the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Effects of Warming and Stratospheric Aerosol Injection on Tropical Cyclone Distribution and Frequency in a High-Resolution Global Circulation Model
Abstract. In recent years, as global circulation models (GCMs) have increased in spatial resolution, increasingly realistic tropical cyclones (TCs) and TC distributions have emerged from them. Where prior research on TC climatologies has relied on proxies like Potential Intensity (PI) and synthetic storm models, the cyclones emerging from the dynamics of newer GCMs can now be analyzed directly, using native model variables.
Such direct analysis may be particularly useful in studying possible global storm distributions under radically altered future climates, including high-emissions warming scenarios, and even those shaped by climate interventions. These interventions include various directed changes in global albedo, such as Stratospheric Aerosol Injection (SAI), with only limited precedent in the historical period. GCMs simulating realistic climate intervention scenarios, have not as of yet paired storm-resolving resolution with realistic intervention scenario construction. This has left gaps in our understanding as to how interventions might affect global storm/TC distributions.
In this paper, we utilize a new high-resolution model configuration to conduct experiments examining the effects of SAI, on tropical cyclones and global storm physics more broadly. These experiments are constructed based on prior work on SAI, using the GLENS GCM ensemble. Our analysis centers on 3 10-year experiments conducted using 30-km grid spacing. These include a recent-past calibration run; the Intergovernmental Panel on Climate Change climate pathway SSP (IPCC, 2021), for the years 2090–2099, with no SAI; and SSP 8.5, with SAI having begun in 2020 to maintain a global temperature rise of no more than 1.5 °C, also simulated for the years 2090–2099. With the resulting data sets, we deploy a novel TC-tracking algorithm to analyze resulting changes in storm tracks and properties. Based on our results for these different scenarios, we find that SAI, while in some ways restoring global storm patterns to a pre-warming state, may also create unique basin-scale TC distribution features and pose novel related hazards.
- Preprint
(27715 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2024-3526', Anonymous Referee #1, 23 Dec 2024
General comments
In this study, the authors look at the effect of injecting aerosols into the stratosphere on the distribution and frequency of tropical cyclones in the SSP5-8.5 scenario. For their study, they conduct and compare three simulations with the EarthWorks model: (1) a control simulation for the period 1990-1999, (2) an SSP5-8.5 simulation from 2090 to 2099 with forcings from CMIP6 and (3) an SSP5-8.5 simulations with stratospheric aerosol injection (SAI) from 2090 to 2099. While the question raised by the paper certainly falls within the journal’s scope, it is not at all clear where the new scientific contribution lies. This is down to the issues including the following: a convoluted writing style lacking structure, an introduction that does not clearly define and state the research gap, a lack of precision with scientific terms, a lack of references and claims that are questionable at best and untrue at worst (see below for more details). For the above reasons, it was extremely difficult to read the paper and pinpoint the new scientific contribution. Overall, the paper is of poor quality and should be rejected in its current form. I am extremely surprised that the paper was not rejected earlier.
The main hindrance to evaluating the scientific merits of this paper is the convoluted writing style and structure. The introduction is far too long. Dividing the introduction into subsections is also highly unusual and unnecessary for a research article. The paper also suffers from repetition in many places (see detailed comments for examples) and I suspect that the length of the paper can be reduced by at least half by removing unnecessary repetitions. While putting results in the context of the existing literature is an essential part of any scientific work, excessive citation of results from other studies in the analysis section makes it difficult to distinguish the authors’ own results from those of previous studies. I suggest the authors substantially shorten the introduction and remove the subsections. For the analysis section, the paper would benefit from a clearer separation between results and discussion. I suggest limiting the description of the authors’ own work to a results section, whereas comparison with results from other people’s work should be limited to a discussion section.
By not clearly defining the research gap in the introduction, the reader is left wondering as to what the new scientific contribution is. To be frank, the introduction reads more like a review paper than a research article; the introduction deals with several broad topics one after the other rather than honing in on a research gap. As a result, the last two paragraphs in the introduction describing the work done in this study appear out of place and do not follow naturally from the preceding paragraphs. At the end of the introduction, I did not understand why the authors had chosen the EarthWorks model in particular and the three simulations. I was also expecting a treatment of the literature dealing with the effect of stratospheric aerosols on TC distribution and frequency in the current climate, before discussing the effect of stratospheric aerosol injection in the future climate. I find it very odd that the authors do not touch upon the work done on the effect of volcanic eruptions on TC distribution and frequency.
Another major issue I have with this paper is the lack of precision with scientific and non-scientific terms. At several points in the paper, the term ‘significant’ is used to mean important or substantial. However, the term should be reserved for the statistical definition of ‘significant’. Another example is using the term ‘mechanism’ to describe a correlation without any reference to the underlying physics. Also, the term ‘explain’ is incorrectly used in many places where two variables correlate as might be expected. However, a correlation without any context and supporting evidence does not explain anything, let alone an underlying mechanism. More examples are listed in the specific comments below.
Although the authors excessively cite several papers throughout the manuscript, there are many places where references are required but not provided. No reference is provided for the EarthWorks model nor the Community Earth System Model, compromising the reproducibility of the authors’ work. At another point, the authors assert that this study ‘constitutes among the first attempts to directly model tropical storms in high resolution under climate intervention’. What were the previous attempts? Elsewhere in the text, one reference is described as ‘the literature’. I very much doubt that one paper alone can ever represent the entire body of scientific work. Without reference or evidence, the authors claim that the ‘TC pressures in EarthWorks outputs map more closely to real-world distributions..’. Further examples are detailed in the specific comments below.
The paper makes a series of questionable claims. In my mind, the most serious is the claim to have developed a novel tracking algorithm. Upon closer inspection, the authors themselves state in section 2.3 on the PHAST tracking algorithm that they ‘update an existing tracking algorithm.’ As far as I can make out, their modification to the existing tracking algorithm consists of converting the central pressure low to a wind speed via an empirical relationship between max wind speed and the central pressure. They then apply constraints on both the original wind field and the derived wind field. I fail to see how this modification results in a novel tracking algorithm. Fundamentally, the algorithm still depends on the near-surface wind speed and mean sea-level pressure, which is also the case for numerous other algorithms. Moreover, no real evidence is provided that the tracking algorithm does better than existing algorithms. The authors should remove the claim that the tracking algorithm is novel, not least because it does not properly credit the work of those who developed the original algorithms. Instead, they should characterise the algorithm as a modification/rehash of existing algorithms.
Finally, the paper suffers from a lack of discussion of limitations and errors. More details in the specific comments below.
Specific comments
L17-19: In what way exactly does SAI restore global storm patterns? And what is meant by novel hazards? Please be more specific.
L22: Strictly speaking, Emmanuel’s theory applies to a mature, axisymmetric TC in a steady state.
L29: What do you mean by available energy? Can you define this?
L41: What is the physical structure of a TC. Do you just mean the structure of the TC circulation? Or are you referring to small-scale processes, such as convection, that are not sufficiently simulated in coarse resolution simulations?
L50: Parsimonious means stingy and unwilling to spend money. However, I have absolutely no idea what a parsimonious relationship is. Do you mean a ‘tenuous’ relationship? Please select an alternative term that is appropriate for the context.
L50: What attributes are you referring to with the term ‘TC behaviour’? And how is ‘TC behaviour’ related to model skill? How do you define model skill?
L52: Tracking TCs in GCMs is standard practice. There is no need to use the term ‘direct’ here.’ It sounds like you are trying to articulate the advantages of using GCMs that explicitly resolve TCs.
L53: Define the term non-canonical cyclogenesis.
L66: Having read the paper, I fail to see how the tracking algorithm is novel.
L77: If you cannot quantify what you mean by ‘perhaps considerably’, please remove these two words.
L84: ‘Produced TC tracking results’ sounds like Bhatia et al. (2018) were testing tracking algorithms. However, this is not the case.
L88: Is it not obvious that TCs arise spontaneously? How else would they arise? I guess you are saying that they are not seeded in the simulation.
L94: What mechanisms for TC intensification are you referring to here? Why would one assume that the mechanisms for cyclogenesis would be similar when there exists no theory that can explain why we see different numbers of TCs from one year to the next?
L95: Again, I do not understand what is meant by a parsimonious relationship.
L96: Why do you refer to ‘a physics of TCs’? The field of physics is not different from one regime to another. Are you talking about mechanisms of some sort here?
L107-115: Yes, it is true that factors such as SST and vorticity feature in genesis indices and that these are empirically derived. Precisely for this reason, they have limited applicability to modelling studies, with the correlation between model TCs and genesis indices varying across models. Please mention this.
L114: Reference required for statement that such indices may not hold in future climate.
L116: Is this correlation positive or negative? What do you mean by a robust correlation?
L133: What are feedback algorithms?
L148: Why may the SSP5-8.5 be unlikely? If the scenario is unlikely, why do the authors focus on SAI in this scenario?
L168: What is loft SOO2?
L175: Since Kerry Emanuel’s theory of TCs being a carnot engine applies to a steady-state, mature TC, does it really make sense to invoke this theory for TC formation? What is exactly is the heat engine understanding of TC formation and propagation?
L175: Again, I do not know what a parsimonious relationship is.
L176-177: Please avoid using phrases such as ‘volcanism/sulfur’ and ‘hypothesized/observed’. As a reader, it is not my job to decide how to apply each word to the context.
L179: What do you mean by unexpected mechanism? What is the ‘simple SST-storm relationship’?
L191: Why say ‘may robustly extend to TC activity’ rather than ‘may extend to TC activity’? The word ‘robust’ is automatically negated by the word ‘may’.
L191: In which hemisphere is the SAI deployed?
L196: Why are the results contradictory?
L198: What exactly are heat-engine effects? I have never heard this term before. Do you just mean an increase in SST? If so, then say that.
L200: You refer to mechanisms without explaining the underlying physics that lead to the correlations that you describe. Without any explanation of the processes behind changes in the ITCZ and TC distributions, one can only really talk of correlations and not mechanisms.
L201: Is it not obvious that similar processes affect the distribution and frequency of TCs in the event of volcanic eruptions and aerosol-based interventions. After all, aerosols can be released into the stratosphere in both cases.
L203: Please remove the reference to prediction. The term prediction is reserved for forecasting. You cannot predict storm responses to SAI in the years 2090-2099. Instead you are trying to better understand the response of TC distribution and frequency to SAI.
L204: What effects and relationships are you talking about here? Please be more specific.
L205-212: Here you list a series of relationships or processes without justifying why they are being singled out for examination. There is also no description of how exactly you ‘will attempt to link any model-observed changes in TC statistics to one or more of theorized mechanisms’. I fail to see how looking at these various processes in your own simulations will provide a ‘framework’. At the most, looking at these processes will provide insight into the mechanisms behind the changes in TC distribution and frequency linked to SAI in the SSP5-8.5 scenario. The word ‘framework’ overstates the importance of your work.
L213-218: Again you are overstating the importance of your own work. I suggest you tone down the claim that you are in a position to ‘create outlines of potential TC hazards with salience well beyond the results of these specific runs’. Firstly, your study is conducted using a single model and single realisations, so the claim of applicability beyond the EarthWorks model is questionable. Secondly, this study is not about characterizing or quantifying TC hazards but the broad changes in TC distribution and frequency.
L219: Please provide reference for EarthWorks model.
L220: What are these new multiple techniques? The tracking algorithm itself is modified but not novel.
L229-230: Again, please provide a reference for the EarthWorks model.
L231: Mentioning 4-km horizontal grid spacing here is misleading, because the actual grid spacing for the simulations conducted here is ~30 km. Information about project goals do not belong in this paper, unless they are relevant for the study at hand.
L236: Please provide a reference for the Community Earth System Model. Please also Otherwise, it is impossible to ascertain how the EarthWorks
L243: Why are the MPAS dynamical cores notable? Are they better than the dynamical cores of other models? How is this relevant for your study?
L247: What metric are you using to come to the conclusion that the horizontal grid spacing yields realistic numbers of TCs. What is a realistic number and what is not?
L252: Please define ‘dynamical simulation properties’. Or are you referring to the dynamical core?
L254: What are scenario presets?
L255: So if SST is prescribed, then EarthWorks is an atmosphere-only model with no ocean component. This should be made clearer earlier on.
L256: We already know that SST is important for TC formation. Is it necessary to repeat this?
L258: How can ocean cooling be self-limiting? Firstly, I believe you mean sea-surface cooling. Secondly, the ocean cooling is not limited by the ocean cooling.
L266: What exactly do you mean by ‘likely before most appreciable change in climatic TC influences’? What climatic TC influences are you referring to. What do you define as appreciable change?
L274: Why is ‘Analysis’ capitalized.
L279: What are storm-generating SSTs? This is not a standard term.
L280: ‘Reduce/alter’. Which one is it?
L281: You say that this study ‘constitutes among the first attempts to directly model tropical storms in high resolution under climate intervention, and the first to [do] so utilizing a realistic SAI regime’. Have you discussed the previous attempts and referenced them in the text? What do you mean by a realistic SAI regime?
L284: I do not understand the term ‘storm-generating resolution’. Resolution itself does not generate storms. I think you mean ‘storm-resolving resolutions’.
L289: I am confused as to how one paper (Knaff and Zehr 2007) constitutes ‘the literature’.
