accurate to say that a parameterisation has been tuned.

Line 355: Fig. S3 should be S2?

Line 367: ‘simple background’ meaning homogeneous background?

Line 371: ‘We find that the commonly used activation…’. Wasn’t this already known? Perhaps the study has confirmed that finding on Abdul-Razzak & Ghan.

Line 375: ‘at long timescales’. I would argue that the ‘workaround’ undermines the physical realism at all timescales.

Line 377: ‘which governs the conversion of cloud droplets to rain…’. Is the KK00 param the approach used in CASIM? A more thorough model introduction would make this clear.

Line 383: ‘an increases of Nd’, should be ‘an increase’?

Line 396: ‘where high-resolution global cloud resolving models are being employed’. Global climate models are not yet approaching resolutions that can be considered high nor cloud resolving?

Line 399: Do you have any insights or suggestions as to how autoconversion schemes could be improved?

Line 404: ‘In order to evaluate the model representation of these processes…’. I would suggest that the model representation hasn’t been evaluated, rather the simulated responses to a perturbation have been evaluated.

Line 410: ‘very clean background conditions’. I would make the argument that the region in question cannot be considered very clean background. As the study shows, there is considerable ship traffic in the area, and with aerosol lifetimes in the range of days to weeks, there is a strong possibility that the background conditions are anthropogenically influenced.

Paragraph beginning line 412: Can you suggest alternatives, or suggest how ARG can be revised / improved?

Line 438: ‘This has particular relevance’… It is not clear what ‘this’ refers to, especially as it used at the start of a paragraph.

Acknowledgements: are incomplete