L294: How do you determine that ‘TC central pressures in EarthWorks outputs map more closely to a real-world distribution’ compared to max winds?
L322: Please quantify what you mean by the track numbers being ‘virtually unchanged’ with the filtering? Otherwise, please remove this statement.
L328: How does filtering out storms weaker than category 1 increase the reliability of the best track data? Please explain or provide references. I guess you are alluding to the tracking of tropical storms (< category 1) being associated with larger errors compared to stronger storms.
L334: How do you define high fidelity here? From what I see, the model does a reasonable job. ‘High fidelity’ is an overstatement of the model’s ability to capture ‘historical TC behaviour’. The 605 TCs in the model is more than 10% more than the 517 in IBTrACS.
L343: The assertion that the model can simulate the full spectrum of intensity is misleading. The model can simulate the range of central low pressures including for category 5 storms. However, the model does not actually simulate maximum wind speeds corresponding to category 5 storms. Therefore, it is incorrect to claim that the model can simulate the full spectrum of intensity.
L343: What are downward intensity biases
L348: The order of the figures should correspond to the order in which they are introduced in the text. Why are Figure 4 and 6 mentioned before Figure 1, 2 and 3?
L349: Define observational data rounding.
L351: The term ‘great affinity’ again overstates the agreement between model and observations.
L355: Are these differences in lifetime due to the tracking algorithm or the EarthWorks model generating TCs further east compared to observations?
L360: What does CAT1 + IBTrACS mean? Are you averaging over the EarthWorks TCs and IBTrACS TCs?
L371: ‘Donuts’ is not an appropriate term to describe the TCs.
L373: What are ‘fundamental storm physics’?
Figure 1: It is strange to see a U10 plots with wind speeds corresponding to category 3 and be told that one is looking at a category 5 storm. Why do we need to convert pressure to wind speed using the wind-pressure relationship? Does it not make more sense to conduct the whole analysis using the mean sea-level pressure instead. Converting pressure to wind speed adds another level of abstraction to this study and makes it harder to interpret what is actually going on. Also, is there any reason to believe that the current, empirical wind-pressure relationship will hold in the SSP5-8.5 scenario?
L398: You say you see a significant correlation between RI and lifetime maximum intensity. I don’t. Please provide evidence of this statement. Either here or in the supplementary materials.
L401: How can rapid intensification drive storm development? Isn’t rapid intensification storm development itself?
L405: What is ‘modeling/earlier’ literature supposed to mean?
L411: Are you using ‘significant’ in the statistical sense? If not, please remove this word.
L415: What is a parsimonious relationship?
L421: What is a ‘linear part-way state’?
L426: I do not understand what ‘confounded’ means in the context of results.
L428: Why is this result unexpected? Using the term unexpected creates false expectations as to the importance of this result.
L428: You mean section 5 not Chapter 5, right?
L433: You say results of a similar magnitude were found by Bacmeister et al. (2018). What were their results?
Figure 2: The font size is too small.
Figure 4: If you want to compare the track numbers for the different simulations and IBTrACS, would it not be better to plot track density rather than the individual tracks?.
L444: How can you describe the emergence of TC activity in the South Atlantic as novel if TCs have already occurred there in the past? What exactly is the novel aspect here?
L446: Please provide a reference for the lack of TCs forming in the South Atlantic.
L452: Why are you generalising increasing TC activity in the South Atlantic to other ‘regions of the world’ without evidence to support this statement?
L457: Please specify what you mean by thermal energy. Are you referring to higher SSTs? Larger moist static energy? Or larger enthalpy?
L459: I do not agree with the statement that very few places experience an increase in TC frequency. What about the increase in the Pacific evident from the red contours there?
L463: What is the BRACE signal? What does BRACE stand for? Can you quantify the degree of correlation.
L469: How do you define ‘significant density’?
L470: What does ‘69 over the decade’ mean? Is this an increase or a decrease? Also, this number is meaningless without stating how this change relates to the absolute number of TCs in the historical scenario.
L473: Is the result robust? If not, please remove baseless speculation about the result being robust.
L476: What is a consistent correlation across scenarios? Do you mean that the ENSO index is very different between the three simulations, which makes it hard to attribute any differences in TC numbers to ENSO?
Figure 6: Font size is too small.
L478: Please define ‘relative global storm distribution patterns’. ’How can simulations share something that is defined as being relative?
L482: Please quantify what you mean by ‘matching’.
L485-488: You say that translational speeds are largely unchanged. But isn’t largely unchanged the same as saying a small change? ‘Largely unchanged’ is not sufficiently precise a term here.
L498-501: Why do refer to ‘polar amplification’ and ‘mid-level shear’ when you have not shown figures of polar amplification and mid-level shear in your results?
L512: Again, I do not know what heat-engine effects are.
Figure 8: Why are you plotting weekly TC frequency? Monthly frequency is standard and would be easier to compare with other studies?
L516: TC formation is not only dependent on SST, but a range of environmental factors.
L521: You refer to ‘overall hemispheric trends’, but it looks like you are referring to differences not trends.
L525: Why would a change in storm lifetimes affect the monthly distribution of TC frequency?
L526: Change ‘Chapter 5’ to section 5.
L529: What do you mean by ‘dynamically calculated land temperature patterns’?
L532: What do you mean by ‘global storm activity’? Are you referring to TC frequency? Or accumulated cyclone energy (ACE)? And how can you be sure that ‘significant’ increases are likely if only the global surface temperature changes?
L532-534: I do not understand what you are trying to say here.
L535-545: You talk about the effect of changes in SST on the distribution of TCs without any Figure or Table to back this up. Delete this text if you cannot back up these claims.
L549: What is a history-agnostic ENSO index?
L550: Above what SST value is convection permitted?
L553: What do you mean by ‘important new modalities’?
L556: Please provide references for the ocean upwelling argument and the permanent El-Niño arguments.
L559: What exactly is driven by a shift toward much more powerful positive-ENSO events? How can I assert from Figure 9 and 10 that the positive-ENSO events are ‘much more powerful’?
Figure 9: Caption makes no mention that SST anomaly is plotted. Which two simulations are used to calculate the SST anomaly? What does ONI stand for?
L562: What does ‘perhaps entirely’ mean? Please leave out baseless speculation.
L565: Do not ask the reader to check the long-term historical index values calculated in another study by themselves. Provide the comparison in the text.
L569: What does ‘provide an opportunity for this difference in scenario responses’ mean?
L571-576: How is this information about Atlantic TC activity relevant to your results?
L580: Where are these maps that you have produced? Presumably in Figure 11, 12 and 13?
L585: How are the changes ‘profound’?
L593: What is strong local forcing? Forcing of what?
L595-598: Why do you talk about the ‘Australasia storm development region’ and not the other regions. Stick with the global picture or discuss all regions. Why do you use ‘significant’ here if it is not used in the statistical sense?
Figure 13: I find this Figure with a different colorbar scale very misleading. The CIN values are much smaller than the values with the corresponding shades in Figure 11 and 12.
L606: Please be specific what you mean by ‘extreme’ degree of suppression.
L610: You assert that substantial increases in vertical shear in the southern hemisphere ‘can more than explain substantial reduction in SH storm activity’. Yet you cite this result from another study and you do not consider other environmental factors, such as relative humidity, vorticity etc… Please remove the word ‘explain’. You can assert that your results are consistent, but your results do not support the claim that vertical shear changes explain reductions in SH TC activity.
L613: This is another example of a juxtaposition of two words that are meaningless together. What do trends that ‘potentially most exemplify the complexity of cyclogenesis controlling factors’ mean. Either they exemplify the complexity or they do not.
L615: What do you mean by ‘powerfully’?
L618: Define ‘main-season SSTs’
L621: We already know that higher SSTs are more conducive to TC formation. Is it necessary to repeat this again?
L623-624: I don’t see a substantial expansion of a high-CIN region in the western North Pacific. Figure 11 does not display high CIN values here and Figure 13 is misleading with a different colorbar scale. Why do you not apply a consistent definition of what constitutes a high CIN value across all ocean basins?
L626: What do you mean ‘by NH standards’. Please quantify this statement.
L268: Again, ‘could more than …’ is meaningless.
L629: What is the difference between the being ‘likely the primary explanation’ and ‘one possible explanation’. Your wording is overstating the confidence of your results.
L630: What is a dominant explanation? How do you define dominant? How do your results show that this is the case?
L635: I do not understand this sentence at all.
L637-640: This paragraph is not relevant to your results.
L641: You have not examined the results with different paradigms. Instead, you have looked at individual relationships between various variables and processes, such as ENSO, and TC distribution and frequency.
L643: ‘Novel’ here is redundant as results are, by definition, always novel. Please articulate what is the novel contribution of your study to the field.
L643: What are resolved TC physics? And how can a scenario be realistic?
L645-650: I have serious reservations about the novelty of this tracking algorithm. It is not clear why the tracking algorithm features in the conclusions, as it has brought clear benefit to the study.
L651-654: This is not new information and does not belong in the conclusions. Also, you are grossly overstating the agreement between model and observations with the term ‘high fidelity’.
L656: I still do not understand what you mean by a non-linear here. Why would you expect TC frequency to scale linearly global surface temperature anyway?
L661: I am not convinced that computational power was the main limitation of this study. The main limitations stem from the limited analysis, single model, horizontal resolution and any model biases in the EarthWorks model. Please include a more comprehensive and candid discussion of limitations in the main text.
L673-679: I do not see the relevance to your work here.
Citation: https://doi.org/10.5194/egusphere-2024-3526-RC1 -
AC1: 'Reply on RC1', Andrew Feder, 26 Feb 2025
General Response:
So, a lot of the issues here you've pointed out are really fundamentals of how atmospheric science papers should be written, including issues with the non-statistical use of the word "significant," and similar issues, as well as imprecise language when talking about model properties. These I hope I've addressed, I've made the edits suggested in a new draft of the paper, not submitted here in line with the response guidelines, but I've described how I've addressed a lot of those below.
On the bigger-picture stuff, I have tried to get a lot more specific about EarthWorks model components, since this is right now the first EarthWorks paper. I've emphasized that these experiments are following something like a time-slice, storyline modeling approach, and they should be thought of as a first step toward understanding mesoscale and TC behavior arising from feedback algorithm SAI. I've situated algorithmically controlled SAI in the broader research environment, including its upcoming inclusion in GeoMip7.
With respect to the PHAST algorithm, it's true that the detection rules are based on TSTORMS. But, this is something substantially different we've made here. I've toned down describing it as "novel," but these changes solve real problems with TC tracking algorithms while still following very simple heuristics, and they're an important part of the scientific contributions, if any, made by this project.
Specific Comments:
Introduction:
"L17-19: In what way exactly does SAI restore global storm patterns? And what is meant by novel hazards? Please be more specific."
L17-19: Changed end of our intro to be more specific about what changes and what doesn't.
"L22: Strictly speaking, Emmanuel’s theory applies to a mature, axisymmetric TC in a steady state."
L22: TC's have more going on than just an ideal Carnot cycle. But, this provides an important way to understand their potential intensity. I hope "approximated" makes this more epistemologically modest of a claim, but the fundamental idea is important here.
"L29: What do you mean by available energy? Can you define this?"
L29: Available energy in the heat reservoir, Carnot engine sense.
"L41: What is the physical structure of a TC. Do you just mean the structure of the TC circulation? Or are you referring to small-scale processes, such as convection, that are not sufficiently simulated in coarse resolution simulations?"
L41: I guess it's better to say that the convection couldn't be resolved.
"L50: Parsimonious means stingy and unwilling to spend money. However, I have absolutely no idea what a parsimonious relationship is. Do you mean a ‘tenuous’ relationship? Please select an alternative term that is appropriate for the context."
L50: Parsimony is the quality of a scientific explanation reflecting the most straight-forward possible cause and effect relationship. But, it is maybe an uncommon word to use here, so I'll substitute it with some different words.
"L50: What attributes are you referring to with the term ‘TC behaviour’? And how is ‘TC behaviour’ related to model skill? How do you define model skill?"
L50: Added examples of the salient TC properties GCM's can capture, and some parts of large and small scale model dynamics that can influence existing storms.
"L52: Tracking TCs in GCMs is standard practice. There is no need to use the term ‘direct’ here.’ It sounds like you are trying to articulate the advantages of using GCMs that explicitly resolve TCs."
L52: You're right, what I'm really describing here is the advantage of high-resolution GCM's for TC's over the more indirect methods as performed at lower resolution.
"L53: Define the term non-canonical cyclogenesis."
L53: In this case, non-canonical cyclogenesis mostly means storms forming in new places, but I'll let that be covered by the phrase "changes in TC formation areas with shifting global temperatures."
"L66: Having read the paper, I fail to see how the tracking algorithm is novel."
L66: I'll be changing references to our algorithm we used as an adaption or modification of the Vitart et al one.
"L77: If you cannot quantify what you mean by ‘perhaps considerably’, please remove these two words."
L77: No more adjectives here.
"L84: ‘Produced TC tracking results’ sounds like Bhatia et al. (2018) were testing tracking algorithms. However, this is not the case."
L84: Number and intensity of storms, right, not anything about the tracking itself.
"L88: Is it not obvious that TCs arise spontaneously? How else would they arise? I guess you are saying that they are not seeded in the simulation. "
L88: I've changed to model native as our term for these, but I've wanted to emphasize when other studies are using the high-resolution and direct tracking combo, rather than any number of other methods of estimating storm activity or creating synthetic storms.
"L94: What mechanisms for TC intensification are you referring to here? Why would one assume that the mechanisms for cyclogenesis would be similar when there exists no theory that can explain why we see different numbers of TCs from one year to the next?"
L94: I've changed around this whole set of ideas. But basically, the numbers of intense TCs and TCs overall would always go up or down together if we make the assumption that we're not seed limited. I don't believe this is likely true of the real world, but its one of the possible ideas Kerry Emanuel puts forward in his retirement keynote lecture cited in this section, and this co-correlation would be an extrapolation from that.
"L95: Again, I do not understand what is meant by a parsimonious relationship."
L95: Let's just get rid of that whole modifying clause.
"L96: Why do you refer to ‘a physics of TCs’? The field of physics is not different from one regime to another. Are you talking about mechanisms of some sort here?"
L96: Let's probably not refer to "a physics" then.
"L107-115: Yes, it is true that factors such as SST and vorticity feature in genesis indices and that these are empirically derived. Precisely for this reason, they have limited applicability to modelling studies, with the correlation between model TCs and genesis indices varying across models. Please mention this."
L107-115: expanded discussion of their specificity to present observations.
"L114: Reference required for statement that such indices may not hold in future climate."
L114: There's no rigorous way right now for us to presently determine how these will do in a future climate vs. just in modeled future climates. But this statement was maybe too strong, I've softened it.
"L116: Is this correlation positive or negative? What do you mean by a robust correlation?"
L116: a robust positive correlation between vertical velocity and storm formation, stronger than for anything else they tested.
"L133: What are feedback algorithms?"
L133: any number of schemes where, I've added, we adjust the injections over time in response to present conditions.
"L148: Why may the SSP5-8.5 be unlikely? If the scenario is unlikely, why do the authors focus on SAI in this scenario?"
L148: SSP5-8.5 is unlikely if the Paris agreement goals are met. I don't think we're going to go that far, but we're probably significantly more likely to end up using SAI if we do. I've added another citation for a big article about how likely SSP5-8.5 is, but I've mentioned why studying this scenario might be important apart from signal-to-noise reasons.
"L168: What is loft SOO2?"
L168: LaTeX typo on SO2, and loft feels like a good word here since volcanoes shoot sulfur upwards but don't really inject at any particular height.
"L175: Since Kerry Emanuel’s theory of TCs being a carnot engine applies to a steady-state, mature TC, does it really make sense to invoke this theory for TC formation? What is exactly is the heat engine understanding of TC formation and propagation?"
L175: In the cited keynote, Emanuel says roughly, "if you stop throwing lit cigarettes into the forest, you'll stop having fires for a while, but eventually something else might just spark them anyway once there's a higher level of fuel available." When I've quoted the keynote, it's basically to cite Emanuel's own uncertainty about the extent to which TC's are seed limited or "fuel" (potential energy) limited. But I think if I've already referenced this idea, and a lot of this theorizing different relationships among variables under SAI is my own, it's perhaps confusing to keep referring to him speaking here.
"L175: Again, I do not know what a parsimonious relationship is."
"L176-177: Please avoid using phrases such as ‘volcanism/sulfur’ and ‘hypothesized/observed’. As a reader, it is not my job to decide how to apply each word to the context."
L175-177: Removed problematic phrasing
"L179: What do you mean by unexpected mechanism? What is the ‘simple SST-storm relationship’?"
L179: Storm seed number effects could stop lower SSTs from always leading to fewer storms and vice versa.
"L191: Why say ‘may robustly extend to TC activity’ rather than ‘may extend to TC activity’? The word ‘robust’ is automatically negated by the word ‘may’."
"L191: In which hemisphere is the SAI deployed?"
L191: Experiments were done with both NH and SH deployments, and they have counter effects on ITCZ position.
"L196: Why are the results contradictory?"
L196: Not as strong and consistent TC effects as for the SH injection.
"L198: What exactly are heat-engine effects? I have never heard this term before. Do you just mean an increase in SST? If so, then say that."
L198: Done
"L200: You refer to mechanisms without explaining the underlying physics that lead to the correlations that you describe. Without any explanation of the processes behind changes in the ITCZ and TC distributions, one can only really talk of correlations and not mechanisms."
L200: I added another mention of the shear effects of moving the ITCZ. I think I have established this as a mechanism now, though it is true it isn't directly the ITCZ itself effecting TCs.
"L201: Is it not obvious that similar processes affect the distribution and frequency of TCs in the event of volcanic eruptions and aerosol-based interventions. After all, aerosols can be released into the stratosphere in both cases. "
L201: Gone
"L203: Please remove the reference to prediction. The term prediction is reserved for forecasting. You cannot predict storm responses to SAI in the years 2090-2099. Instead you are trying to better understand the response of TC distribution and frequency to SAI."
L203: Changed to just "understand"
"L204: What effects and relationships are you talking about here? Please be more specific."
L204: The point I should have made clearer is that we have a mix of imperfect past proxies, and then more specific modeling scenarios but which are limited by the nature of modeling.
"L205-212: Here you list a series of relationships or processes without justifying why they are being singled out for examination. There is also no description of how exactly you ‘will attempt to link any model-observed changes in TC statistics to one or more of theorized mechanisms’. I fail to see how looking at these various processes in your own simulations will provide a ‘framework’. At the most, looking at these processes will provide insight into the mechanisms behind the changes in TC distribution and frequency linked to SAI in the SSP5-8.5 scenario. The word ‘framework’ overstates the importance of your work."
L205-212: These are the mechanisms/correlations we picked out of the literature, they reflect the concerns the SAI literature has had about ensuring they can recreate a pre-warming climate as closely as possible, and possible climate side effects. There are other possible ways to slice up how to describe different temperature patterns and changes in convection. But, I genuinely do feel that this summarizes the more important ideas at the convergence of the SAI/TC research areas.
"L213-218: Again you are overstating the importance of your own work. I suggest you tone down the claim that you are in a position to ‘create outlines of potential TC hazards with salience well beyond the results of these specific runs’. Firstly, your study is conducted using a single model and single realisations, so the claim of applicability beyond the EarthWorks model is questionable. Secondly, this study is not about characterizing or quantifying TC hazards but the broad changes in TC distribution and frequency. "
L213-218: I hope this is better, the final intro paragraph is now:
'We modify existing heuristics to track TCs occurring in this and other GCMs, and compare the performance of the resulting model-tracker combination to real-world storm data. We then use this baseline to evaluate changes in storm behavior given either advanced climate change or the global climate patterns created by SAI. This includes overall storm number, latitude/basin distributions, life cycle changes, and changes in typical TC intensities. Finally, we examine possible intermediate mechanisms, such as El Niño and changes in regional vertical shear, through which climate and TC formation may interact.'
"L219: Please provide reference for EarthWorks model."
"L220: What are these new multiple techniques? The tracking algorithm itself is modified but not novel."
L219-220: Changed to modified heuristics. The EarthWorks project, we will introduce a bit more in methods, but here with the tracking algorithm its not really that novel.
Methods:
"L229-230: Again, please provide a reference for the EarthWorks model."
L229-230: This is the first journal paper to be published using EarthWorks, our big methods paper will be published when the project is completed at the target resolution. But, I've now cited every component we actually use, our release documentation, and everything that would be necessary to also implement the model configuration we used. Basically, our work here has been making all these different components run together for a long time at higher resolution, without crashing and while shifting a lot of work load onto new GPU systems. The advances made by the EarthWorks project I should really have emphasized are all performance-related, but fundamental model properties should be unchanged for each of these components.
"L231: Mentioning 4-km horizontal grid spacing here is misleading, because the actual grid spacing for the simulations conducted here is ~30 km. Information about project goals do not belong in this paper, unless they are relevant for the study at hand."
L231: 4km mention is removed, we just talk about GSRM resolution more broadly.
"L236: Please provide a reference for the Community Earth System Model. Please also Otherwise, it is impossible to ascertain how the EarthWorks "
L236: This was clever, also, done.
"L243: Why are the MPAS dynamical cores notable? Are they better than the dynamical cores of other models? How is this relevant for your study?"
L243: The main advantage here among other possible ones, is that TCs won't need to take up more cells as they move poleward, as would be the case in a lat-lon grid model.
"L247: What metric are you using to come to the conclusion that the horizontal grid spacing yields realistic numbers of TCs. What is a realistic number and what is not?"
L247: A realistic number is based on best-track data. What do we currently observe in the world each year, and do the storms in the model have similar population properties?
"L252: Please define ‘dynamical simulation properties’. Or are you referring to the dynamical core?"
L252: You're right, I should just say dynamical core here.
"L254: What are scenario presets? "
L254: Like other model software, we have options built in to impose SSP trajectory GHG concentrations from different time periods, but this isn't really worth noting necessarily.
"L255: So if SST is prescribed, then EarthWorks is an atmosphere-only model with no ocean component. This should be made clearer earlier on. "
L255: I'll note this earlier in methods.
"L256: We already know that SST is important for TC formation. Is it necessary to repeat this?"
L256: Removed.
"L258: How can ocean cooling be self-limiting? Firstly, I believe you mean sea-surface cooling. Secondly, the ocean cooling is not limited by the ocean cooling."
L258: Rather, storms that cool the ocean surface might be self-limiting.
"L266: What exactly do you mean by ‘likely before most appreciable change in climatic TC influences’? What climatic TC influences are you referring to. What do you define as appreciable change?"
L266: This is hard, because the period of really good satellite TC records begins after climate change could have plausibly started effecting TC behavior. A more modest thing to say here, is that there's really a lot less carbon in the air already than in 2090 under SSP5-8.5.
"L274: Why is ‘Analysis’ capitalized."
L274: referred to a previous division of this work into three big sections, no longer present. This is a simple typo then, apologies.
"L279: What are storm-generating SSTs? This is not a standard term."
L279: I discuss the idea of threshold temperatures later, but for now, let's say just warm waters.
"L280: ‘Reduce/alter’. Which one is it?"
L280: Alter if we're talking broadly. The general purpose of SAI would be to try to reduce climate-related risks as much as possible, but our scientific question should be how do storms change, not specifically, does this improve things from a human perspective or not.
"L281: You say that this study ‘constitutes among the first attempts to directly model tropical storms in high resolution under climate intervention, and the first to [do] so utilizing a realistic SAI regime’. Have you discussed the previous attempts and referenced them in the text? What do you mean by a realistic SAI regime?"
L281: Added more discussion of the Irvine et al paper, the GeoMIP6 protocol, and the future GeoMIP7 protocol in our introduction. Irvine et al and Jones et al are the two papers we've found that previously simulated TCs under SAI at about our model's resolution. Otherwise, lots of big geoengineering modeling studies, typically at lower resolution, are based on the GeoMIP protocols, which so far haven't included algorithmic injection strategies as in the Tilmes et al 2020 ensemble.
"L284: I do not understand the term ‘storm-generating resolution’. Resolution itself does not generate storms. I think you mean ‘storm-resolving resolutions’."
L284: storm-resolving, that's better
"L289: I am confused as to how one paper (Knaff and Zehr 2007) constitutes ‘the literature’."
L289: They discuss lots of prior WPR work in the Knaff and Zehr paper, but the thing that's important here is that figure cited which illustrates the relationship. I've changed this to just "earlier work" since I don't go into WPRs much otherwise.
"L294: How do you determine that ‘TC central pressures in EarthWorks outputs map more closely to a real-world distribution’ compared to max winds? "
L294: Simply, pressure-based intensity gets us a very realistic humber of storm in each category. Wind speed does not, and based on the Davis paper, really should not in a model at our spacing.
"L322: Please quantify what you mean by the track numbers being ‘virtually unchanged’ with the filtering? Otherwise, please remove this statement."
L322: Made this more descriptive.
"L328: How does filtering out storms weaker than category 1 increase the reliability of the best track data? Please explain or provide references. I guess you are alluding to the tracking of tropical storms (< category 1) being associated with larger errors compared to stronger storms. "
L328: IBtRACS data for weaker systems gets better and more complete over time, but I am more confident in the CAT1+ data.
"L334: How do you define high fidelity here? From what I see, the model does a reasonable job. ‘High fidelity’ is an overstatement of the model’s ability to capture ‘historical TC behaviour’. The 605 TCs in the model is more than 10\% more than the 517 in IBTrACS."
L334: Toned down description here.
"L343: The assertion that the model can simulate the full spectrum of intensity is misleading. The model can simulate the range of central low pressures including for category 5 storms. However, the model does not actually simulate maximum wind speeds corresponding to category 5 storms. Therefore, it is incorrect to claim that the model can simulate the full spectrum of intensity. "
L343: I've changed this to a more modest, direct claim.
"L343: What are downward intensity biases"
L343: Anything that's weakening at least some TCs and stopping us from having all the major storms we would with a better, more realistic model.
"L348: The order of the figures should correspond to the order in which they are introduced in the text. Why are Figure 4 and 6 mentioned before Figure 1, 2 and 3?"
L348: Changed order of figures, now the ones I mentioned first appear first in the document and are numbered first.
"L349: Define observational data rounding."
L349: IBtRACS wind speed data is sometimes but not always rounded up to the nearest 5 knots.
"L351: The term ‘great affinity’ again overstates the agreement between model and observations. "
L351: Changed to, "fairly well."
"L355: Are these differences in lifetime due to the tracking algorithm or the EarthWorks model generating TCs further east compared to observations?"
L355: Looking closely at the dot colors in this region for our new figure 2, there's never a TC-intensity system over land in West Africa, just the minor cyclonic movement that becomes TCs over the ocean. Translational speed differences should really have some effect on actual TC track geometry changes, so I've put the phrase "at least partly" to make clear this isn't the only thing going on, and some other influences are discussed later.
"L360: What does CAT1 + IBTrACS mean? Are you averaging over the EarthWorks TCs and IBTrACS TCs?"
L360: Just means every storm CAT1 and stronger, I've put this in words now in this sentence.
"L371: ‘Donuts’ is not an appropriate term to describe the TCs."
371: I stand by this. I've heard folks use this term at conferences, it's descriptive of the current limitations of GCM TCs, and I think the imagery of it is both accurate and helpful.
"L373: What are ‘fundamental storm physics’?"
L373: Changed to "observed storm behavior," pressure and wind speed vary within and outside the eye wall a lot like in real life.
"Figure 1: It is strange to see a U10 plots with wind speeds corresponding to category 3 and be told that one is looking at a category 5 storm. Why do we need to convert pressure to wind speed using the wind-pressure relationship? Does it not make more sense to conduct the whole analysis using the mean sea-level pressure instead. Converting pressure to wind speed adds another level of abstraction to this study and makes it harder to interpret what is actually going on. Also, is there any reason to believe that the current, empirical wind-pressure relationship will hold in the SSP5-8.5 scenario? "
Figure 1 (now figure 5): I have also wrestled with this. Kerry Emanuel advocates categorizing TC intensity entirely based on central pressure. At the same time, the other studies in this area have described storm intensity in terms of wind speed or derived variables. Commonly used metrics of storm energy like ACE, PDI, etc are explicitly formulated in terms of max wind speeds.
Using the formula to convert between pressure and wind has sometimes been a headache, but other than just imposing an arbitrary wind correction function specific to our current model and resolution, pressure-based categorization is the only way to capture the full range of intensities that I think do meaningfully occur in our outputs. Yet, to be comparable with the existing set of TC and climate change studies, we have to use the CAT1-5 system, which is wind speed based. Some of these other modeling projects have not had this problem because, in CAM5 or HIFLOR, they've relied on GCMs that do produce CAT5 storms based on wind speed, but the Davis paper we cite shows this is not physically accurate at the resolutions they've used anyway, and relying on those wind outputs is just as arbitrary as this. So, we've ended up at this compromise.
The model-native U10 plot is important to show what our conversion really ends up meaning practically. That what we classify as a CAT5 storm based on pressure really does look like one in its wind field in a lot of ways, in terms of overall size, how clear the eye geometry is. It's just missing accurate max wind values, which we've kind of interpolated for it.
I'll add a paragraph about this to the TC tracker subsection.
"L398: You say you see a significant correlation between RI and lifetime maximum intensity. I don’t. Please provide evidence of this statement. Either here or in the supplementary materials."
L398: Shortened this paragraph and focused on the idea that, basing these cursory statistics on the ones Bhatia et al create in their study, we immediately do not see storms experiencing RI as becoming that much stronger than ones that don't. The point here is kind of more what we don't see.
"L401: How can rapid intensification drive storm development? Isn’t rapid intensification storm development itself?"
L401: Not as much a driver of intensity, rather.
"L405: What is ‘modeling/earlier’ literature supposed to mean?"
L405: I'm really mostly talking about the modeling literature, so let's just say that.
"L411: Are you using ‘significant’ in the statistical sense? If not, please remove this word."
L411: Done.
"L415: What is a parsimonious relationship?"
L415: I promise parsimonious is a real word in the philosophy of science, meaning an explanation requiring the fewest possible logical steps. I have overused it here, and in this case, it can be replaced witj "simplest."
"L421: What is a ‘linear part-way state’? "
L421: It's not just an intermediate state between the other two scenarios, there's something unique going on.
"L426: I do not understand what ‘confounded’ means in the context of results."
L426: influenced, particularly by the fact that each of our time slices could be highly affected by, say, one big El Nino cycle or similar event.
"L428: Why is this result unexpected? Using the term unexpected creates false expectations as to the importance of this result."
L428: It's not what I expected, but I guess there's been no big scientific consensus on what should happen here, so I've removed that.
"L428: You mean section 5 not Chapter 5, right? "
L428: Indeed.
"L433: You say results of a similar magnitude were found by Bacmeister et al. (2018). What were their results?"
L433: They divide the basin into 4 degree squares, produce a map of density of storm track through each, and find that the biggest Atlantic densities decrease from 18 storm hours/cell/year to about 8-10 hours.
"Figure 2: The font size is too small."
Figure 2: Made the fonts bigger
"Figure 4: If you want to compare the track numbers for the different simulations and IBTrACS, would it not be better to plot track density rather than the individual tracks?. "
Figure 4: Track density differences between IBtRACS and our model outputs will be heavily influenced by storm lifetime and translational speed biases. I think these maps of individual trajectories show important aspects of storm behavior, formation regions, where the storms end, the paths they collectively take, that are important and cover stuff the next plot with the density differences can't.
"L444: How can you describe the emergence of TC activity in the South Atlantic as novel if TCs have already occurred there in the past? What exactly is the novel aspect here?"
L444: It's novel in that this basin starts regularly seeing named-intensity TCs, versus not at all in the other simulations we do, and only once ever in the entire historical record. I've added this in the phrase "regular activity."
"L446: Please provide a reference for the lack of TCs forming in the South Atlantic. "
L446: The Pezza reference I've used again, also includes all the reasons for this.
"L452: Why are you generalising increasing TC activity in the South Atlantic to other ‘regions of the world’ without evidence to support this statement?"
L452: Got rid of this for simplicity.
"L457: Please specify what you mean by thermal energy. Are you referring to higher SSTs? Larger moist static energy? Or larger enthalpy?"
L457: Higher SSTs.
"L459: I do not agree with the statement that very few places experience an increase in TC frequency. What about the increase in the Pacific evident from the red contours there?"
L459: Now noted there.
"L463: What is the BRACE signal? What does BRACE stand for? Can you quantify the degree of correlation."
L463: BRACE is the "Benefits of Reducing Anthropogenic Climate changE" paradigm, and it studies how climate effects differ between RCP 8.5 and RCP 4.5. This is about 5-6 degrees of warming versus like 2-3 degrees at century end, but I've decided to get rid of this whole paragraph for space and because, as I said there before, the situations are not in fact directly comparable because of SAI specific effects.
"L469: How do you define ‘significant density’?"
L469: Changed to everywhere they typically occur each year presently, which excludes our South Atlantic basin and tracks in SSP5-8.5 that manage to travel southeast of Australia.
"L470: What does ‘69 over the decade’ mean? Is this an increase or a decrease? Also, this number is meaningless without stating how this change relates to the absolute number of TCs in the historical scenario."
L470: Added Control Run number here.
"L473: Is the result robust? If not, please remove baseless speculation about the result being robust."
L473: Removed this clause.
"L476: What is a consistent correlation across scenarios? Do you mean that the ENSO index is very different between the three simulations, which makes it hard to attribute any differences in TC numbers to ENSO?"
L476: It's that supressed Atlantic activity and enhanced east Pacific activity both occur very strongly with the very strong EL Nino response in Intervention, but east Pacific activity is not really enhanced with just advanced warming, while Atlantic activity is still quite suppressed, means something more complicated is going on here.
"Figure 6: Font size is too small."
Figure 6: Fixed this.
"L478: Please define ‘relative global storm distribution patterns’. ’How can simulations share something that is defined as being relative?"
L478: Let's just get rid of this whole paragraph since we added length elsewhere and the same trends are captured on a basin-by-basin scale elsewhere.
"L482: Please quantify what you mean by ‘matching’."
L482: Removed
"L485-488: You say that translational speeds are largely unchanged. But isn’t largely unchanged the same as saying a small change? ‘Largely unchanged’ is not sufficiently precise a term here."
L485-488: Now indicated more specifically here.
"L498-501: Why do refer to ‘polar amplification’ and ‘mid-level shear’ when you have not shown figures of polar amplification and mid-level shear in your results? "
L498-501: Polar amplification is one of the three degrees of freedom the SAI control algorithm used by Tilmes et al specifically tries to limit, and a major difference between the between the SAI climate they produce and the baseline SSP5-8.5, although we don't quantify here the amount of change in this. And in the present mid-level shear is what breaks storms apart as they north and south, but I don't show that's also what happens in our models. So, I've removed both these ways of describing what's happening to storms. Now it's just higher SSTs at mid latitudes, and whatever it is that kills storms as they travel poleward that isn't quite overcome with warming, seemingly.
"L512: Again, I do not know what heat-engine effects are."
L512: Changed to just ocean heat here.
"Figure 8: Why are you plotting weekly TC frequency? Monthly frequency is standard and would be easier to compare with other studies?"
Figure 8: Changed to monthly.
"L516: TC formation is not only dependent on SST, but a range of environmental factors."
L516: I'll just remove this small paragraph, it kind of just restates something you can see in the figure.
"L521: You refer to ‘overall hemispheric trends’, but it looks like you are referring to differences not trends."
L521: Let's say hemisphere-level results here.
"L525: Why would a change in storm lifetimes affect the monthly distribution of TC frequency? "
L525: If storms were forming at the same time, but each lived a few days longer, say, median storm occurrence dates would also stretch back that far. I've added a fuller explanation of this possibility.
"L526: Change ‘Chapter 5’ to section 5."
L526: Changed to "Bulk Variable Analysis."
"L529: What do you mean by ‘dynamically calculated land temperature patterns’?"
L529: Basically that these came out of EarthWorks, changed to that.
"L532: What do you mean by ‘global storm activity’? Are you referring to TC frequency? Or accumulated cyclone energy (ACE)? And how can you be sure that ‘significant’ increases are likely if only the global surface temperature changes?"
L532: Changed to storm frequency, cited our SST threshold paper again (Tory and Dare), and removed non-statistical use of word "significant." But, increasing SSTs, all else equal, should enable more TC seeds to form into storms.
"L532-534: I do not understand what you are trying to say here."
L532-534: Let's just get rid of this.
"L535-545: You talk about the effect of changes in SST on the distribution of TCs without any Figure or Table to back this up. Delete this text if you cannot back up these claims."
L535-545: Added global temperature change maps (which I previously got rid of because I thought my paper was running too long). I've also changed the phrasing here, just to say that SST changes must be acting counter to some other inhibition on TC development.
"L549: What is a history-agnostic ENSO index?"
L549: Doesn't depend on a long-term climatology of, say, the Nino 3.4 region.
"L550: Above what SST value is convection permitted?"
L550: As described by Williams, and now also here, they decided that because deep convection in one longitude has to be balanced by downwelling elsewhere, the actual global mean surface temperature in a given latitude is a good proxy for a convective threshold.
"L553: What do you mean by ‘important new modalities’? "
L553: I've deleted that, but basically if we were just judging ENSO strength by temperature in the ENSO region, we'd have trouble telling what's, say, a strong El Nino, and what's just a result of global temperature changes manifest in the same region.
"L556: Please provide references for the ocean upwelling argument and the permanent El-Niño arguments."
L556: The mechanisms behind permanent El Nino don't really matter here, but Pacific temeperature patterns like this are what other research describes as a permanent El Nino, see the forcings used in the Duque-Villegas et al paper.
"L559: What exactly is driven by a shift toward much more powerful positive-ENSO events? How can I assert from Figure 9 and 10 that the positive-ENSO events are ‘much more powerful’?"
L559: The ELI index value mean is much higher for the whole decade. But this is driven by very large temporary states where the ELI index moves east. ELI is a fairly good proxy for El Nino as measured by SSTs in the present climate. And so using it here, when we get those big positive monthly values in what's now figure 11, those are powerful El Nino events.
"Figure 9: Caption makes no mention that SST anomaly is plotted. Which two simulations are used to calculate the SST anomaly? What does ONI stand for?"
Figure 9: Done.
"L562: What does ‘perhaps entirely’ mean? Please leave out baseless speculation."
L562: Removed this phrasing.
"L565: Do not ask the reader to check the long-term historical index values calculated in another study by themselves. Provide the comparison in the text."
L565: Added comparison values.
"L569: What does ‘provide an opportunity for this difference in scenario responses’ mean?"
L569: Basically just that these intermediate things are how ENSO affects TC activity in the region, but we don't need that clause.
"L571-576: How is this information about Atlantic TC activity relevant to your results?"
L571-576: This is important because it shows that, while El Nino events suppress Atlantic TC activity, this also seems to hold for permanent El Nino/El Nino like warming too. This isn't dependent on any mechanisms that can only be temporary.
"L580: Where are these maps that you have produced? Presumably in Figure 11, 12 and 13?"
L580: Yup, now 12 13 14, now indicated.
"L585: How are the changes ‘profound’?"
L585: They just appear big, got rid of that adjective.
"L593: What is strong local forcing? Forcing of what?"
L593: Basically local changes in shear between SSP5-8.5 and control, changed to just no strong local increase.
"L595-598: Why do you talk about the ‘Australasia storm development region’ and not the other regions. Stick with the global picture or discuss all regions. Why do you use ‘significant’ here if it is not used in the statistical sense?"
L595-598: Got rid of this altogether.
"Figure 13: I find this Figure with a different colorbar scale very misleading. The CIN values are much smaller than the values with the corresponding shades in Figure 11 and 12."
Figure 13: I've noted this change in colorbar more directly, and also changed the color here to orange instead of red. But this is world's most active storm area, I feel like I do need this third focus graph.
"L606: Please be specific what you mean by ‘extreme’ degree of suppression."
L606: Let's not call it "extreme," although it is our biggest basin-level change.
"L610: You assert that substantial increases in vertical shear in the southern hemisphere ‘can more than explain substantial reduction in SH storm activity’. Yet you cite this result from another study and you do not consider other environmental factors, such as relative humidity, vorticity etc… Please remove the word ‘explain’. You can assert that your results are consistent, but your results do not support the claim that vertical shear changes explain reductions in SH TC activity."
L610: This can be, "well in line with."
"L613: This is another example of a juxtaposition of two words that are meaningless together. What do trends that ‘potentially most exemplify the complexity of cyclogenesis controlling factors’ mean. Either they exemplify the complexity or they do not."
L613: I've changed this to, "illustrate competing basin-level influences."
" L615: What do you mean by ‘powerfully’?"
L615: We'll change this to, high ENSO indices.
"L618: Define ‘main-season SSTs’"
L618: Aug-Oct
"L621: We already know that higher SSTs are more conducive to TC formation. Is it necessary to repeat this again?"
L621: Removed.
"L623-624: I don’t see a substantial expansion of a high-CIN region in the western North Pacific. Figure 11 does not display high CIN values here and Figure 13 is misleading with a different colorbar scale. Why do you not apply a consistent definition of what constitutes a high CIN value across all ocean basins?"
L623-624: There's definitely an expansion visible in figure 12, though I've removed substantial.
"L626: What do you mean ‘by NH standards’. Please quantify this statement."
L626: There's bands of very high shear in the southern ocean, but I've removed this qualifier.
"L268: Again, ‘could more than …’ is meaningless."
L628: Removed.
"L629: What is the difference between the being ‘likely the primary explanation’ and ‘one possible explanation’. Your wording is overstating the confidence of your results."
L629: Removed.
"L630: What is a dominant explanation? How do you define dominant? How do your results show that this is the case?"
L630: Changed to overarching, but big positive ENSO shifts are consistent with lots of the regional condition changes that end up affecting TCs globally, and we do show that above.
"L635: I do not understand this sentence at all."
L635: Removed the last sentence, big idea of this conclusion is that ENSO effects are the main things changing TC distributions globally between our Geoengineering and Control scenarios, but not the only thing, especially for the east Pacific.
"L637-640: This paragraph is not relevant to your results."
L637-640: Removed.
"L641: You have not examined the results with different paradigms. Instead, you have looked at individual relationships between various variables and processes, such as ENSO, and TC distribution and frequency.
L643: ‘Novel’ here is redundant as results are, by definition, always novel. Please articulate what is the novel contribution of your study to the field.
L643: What are resolved TC physics? And how can a scenario be realistic?"
L641-643: I think this is better. Maybe we haven't used any real new analytical paradigms, but certainly this is where geoengineering research has advanced to now, and this is the first study combining such with TC tracking.
"L645-650: I have serious reservations about the novelty of this tracking algorithm. It is not clear why the tracking algorithm features in the conclusions, as it has brought clear benefit to the study."L645-650: We've created a tracking algorithm based on really simple, easily describable rules like TSTORMS, but which also has features to let storms be detected much closer to each other in space without interference, exclude subtropical systems, and have intensity measured based on pressure instead of wind speed, as called for in the Davis (2018) paper. It's about as different from TSTORMS as other tracking algorithms are from each other, though that was our basis to start. I stand by the importance of the work I did looking through the literature, talking with folks about problems with TSTORMS and similar algorithms, and I feel this is in fact an important contribution.
"L651-654: This is not new information and does not belong in the conclusions. Also, you are grossly overstating the agreement between model and observations with the term ‘high fidelity’."
L651-654: I've changed high fidelity to good agreement. This does belong in the conclusion because this is the first paper using EarthWorks for research, though others at higher resolution are on the way. The fact that this suite of components, together, has done so well representing the world's storm basin geometries and overall activity in our Control Run is a very good sign for future experiments trying to study the intersection of weather and climate with EarthWorks, I stand by this paragraph.
"L656: I still do not understand what you mean by a non-linear here. Why would you expect TC frequency to scale linearly global surface temperature anyway?"
L656: Here, basically the big thing is that SAI doesn't return us to the global TC profile we had before warming. It's really this third, other tempestology, perhaps because of side effects of SAI or stuff it fails to prevent, like permanent El Nino.
"L661: I am not convinced that computational power was the main limitation of this study. The main limitations stem from the limited analysis, single model, horizontal resolution and any model biases in the EarthWorks model. Please include a more comprehensive and candid discussion of limitations in the main text."
L661: Computing power is what's limited our ensemble size, simulation run time, and thus many forms of analysis that would depend on having more data. I've expressed this relationship more directly in this paragraph now. And also, a new paragraph in Section 2 about our lack of an ensemble here.
"L673-679: I do not see the relevance to your work here."
L673-679: Replaced our last paragraph:
'Thus, in both the specific questions we have sought to address here, and the tools and resources we have marshaled to do so, this work can only be understood as the very beginning of efforts to understand how different forms of targeted SAI implementation could impact TC and other mesoscale event risk. As other researchers pursue impacts research in this area, perhaps with new, standardized protocols \citep{GeoMIP7}, they will face similar methodological and other challenges, and we hope that this could serve as a starting point for developing techniques and research questions in this space.'
Citation: https://doi.org/10.5194/egusphere-2024-3526-AC1
-
AC1: 'Reply on RC1', Andrew Feder, 26 Feb 2025
-
RC2: 'Comment on egusphere-2024-3526', Anonymous Referee #2, 17 Jan 2025
In this paper, the authors "describe" (for the reason why this is in quote, see the rest of the review) the results of a set of simulations at high resolution (30km) and investigate Tropical Cyclones changes under global warming and under SAI. I am not a TC expert, therefore there are some part of the manuscript I am less likely to have good insight into, so I hope other reviewers can make up for my lack there (such as on Section 2.3).
I think the paper has the potential eventually to be very good, but there is a rather large problem that needs addressing before anything else, and that makes this paper unpublishable at this stage: EarthWorks is announced and praised, and TC results are validated, but there are no references in the Methods about the model itself. No effort is done to show results for the surface climate (what's the warming in EarthWorks under the GHG scenarios? The first phrase where warming is discussed is "In line with the bulk of prior modeling/earlier literature, severe warming in EarthWorks yields a world with fewer TCs, but a greater number of CAT5 and outlier/extreme intensity storms" but... there's no discussion of how much warming there is? How does the ITCZ look like? Do clouds make sense? Does the model has a stratospheric circulation, or if the aerosols are prescribed, what radiative code is used? What's the resulting TOA imbalance?) If this is a new model, then there needs to be a serious validation behind that can't jump straight to the TC results. The only "paper" cited about EarthWorks is a proposal to NSF as a PDF: to say this is not best practice is definitely not enough. I can't judge the maps I see about shear if I can't see what the surface temperature in this model is. The authors also don't explain how they use the forcing from CESM2: sometimes they talk about SO2, but does it means you're using the SO2 injection locations from the Tilmes et al. (2020) paper, or the aerosols forcing in 3D? How is it assimilated, and have you verified it works? If so, why are you not showing this?
In no way the way the authors present their data about a new model (also considering the lack of a data availability statement about the models result, not to mention the code) is in line with the practices outlined by ESD (or by any other reputable journal in 2025): https://www.earth-system-dynamics.net/policies/data_policy.html Their description of EarthWorks is non-existent.
With all this said, I really can't recommend anything more than a sound rejection at this point. I have other comments about the manuscript and some of the analyses (most of them positive!) but I don't think I can engage with a paper that ignores any guideline set in the last decade about sharing underlying data, model validation and transparency. If the authors have a manuscript of validation of EarthWorks (which I didn't find also by Googling, at most I found a GitHub page the authors didn't even bother adding here), they should cite it, and they should explain how the forcing from SAI works in their model (if this is a new model, even if based on CESM, they should show it in depth). Only after all of this can this manuscript, which is downstream of those, be assessed for its scientific merit.
Citation: https://doi.org/10.5194/egusphere-2024-3526-RC2 -
AC2: 'Reply on RC2', Andrew Feder, 26 Feb 2025
So, this is a little difficult. This is the first paper describing EarthWorks results because it's a preliminary set of experiments before we get to our target resolution. But, all the components used here (MPAS land, ice/ocean, and atmosphere, the community land model) exist independently, and could be used all together without EarthWorks. They would just run very slowly, largely on CPUs. The important thing noted here now, in my latest set of edits to the paper, is that in terms of outputs, EarthWorks is identical to the sum of these existing components. This may not be enough for you, understandably, in which case the only remedy would be to only publish this paper later once there's a big methods paper for the EarthWorks configuration describing a lot more of the relevant model/climate properties. But as you say, that can't really be what this paper is, short of adding really a lot more stuff with a different focus.
In terms of stuff that is immediately fixable, I've added temperature delta maps (that I previously removed for space reasons). I try to explicitly state that we are forcing SO2 fluxes, these undergo chemical changes, and then do eventually fall out. EarthWorks model code is in the EarthWorks Github, now in the code availability section alongside the tracking algorithm code.
Citation: https://doi.org/10.5194/egusphere-2024-3526-AC2
-
AC2: 'Reply on RC2', Andrew Feder, 26 Feb 2025
Status: closed
-
RC1: 'Comment on egusphere-2024-3526', Anonymous Referee #1, 23 Dec 2024
General comments
In this study, the authors look at the effect of injecting aerosols into the stratosphere on the distribution and frequency of tropical cyclones in the SSP5-8.5 scenario. For their study, they conduct and compare three simulations with the EarthWorks model: (1) a control simulation for the period 1990-1999, (2) an SSP5-8.5 simulation from 2090 to 2099 with forcings from CMIP6 and (3) an SSP5-8.5 simulations with stratospheric aerosol injection (SAI) from 2090 to 2099. While the question raised by the paper certainly falls within the journal’s scope, it is not at all clear where the new scientific contribution lies. This is down to the issues including the following: a convoluted writing style lacking structure, an introduction that does not clearly define and state the research gap, a lack of precision with scientific terms, a lack of references and claims that are questionable at best and untrue at worst (see below for more details). For the above reasons, it was extremely difficult to read the paper and pinpoint the new scientific contribution. Overall, the paper is of poor quality and should be rejected in its current form. I am extremely surprised that the paper was not rejected earlier.
The main hindrance to evaluating the scientific merits of this paper is the convoluted writing style and structure. The introduction is far too long. Dividing the introduction into subsections is also highly unusual and unnecessary for a research article. The paper also suffers from repetition in many places (see detailed comments for examples) and I suspect that the length of the paper can be reduced by at least half by removing unnecessary repetitions. While putting results in the context of the existing literature is an essential part of any scientific work, excessive citation of results from other studies in the analysis section makes it difficult to distinguish the authors’ own results from those of previous studies. I suggest the authors substantially shorten the introduction and remove the subsections. For the analysis section, the paper would benefit from a clearer separation between results and discussion. I suggest limiting the description of the authors’ own work to a results section, whereas comparison with results from other people’s work should be limited to a discussion section.
By not clearly defining the research gap in the introduction, the reader is left wondering as to what the new scientific contribution is. To be frank, the introduction reads more like a review paper than a research article; the introduction deals with several broad topics one after the other rather than honing in on a research gap. As a result, the last two paragraphs in the introduction describing the work done in this study appear out of place and do not follow naturally from the preceding paragraphs. At the end of the introduction, I did not understand why the authors had chosen the EarthWorks model in particular and the three simulations. I was also expecting a treatment of the literature dealing with the effect of stratospheric aerosols on TC distribution and frequency in the current climate, before discussing the effect of stratospheric aerosol injection in the future climate. I find it very odd that the authors do not touch upon the work done on the effect of volcanic eruptions on TC distribution and frequency.
Another major issue I have with this paper is the lack of precision with scientific and non-scientific terms. At several points in the paper, the term ‘significant’ is used to mean important or substantial. However, the term should be reserved for the statistical definition of ‘significant’. Another example is using the term ‘mechanism’ to describe a correlation without any reference to the underlying physics. Also, the term ‘explain’ is incorrectly used in many places where two variables correlate as might be expected. However, a correlation without any context and supporting evidence does not explain anything, let alone an underlying mechanism. More examples are listed in the specific comments below.
Although the authors excessively cite several papers throughout the manuscript, there are many places where references are required but not provided. No reference is provided for the EarthWorks model nor the Community Earth System Model, compromising the reproducibility of the authors’ work. At another point, the authors assert that this study ‘constitutes among the first attempts to directly model tropical storms in high resolution under climate intervention’. What were the previous attempts? Elsewhere in the text, one reference is described as ‘the literature’. I very much doubt that one paper alone can ever represent the entire body of scientific work. Without reference or evidence, the authors claim that the ‘TC pressures in EarthWorks outputs map more closely to real-world distributions..’. Further examples are detailed in the specific comments below.
The paper makes a series of questionable claims. In my mind, the most serious is the claim to have developed a novel tracking algorithm. Upon closer inspection, the authors themselves state in section 2.3 on the PHAST tracking algorithm that they ‘update an existing tracking algorithm.’ As far as I can make out, their modification to the existing tracking algorithm consists of converting the central pressure low to a wind speed via an empirical relationship between max wind speed and the central pressure. They then apply constraints on both the original wind field and the derived wind field. I fail to see how this modification results in a novel tracking algorithm. Fundamentally, the algorithm still depends on the near-surface wind speed and mean sea-level pressure, which is also the case for numerous other algorithms. Moreover, no real evidence is provided that the tracking algorithm does better than existing algorithms. The authors should remove the claim that the tracking algorithm is novel, not least because it does not properly credit the work of those who developed the original algorithms. Instead, they should characterise the algorithm as a modification/rehash of existing algorithms.
Finally, the paper suffers from a lack of discussion of limitations and errors. More details in the specific comments below.
Specific comments
L17-19: In what way exactly does SAI restore global storm patterns? And what is meant by novel hazards? Please be more specific.
L22: Strictly speaking, Emmanuel’s theory applies to a mature, axisymmetric TC in a steady state.
L29: What do you mean by available energy? Can you define this?
L41: What is the physical structure of a TC. Do you just mean the structure of the TC circulation? Or are you referring to small-scale processes, such as convection, that are not sufficiently simulated in coarse resolution simulations?
L50: Parsimonious means stingy and unwilling to spend money. However, I have absolutely no idea what a parsimonious relationship is. Do you mean a ‘tenuous’ relationship? Please select an alternative term that is appropriate for the context.
L50: What attributes are you referring to with the term ‘TC behaviour’? And how is ‘TC behaviour’ related to model skill? How do you define model skill?
L52: Tracking TCs in GCMs is standard practice. There is no need to use the term ‘direct’ here.’ It sounds like you are trying to articulate the advantages of using GCMs that explicitly resolve TCs.
L53: Define the term non-canonical cyclogenesis.
L66: Having read the paper, I fail to see how the tracking algorithm is novel.
L77: If you cannot quantify what you mean by ‘perhaps considerably’, please remove these two words.
L84: ‘Produced TC tracking results’ sounds like Bhatia et al. (2018) were testing tracking algorithms. However, this is not the case.
L88: Is it not obvious that TCs arise spontaneously? How else would they arise? I guess you are saying that they are not seeded in the simulation.
L94: What mechanisms for TC intensification are you referring to here? Why would one assume that the mechanisms for cyclogenesis would be similar when there exists no theory that can explain why we see different numbers of TCs from one year to the next?
L95: Again, I do not understand what is meant by a parsimonious relationship.
L96: Why do you refer to ‘a physics of TCs’? The field of physics is not different from one regime to another. Are you talking about mechanisms of some sort here?
L107-115: Yes, it is true that factors such as SST and vorticity feature in genesis indices and that these are empirically derived. Precisely for this reason, they have limited applicability to modelling studies, with the correlation between model TCs and genesis indices varying across models. Please mention this.
L114: Reference required for statement that such indices may not hold in future climate.
L116: Is this correlation positive or negative? What do you mean by a robust correlation?
L133: What are feedback algorithms?
L148: Why may the SSP5-8.5 be unlikely? If the scenario is unlikely, why do the authors focus on SAI in this scenario?
L168: What is loft SOO2?
L175: Since Kerry Emanuel’s theory of TCs being a carnot engine applies to a steady-state, mature TC, does it really make sense to invoke this theory for TC formation? What is exactly is the heat engine understanding of TC formation and propagation?
L175: Again, I do not know what a parsimonious relationship is.
L176-177: Please avoid using phrases such as ‘volcanism/sulfur’ and ‘hypothesized/observed’. As a reader, it is not my job to decide how to apply each word to the context.
L179: What do you mean by unexpected mechanism? What is the ‘simple SST-storm relationship’?
L191: Why say ‘may robustly extend to TC activity’ rather than ‘may extend to TC activity’? The word ‘robust’ is automatically negated by the word ‘may’.
L191: In which hemisphere is the SAI deployed?
L196: Why are the results contradictory?
L198: What exactly are heat-engine effects? I have never heard this term before. Do you just mean an increase in SST? If so, then say that.
L200: You refer to mechanisms without explaining the underlying physics that lead to the correlations that you describe. Without any explanation of the processes behind changes in the ITCZ and TC distributions, one can only really talk of correlations and not mechanisms.
L201: Is it not obvious that similar processes affect the distribution and frequency of TCs in the event of volcanic eruptions and aerosol-based interventions. After all, aerosols can be released into the stratosphere in both cases.
L203: Please remove the reference to prediction. The term prediction is reserved for forecasting. You cannot predict storm responses to SAI in the years 2090-2099. Instead you are trying to better understand the response of TC distribution and frequency to SAI.
L204: What effects and relationships are you talking about here? Please be more specific.
L205-212: Here you list a series of relationships or processes without justifying why they are being singled out for examination. There is also no description of how exactly you ‘will attempt to link any model-observed changes in TC statistics to one or more of theorized mechanisms’. I fail to see how looking at these various processes in your own simulations will provide a ‘framework’. At the most, looking at these processes will provide insight into the mechanisms behind the changes in TC distribution and frequency linked to SAI in the SSP5-8.5 scenario. The word ‘framework’ overstates the importance of your work.
L213-218: Again you are overstating the importance of your own work. I suggest you tone down the claim that you are in a position to ‘create outlines of potential TC hazards with salience well beyond the results of these specific runs’. Firstly, your study is conducted using a single model and single realisations, so the claim of applicability beyond the EarthWorks model is questionable. Secondly, this study is not about characterizing or quantifying TC hazards but the broad changes in TC distribution and frequency.
L219: Please provide reference for EarthWorks model.
L220: What are these new multiple techniques? The tracking algorithm itself is modified but not novel.
L229-230: Again, please provide a reference for the EarthWorks model.
L231: Mentioning 4-km horizontal grid spacing here is misleading, because the actual grid spacing for the simulations conducted here is ~30 km. Information about project goals do not belong in this paper, unless they are relevant for the study at hand.
L236: Please provide a reference for the Community Earth System Model. Please also Otherwise, it is impossible to ascertain how the EarthWorks
L243: Why are the MPAS dynamical cores notable? Are they better than the dynamical cores of other models? How is this relevant for your study?
L247: What metric are you using to come to the conclusion that the horizontal grid spacing yields realistic numbers of TCs. What is a realistic number and what is not?
L252: Please define ‘dynamical simulation properties’. Or are you referring to the dynamical core?
L254: What are scenario presets?
L255: So if SST is prescribed, then EarthWorks is an atmosphere-only model with no ocean component. This should be made clearer earlier on.
L256: We already know that SST is important for TC formation. Is it necessary to repeat this?
L258: How can ocean cooling be self-limiting? Firstly, I believe you mean sea-surface cooling. Secondly, the ocean cooling is not limited by the ocean cooling.
L266: What exactly do you mean by ‘likely before most appreciable change in climatic TC influences’? What climatic TC influences are you referring to. What do you define as appreciable change?
L274: Why is ‘Analysis’ capitalized.
L279: What are storm-generating SSTs? This is not a standard term.
L280: ‘Reduce/alter’. Which one is it?
L281: You say that this study ‘constitutes among the first attempts to directly model tropical storms in high resolution under climate intervention, and the first to [do] so utilizing a realistic SAI regime’. Have you discussed the previous attempts and referenced them in the text? What do you mean by a realistic SAI regime?
L284: I do not understand the term ‘storm-generating resolution’. Resolution itself does not generate storms. I think you mean ‘storm-resolving resolutions’.
L289: I am confused as to how one paper (Knaff and Zehr 2007) constitutes ‘the literature’.
L294: How do you determine that ‘TC central pressures in EarthWorks outputs map more closely to a real-world distribution’ compared to max winds?
L322: Please quantify what you mean by the track numbers being ‘virtually unchanged’ with the filtering? Otherwise, please remove this statement.
L328: How does filtering out storms weaker than category 1 increase the reliability of the best track data? Please explain or provide references. I guess you are alluding to the tracking of tropical storms (< category 1) being associated with larger errors compared to stronger storms.
L334: How do you define high fidelity here? From what I see, the model does a reasonable job. ‘High fidelity’ is an overstatement of the model’s ability to capture ‘historical TC behaviour’. The 605 TCs in the model is more than 10% more than the 517 in IBTrACS.
L343: The assertion that the model can simulate the full spectrum of intensity is misleading. The model can simulate the range of central low pressures including for category 5 storms. However, the model does not actually simulate maximum wind speeds corresponding to category 5 storms. Therefore, it is incorrect to claim that the model can simulate the full spectrum of intensity.
L343: What are downward intensity biases
L348: The order of the figures should correspond to the order in which they are introduced in the text. Why are Figure 4 and 6 mentioned before Figure 1, 2 and 3?
L349: Define observational data rounding.
L351: The term ‘great affinity’ again overstates the agreement between model and observations.
L355: Are these differences in lifetime due to the tracking algorithm or the EarthWorks model generating TCs further east compared to observations?
L360: What does CAT1 + IBTrACS mean? Are you averaging over the EarthWorks TCs and IBTrACS TCs?
L371: ‘Donuts’ is not an appropriate term to describe the TCs.
L373: What are ‘fundamental storm physics’?
Figure 1: It is strange to see a U10 plots with wind speeds corresponding to category 3 and be told that one is looking at a category 5 storm. Why do we need to convert pressure to wind speed using the wind-pressure relationship? Does it not make more sense to conduct the whole analysis using the mean sea-level pressure instead. Converting pressure to wind speed adds another level of abstraction to this study and makes it harder to interpret what is actually going on. Also, is there any reason to believe that the current, empirical wind-pressure relationship will hold in the SSP5-8.5 scenario?
L398: You say you see a significant correlation between RI and lifetime maximum intensity. I don’t. Please provide evidence of this statement. Either here or in the supplementary materials.
L401: How can rapid intensification drive storm development? Isn’t rapid intensification storm development itself?
L405: What is ‘modeling/earlier’ literature supposed to mean?
L411: Are you using ‘significant’ in the statistical sense? If not, please remove this word.
L415: What is a parsimonious relationship?
L421: What is a ‘linear part-way state’?
L426: I do not understand what ‘confounded’ means in the context of results.
L428: Why is this result unexpected? Using the term unexpected creates false expectations as to the importance of this result.
L428: You mean section 5 not Chapter 5, right?
L433: You say results of a similar magnitude were found by Bacmeister et al. (2018). What were their results?
Figure 2: The font size is too small.
Figure 4: If you want to compare the track numbers for the different simulations and IBTrACS, would it not be better to plot track density rather than the individual tracks?.
L444: How can you describe the emergence of TC activity in the South Atlantic as novel if TCs have already occurred there in the past? What exactly is the novel aspect here?
L446: Please provide a reference for the lack of TCs forming in the South Atlantic.
L452: Why are you generalising increasing TC activity in the South Atlantic to other ‘regions of the world’ without evidence to support this statement?
L457: Please specify what you mean by thermal energy. Are you referring to higher SSTs? Larger moist static energy? Or larger enthalpy?
L459: I do not agree with the statement that very few places experience an increase in TC frequency. What about the increase in the Pacific evident from the red contours there?
L463: What is the BRACE signal? What does BRACE stand for? Can you quantify the degree of correlation.
L469: How do you define ‘significant density’?
L470: What does ‘69 over the decade’ mean? Is this an increase or a decrease? Also, this number is meaningless without stating how this change relates to the absolute number of TCs in the historical scenario.
L473: Is the result robust? If not, please remove baseless speculation about the result being robust.
L476: What is a consistent correlation across scenarios? Do you mean that the ENSO index is very different between the three simulations, which makes it hard to attribute any differences in TC numbers to ENSO?
Figure 6: Font size is too small.
L478: Please define ‘relative global storm distribution patterns’. ’How can simulations share something that is defined as being relative?
L482: Please quantify what you mean by ‘matching’.
L485-488: You say that translational speeds are largely unchanged. But isn’t largely unchanged the same as saying a small change? ‘Largely unchanged’ is not sufficiently precise a term here.
L498-501: Why do refer to ‘polar amplification’ and ‘mid-level shear’ when you have not shown figures of polar amplification and mid-level shear in your results?
L512: Again, I do not know what heat-engine effects are.
Figure 8: Why are you plotting weekly TC frequency? Monthly frequency is standard and would be easier to compare with other studies?
L516: TC formation is not only dependent on SST, but a range of environmental factors.
L521: You refer to ‘overall hemispheric trends’, but it looks like you are referring to differences not trends.
L525: Why would a change in storm lifetimes affect the monthly distribution of TC frequency?
L526: Change ‘Chapter 5’ to section 5.
L529: What do you mean by ‘dynamically calculated land temperature patterns’?
L532: What do you mean by ‘global storm activity’? Are you referring to TC frequency? Or accumulated cyclone energy (ACE)? And how can you be sure that ‘significant’ increases are likely if only the global surface temperature changes?
L532-534: I do not understand what you are trying to say here.
L535-545: You talk about the effect of changes in SST on the distribution of TCs without any Figure or Table to back this up. Delete this text if you cannot back up these claims.
L549: What is a history-agnostic ENSO index?
L550: Above what SST value is convection permitted?
L553: What do you mean by ‘important new modalities’?
L556: Please provide references for the ocean upwelling argument and the permanent El-Niño arguments.
L559: What exactly is driven by a shift toward much more powerful positive-ENSO events? How can I assert from Figure 9 and 10 that the positive-ENSO events are ‘much more powerful’?
Figure 9: Caption makes no mention that SST anomaly is plotted. Which two simulations are used to calculate the SST anomaly? What does ONI stand for?
L562: What does ‘perhaps entirely’ mean? Please leave out baseless speculation.
L565: Do not ask the reader to check the long-term historical index values calculated in another study by themselves. Provide the comparison in the text.
L569: What does ‘provide an opportunity for this difference in scenario responses’ mean?
L571-576: How is this information about Atlantic TC activity relevant to your results?
L580: Where are these maps that you have produced? Presumably in Figure 11, 12 and 13?
L585: How are the changes ‘profound’?
L593: What is strong local forcing? Forcing of what?
L595-598: Why do you talk about the ‘Australasia storm development region’ and not the other regions. Stick with the global picture or discuss all regions. Why do you use ‘significant’ here if it is not used in the statistical sense?
Figure 13: I find this Figure with a different colorbar scale very misleading. The CIN values are much smaller than the values with the corresponding shades in Figure 11 and 12.
L606: Please be specific what you mean by ‘extreme’ degree of suppression.
L610: You assert that substantial increases in vertical shear in the southern hemisphere ‘can more than explain substantial reduction in SH storm activity’. Yet you cite this result from another study and you do not consider other environmental factors, such as relative humidity, vorticity etc… Please remove the word ‘explain’. You can assert that your results are consistent, but your results do not support the claim that vertical shear changes explain reductions in SH TC activity.
L613: This is another example of a juxtaposition of two words that are meaningless together. What do trends that ‘potentially most exemplify the complexity of cyclogenesis controlling factors’ mean. Either they exemplify the complexity or they do not.
L615: What do you mean by ‘powerfully’?
L618: Define ‘main-season SSTs’
L621: We already know that higher SSTs are more conducive to TC formation. Is it necessary to repeat this again?
L623-624: I don’t see a substantial expansion of a high-CIN region in the western North Pacific. Figure 11 does not display high CIN values here and Figure 13 is misleading with a different colorbar scale. Why do you not apply a consistent definition of what constitutes a high CIN value across all ocean basins?
L626: What do you mean ‘by NH standards’. Please quantify this statement.
L268: Again, ‘could more than …’ is meaningless.
L629: What is the difference between the being ‘likely the primary explanation’ and ‘one possible explanation’. Your wording is overstating the confidence of your results.
L630: What is a dominant explanation? How do you define dominant? How do your results show that this is the case?
L635: I do not understand this sentence at all.
L637-640: This paragraph is not relevant to your results.
L641: You have not examined the results with different paradigms. Instead, you have looked at individual relationships between various variables and processes, such as ENSO, and TC distribution and frequency.
L643: ‘Novel’ here is redundant as results are, by definition, always novel. Please articulate what is the novel contribution of your study to the field.
L643: What are resolved TC physics? And how can a scenario be realistic?
L645-650: I have serious reservations about the novelty of this tracking algorithm. It is not clear why the tracking algorithm features in the conclusions, as it has brought clear benefit to the study.
L651-654: This is not new information and does not belong in the conclusions. Also, you are grossly overstating the agreement between model and observations with the term ‘high fidelity’.
L656: I still do not understand what you mean by a non-linear here. Why would you expect TC frequency to scale linearly global surface temperature anyway?
L661: I am not convinced that computational power was the main limitation of this study. The main limitations stem from the limited analysis, single model, horizontal resolution and any model biases in the EarthWorks model. Please include a more comprehensive and candid discussion of limitations in the main text.
L673-679: I do not see the relevance to your work here.
Citation: https://doi.org/10.5194/egusphere-2024-3526-RC1 -
AC1: 'Reply on RC1', Andrew Feder, 26 Feb 2025
General Response:
So, a lot of the issues here you've pointed out are really fundamentals of how atmospheric science papers should be written, including issues with the non-statistical use of the word "significant," and similar issues, as well as imprecise language when talking about model properties. These I hope I've addressed, I've made the edits suggested in a new draft of the paper, not submitted here in line with the response guidelines, but I've described how I've addressed a lot of those below.
On the bigger-picture stuff, I have tried to get a lot more specific about EarthWorks model components, since this is right now the first EarthWorks paper. I've emphasized that these experiments are following something like a time-slice, storyline modeling approach, and they should be thought of as a first step toward understanding mesoscale and TC behavior arising from feedback algorithm SAI. I've situated algorithmically controlled SAI in the broader research environment, including its upcoming inclusion in GeoMip7.
With respect to the PHAST algorithm, it's true that the detection rules are based on TSTORMS. But, this is something substantially different we've made here. I've toned down describing it as "novel," but these changes solve real problems with TC tracking algorithms while still following very simple heuristics, and they're an important part of the scientific contributions, if any, made by this project.
Specific Comments:
Introduction:
"L17-19: In what way exactly does SAI restore global storm patterns? And what is meant by novel hazards? Please be more specific."
L17-19: Changed end of our intro to be more specific about what changes and what doesn't.
"L22: Strictly speaking, Emmanuel’s theory applies to a mature, axisymmetric TC in a steady state."
L22: TC's have more going on than just an ideal Carnot cycle. But, this provides an important way to understand their potential intensity. I hope "approximated" makes this more epistemologically modest of a claim, but the fundamental idea is important here.
"L29: What do you mean by available energy? Can you define this?"
L29: Available energy in the heat reservoir, Carnot engine sense.
"L41: What is the physical structure of a TC. Do you just mean the structure of the TC circulation? Or are you referring to small-scale processes, such as convection, that are not sufficiently simulated in coarse resolution simulations?"
L41: I guess it's better to say that the convection couldn't be resolved.
"L50: Parsimonious means stingy and unwilling to spend money. However, I have absolutely no idea what a parsimonious relationship is. Do you mean a ‘tenuous’ relationship? Please select an alternative term that is appropriate for the context."
L50: Parsimony is the quality of a scientific explanation reflecting the most straight-forward possible cause and effect relationship. But, it is maybe an uncommon word to use here, so I'll substitute it with some different words.
"L50: What attributes are you referring to with the term ‘TC behaviour’? And how is ‘TC behaviour’ related to model skill? How do you define model skill?"
L50: Added examples of the salient TC properties GCM's can capture, and some parts of large and small scale model dynamics that can influence existing storms.
"L52: Tracking TCs in GCMs is standard practice. There is no need to use the term ‘direct’ here.’ It sounds like you are trying to articulate the advantages of using GCMs that explicitly resolve TCs."
L52: You're right, what I'm really describing here is the advantage of high-resolution GCM's for TC's over the more indirect methods as performed at lower resolution.
"L53: Define the term non-canonical cyclogenesis."
L53: In this case, non-canonical cyclogenesis mostly means storms forming in new places, but I'll let that be covered by the phrase "changes in TC formation areas with shifting global temperatures."
"L66: Having read the paper, I fail to see how the tracking algorithm is novel."
L66: I'll be changing references to our algorithm we used as an adaption or modification of the Vitart et al one.
"L77: If you cannot quantify what you mean by ‘perhaps considerably’, please remove these two words."
L77: No more adjectives here.
"L84: ‘Produced TC tracking results’ sounds like Bhatia et al. (2018) were testing tracking algorithms. However, this is not the case."
L84: Number and intensity of storms, right, not anything about the tracking itself.
"L88: Is it not obvious that TCs arise spontaneously? How else would they arise? I guess you are saying that they are not seeded in the simulation. "
L88: I've changed to model native as our term for these, but I've wanted to emphasize when other studies are using the high-resolution and direct tracking combo, rather than any number of other methods of estimating storm activity or creating synthetic storms.
"L94: What mechanisms for TC intensification are you referring to here? Why would one assume that the mechanisms for cyclogenesis would be similar when there exists no theory that can explain why we see different numbers of TCs from one year to the next?"
L94: I've changed around this whole set of ideas. But basically, the numbers of intense TCs and TCs overall would always go up or down together if we make the assumption that we're not seed limited. I don't believe this is likely true of the real world, but its one of the possible ideas Kerry Emanuel puts forward in his retirement keynote lecture cited in this section, and this co-correlation would be an extrapolation from that.
"L95: Again, I do not understand what is meant by a parsimonious relationship."
L95: Let's just get rid of that whole modifying clause.
"L96: Why do you refer to ‘a physics of TCs’? The field of physics is not different from one regime to another. Are you talking about mechanisms of some sort here?"
L96: Let's probably not refer to "a physics" then.
"L107-115: Yes, it is true that factors such as SST and vorticity feature in genesis indices and that these are empirically derived. Precisely for this reason, they have limited applicability to modelling studies, with the correlation between model TCs and genesis indices varying across models. Please mention this."
L107-115: expanded discussion of their specificity to present observations.
"L114: Reference required for statement that such indices may not hold in future climate."
L114: There's no rigorous way right now for us to presently determine how these will do in a future climate vs. just in modeled future climates. But this statement was maybe too strong, I've softened it.
"L116: Is this correlation positive or negative? What do you mean by a robust correlation?"
L116: a robust positive correlation between vertical velocity and storm formation, stronger than for anything else they tested.
"L133: What are feedback algorithms?"
L133: any number of schemes where, I've added, we adjust the injections over time in response to present conditions.
"L148: Why may the SSP5-8.5 be unlikely? If the scenario is unlikely, why do the authors focus on SAI in this scenario?"
L148: SSP5-8.5 is unlikely if the Paris agreement goals are met. I don't think we're going to go that far, but we're probably significantly more likely to end up using SAI if we do. I've added another citation for a big article about how likely SSP5-8.5 is, but I've mentioned why studying this scenario might be important apart from signal-to-noise reasons.
"L168: What is loft SOO2?"
L168: LaTeX typo on SO2, and loft feels like a good word here since volcanoes shoot sulfur upwards but don't really inject at any particular height.
"L175: Since Kerry Emanuel’s theory of TCs being a carnot engine applies to a steady-state, mature TC, does it really make sense to invoke this theory for TC formation? What is exactly is the heat engine understanding of TC formation and propagation?"
L175: In the cited keynote, Emanuel says roughly, "if you stop throwing lit cigarettes into the forest, you'll stop having fires for a while, but eventually something else might just spark them anyway once there's a higher level of fuel available." When I've quoted the keynote, it's basically to cite Emanuel's own uncertainty about the extent to which TC's are seed limited or "fuel" (potential energy) limited. But I think if I've already referenced this idea, and a lot of this theorizing different relationships among variables under SAI is my own, it's perhaps confusing to keep referring to him speaking here.
"L175: Again, I do not know what a parsimonious relationship is."
"L176-177: Please avoid using phrases such as ‘volcanism/sulfur’ and ‘hypothesized/observed’. As a reader, it is not my job to decide how to apply each word to the context."
L175-177: Removed problematic phrasing
"L179: What do you mean by unexpected mechanism? What is the ‘simple SST-storm relationship’?"
L179: Storm seed number effects could stop lower SSTs from always leading to fewer storms and vice versa.
"L191: Why say ‘may robustly extend to TC activity’ rather than ‘may extend to TC activity’? The word ‘robust’ is automatically negated by the word ‘may’."
"L191: In which hemisphere is the SAI deployed?"
L191: Experiments were done with both NH and SH deployments, and they have counter effects on ITCZ position.
"L196: Why are the results contradictory?"
L196: Not as strong and consistent TC effects as for the SH injection.
"L198: What exactly are heat-engine effects? I have never heard this term before. Do you just mean an increase in SST? If so, then say that."
L198: Done
"L200: You refer to mechanisms without explaining the underlying physics that lead to the correlations that you describe. Without any explanation of the processes behind changes in the ITCZ and TC distributions, one can only really talk of correlations and not mechanisms."
L200: I added another mention of the shear effects of moving the ITCZ. I think I have established this as a mechanism now, though it is true it isn't directly the ITCZ itself effecting TCs.
"L201: Is it not obvious that similar processes affect the distribution and frequency of TCs in the event of volcanic eruptions and aerosol-based interventions. After all, aerosols can be released into the stratosphere in both cases. "
L201: Gone
"L203: Please remove the reference to prediction. The term prediction is reserved for forecasting. You cannot predict storm responses to SAI in the years 2090-2099. Instead you are trying to better understand the response of TC distribution and frequency to SAI."
L203: Changed to just "understand"
"L204: What effects and relationships are you talking about here? Please be more specific."
L204: The point I should have made clearer is that we have a mix of imperfect past proxies, and then more specific modeling scenarios but which are limited by the nature of modeling.
"L205-212: Here you list a series of relationships or processes without justifying why they are being singled out for examination. There is also no description of how exactly you ‘will attempt to link any model-observed changes in TC statistics to one or more of theorized mechanisms’. I fail to see how looking at these various processes in your own simulations will provide a ‘framework’. At the most, looking at these processes will provide insight into the mechanisms behind the changes in TC distribution and frequency linked to SAI in the SSP5-8.5 scenario. The word ‘framework’ overstates the importance of your work."
L205-212: These are the mechanisms/correlations we picked out of the literature, they reflect the concerns the SAI literature has had about ensuring they can recreate a pre-warming climate as closely as possible, and possible climate side effects. There are other possible ways to slice up how to describe different temperature patterns and changes in convection. But, I genuinely do feel that this summarizes the more important ideas at the convergence of the SAI/TC research areas.
"L213-218: Again you are overstating the importance of your own work. I suggest you tone down the claim that you are in a position to ‘create outlines of potential TC hazards with salience well beyond the results of these specific runs’. Firstly, your study is conducted using a single model and single realisations, so the claim of applicability beyond the EarthWorks model is questionable. Secondly, this study is not about characterizing or quantifying TC hazards but the broad changes in TC distribution and frequency. "
L213-218: I hope this is better, the final intro paragraph is now:
'We modify existing heuristics to track TCs occurring in this and other GCMs, and compare the performance of the resulting model-tracker combination to real-world storm data. We then use this baseline to evaluate changes in storm behavior given either advanced climate change or the global climate patterns created by SAI. This includes overall storm number, latitude/basin distributions, life cycle changes, and changes in typical TC intensities. Finally, we examine possible intermediate mechanisms, such as El Niño and changes in regional vertical shear, through which climate and TC formation may interact.'
"L219: Please provide reference for EarthWorks model."
"L220: What are these new multiple techniques? The tracking algorithm itself is modified but not novel."
L219-220: Changed to modified heuristics. The EarthWorks project, we will introduce a bit more in methods, but here with the tracking algorithm its not really that novel.
Methods:
"L229-230: Again, please provide a reference for the EarthWorks model."
L229-230: This is the first journal paper to be published using EarthWorks, our big methods paper will be published when the project is completed at the target resolution. But, I've now cited every component we actually use, our release documentation, and everything that would be necessary to also implement the model configuration we used. Basically, our work here has been making all these different components run together for a long time at higher resolution, without crashing and while shifting a lot of work load onto new GPU systems. The advances made by the EarthWorks project I should really have emphasized are all performance-related, but fundamental model properties should be unchanged for each of these components.
"L231: Mentioning 4-km horizontal grid spacing here is misleading, because the actual grid spacing for the simulations conducted here is ~30 km. Information about project goals do not belong in this paper, unless they are relevant for the study at hand."
L231: 4km mention is removed, we just talk about GSRM resolution more broadly.
"L236: Please provide a reference for the Community Earth System Model. Please also Otherwise, it is impossible to ascertain how the EarthWorks "
L236: This was clever, also, done.
"L243: Why are the MPAS dynamical cores notable? Are they better than the dynamical cores of other models? How is this relevant for your study?"
L243: The main advantage here among other possible ones, is that TCs won't need to take up more cells as they move poleward, as would be the case in a lat-lon grid model.
"L247: What metric are you using to come to the conclusion that the horizontal grid spacing yields realistic numbers of TCs. What is a realistic number and what is not?"
L247: A realistic number is based on best-track data. What do we currently observe in the world each year, and do the storms in the model have similar population properties?
"L252: Please define ‘dynamical simulation properties’. Or are you referring to the dynamical core?"
L252: You're right, I should just say dynamical core here.
"L254: What are scenario presets? "
L254: Like other model software, we have options built in to impose SSP trajectory GHG concentrations from different time periods, but this isn't really worth noting necessarily.
"L255: So if SST is prescribed, then EarthWorks is an atmosphere-only model with no ocean component. This should be made clearer earlier on. "
L255: I'll note this earlier in methods.
"L256: We already know that SST is important for TC formation. Is it necessary to repeat this?"
L256: Removed.
"L258: How can ocean cooling be self-limiting? Firstly, I believe you mean sea-surface cooling. Secondly, the ocean cooling is not limited by the ocean cooling."
L258: Rather, storms that cool the ocean surface might be self-limiting.
"L266: What exactly do you mean by ‘likely before most appreciable change in climatic TC influences’? What climatic TC influences are you referring to. What do you define as appreciable change?"
L266: This is hard, because the period of really good satellite TC records begins after climate change could have plausibly started effecting TC behavior. A more modest thing to say here, is that there's really a lot less carbon in the air already than in 2090 under SSP5-8.5.
"L274: Why is ‘Analysis’ capitalized."
L274: referred to a previous division of this work into three big sections, no longer present. This is a simple typo then, apologies.
"L279: What are storm-generating SSTs? This is not a standard term."
L279: I discuss the idea of threshold temperatures later, but for now, let's say just warm waters.
"L280: ‘Reduce/alter’. Which one is it?"
L280: Alter if we're talking broadly. The general purpose of SAI would be to try to reduce climate-related risks as much as possible, but our scientific question should be how do storms change, not specifically, does this improve things from a human perspective or not.
"L281: You say that this study ‘constitutes among the first attempts to directly model tropical storms in high resolution under climate intervention, and the first to [do] so utilizing a realistic SAI regime’. Have you discussed the previous attempts and referenced them in the text? What do you mean by a realistic SAI regime?"
L281: Added more discussion of the Irvine et al paper, the GeoMIP6 protocol, and the future GeoMIP7 protocol in our introduction. Irvine et al and Jones et al are the two papers we've found that previously simulated TCs under SAI at about our model's resolution. Otherwise, lots of big geoengineering modeling studies, typically at lower resolution, are based on the GeoMIP protocols, which so far haven't included algorithmic injection strategies as in the Tilmes et al 2020 ensemble.
"L284: I do not understand the term ‘storm-generating resolution’. Resolution itself does not generate storms. I think you mean ‘storm-resolving resolutions’."
L284: storm-resolving, that's better
"L289: I am confused as to how one paper (Knaff and Zehr 2007) constitutes ‘the literature’."
L289: They discuss lots of prior WPR work in the Knaff and Zehr paper, but the thing that's important here is that figure cited which illustrates the relationship. I've changed this to just "earlier work" since I don't go into WPRs much otherwise.
"L294: How do you determine that ‘TC central pressures in EarthWorks outputs map more closely to a real-world distribution’ compared to max winds? "
L294: Simply, pressure-based intensity gets us a very realistic humber of storm in each category. Wind speed does not, and based on the Davis paper, really should not in a model at our spacing.
"L322: Please quantify what you mean by the track numbers being ‘virtually unchanged’ with the filtering? Otherwise, please remove this statement."
L322: Made this more descriptive.
"L328: How does filtering out storms weaker than category 1 increase the reliability of the best track data? Please explain or provide references. I guess you are alluding to the tracking of tropical storms (< category 1) being associated with larger errors compared to stronger storms. "
L328: IBtRACS data for weaker systems gets better and more complete over time, but I am more confident in the CAT1+ data.
"L334: How do you define high fidelity here? From what I see, the model does a reasonable job. ‘High fidelity’ is an overstatement of the model’s ability to capture ‘historical TC behaviour’. The 605 TCs in the model is more than 10\% more than the 517 in IBTrACS."
L334: Toned down description here.
"L343: The assertion that the model can simulate the full spectrum of intensity is misleading. The model can simulate the range of central low pressures including for category 5 storms. However, the model does not actually simulate maximum wind speeds corresponding to category 5 storms. Therefore, it is incorrect to claim that the model can simulate the full spectrum of intensity. "
L343: I've changed this to a more modest, direct claim.
"L343: What are downward intensity biases"
L343: Anything that's weakening at least some TCs and stopping us from having all the major storms we would with a better, more realistic model.
"L348: The order of the figures should correspond to the order in which they are introduced in the text. Why are Figure 4 and 6 mentioned before Figure 1, 2 and 3?"
L348: Changed order of figures, now the ones I mentioned first appear first in the document and are numbered first.
"L349: Define observational data rounding."
L349: IBtRACS wind speed data is sometimes but not always rounded up to the nearest 5 knots.
"L351: The term ‘great affinity’ again overstates the agreement between model and observations. "
L351: Changed to, "fairly well."
"L355: Are these differences in lifetime due to the tracking algorithm or the EarthWorks model generating TCs further east compared to observations?"
L355: Looking closely at the dot colors in this region for our new figure 2, there's never a TC-intensity system over land in West Africa, just the minor cyclonic movement that becomes TCs over the ocean. Translational speed differences should really have some effect on actual TC track geometry changes, so I've put the phrase "at least partly" to make clear this isn't the only thing going on, and some other influences are discussed later.
"L360: What does CAT1 + IBTrACS mean? Are you averaging over the EarthWorks TCs and IBTrACS TCs?"
L360: Just means every storm CAT1 and stronger, I've put this in words now in this sentence.
"L371: ‘Donuts’ is not an appropriate term to describe the TCs."
371: I stand by this. I've heard folks use this term at conferences, it's descriptive of the current limitations of GCM TCs, and I think the imagery of it is both accurate and helpful.
"L373: What are ‘fundamental storm physics’?"
L373: Changed to "observed storm behavior," pressure and wind speed vary within and outside the eye wall a lot like in real life.
"Figure 1: It is strange to see a U10 plots with wind speeds corresponding to category 3 and be told that one is looking at a category 5 storm. Why do we need to convert pressure to wind speed using the wind-pressure relationship? Does it not make more sense to conduct the whole analysis using the mean sea-level pressure instead. Converting pressure to wind speed adds another level of abstraction to this study and makes it harder to interpret what is actually going on. Also, is there any reason to believe that the current, empirical wind-pressure relationship will hold in the SSP5-8.5 scenario? "
Figure 1 (now figure 5): I have also wrestled with this. Kerry Emanuel advocates categorizing TC intensity entirely based on central pressure. At the same time, the other studies in this area have described storm intensity in terms of wind speed or derived variables. Commonly used metrics of storm energy like ACE, PDI, etc are explicitly formulated in terms of max wind speeds.
Using the formula to convert between pressure and wind has sometimes been a headache, but other than just imposing an arbitrary wind correction function specific to our current model and resolution, pressure-based categorization is the only way to capture the full range of intensities that I think do meaningfully occur in our outputs. Yet, to be comparable with the existing set of TC and climate change studies, we have to use the CAT1-5 system, which is wind speed based. Some of these other modeling projects have not had this problem because, in CAM5 or HIFLOR, they've relied on GCMs that do produce CAT5 storms based on wind speed, but the Davis paper we cite shows this is not physically accurate at the resolutions they've used anyway, and relying on those wind outputs is just as arbitrary as this. So, we've ended up at this compromise.
The model-native U10 plot is important to show what our conversion really ends up meaning practically. That what we classify as a CAT5 storm based on pressure really does look like one in its wind field in a lot of ways, in terms of overall size, how clear the eye geometry is. It's just missing accurate max wind values, which we've kind of interpolated for it.
I'll add a paragraph about this to the TC tracker subsection.
"L398: You say you see a significant correlation between RI and lifetime maximum intensity. I don’t. Please provide evidence of this statement. Either here or in the supplementary materials."
L398: Shortened this paragraph and focused on the idea that, basing these cursory statistics on the ones Bhatia et al create in their study, we immediately do not see storms experiencing RI as becoming that much stronger than ones that don't. The point here is kind of more what we don't see.
"L401: How can rapid intensification drive storm development? Isn’t rapid intensification storm development itself?"
L401: Not as much a driver of intensity, rather.
"L405: What is ‘modeling/earlier’ literature supposed to mean?"
L405: I'm really mostly talking about the modeling literature, so let's just say that.
"L411: Are you using ‘significant’ in the statistical sense? If not, please remove this word."
L411: Done.
"L415: What is a parsimonious relationship?"
L415: I promise parsimonious is a real word in the philosophy of science, meaning an explanation requiring the fewest possible logical steps. I have overused it here, and in this case, it can be replaced witj "simplest."
"L421: What is a ‘linear part-way state’? "
L421: It's not just an intermediate state between the other two scenarios, there's something unique going on.
"L426: I do not understand what ‘confounded’ means in the context of results."
L426: influenced, particularly by the fact that each of our time slices could be highly affected by, say, one big El Nino cycle or similar event.
"L428: Why is this result unexpected? Using the term unexpected creates false expectations as to the importance of this result."
L428: It's not what I expected, but I guess there's been no big scientific consensus on what should happen here, so I've removed that.
"L428: You mean section 5 not Chapter 5, right? "
L428: Indeed.
"L433: You say results of a similar magnitude were found by Bacmeister et al. (2018). What were their results?"
L433: They divide the basin into 4 degree squares, produce a map of density of storm track through each, and find that the biggest Atlantic densities decrease from 18 storm hours/cell/year to about 8-10 hours.
"Figure 2: The font size is too small."
Figure 2: Made the fonts bigger
"Figure 4: If you want to compare the track numbers for the different simulations and IBTrACS, would it not be better to plot track density rather than the individual tracks?. "
Figure 4: Track density differences between IBtRACS and our model outputs will be heavily influenced by storm lifetime and translational speed biases. I think these maps of individual trajectories show important aspects of storm behavior, formation regions, where the storms end, the paths they collectively take, that are important and cover stuff the next plot with the density differences can't.
"L444: How can you describe the emergence of TC activity in the South Atlantic as novel if TCs have already occurred there in the past? What exactly is the novel aspect here?"
L444: It's novel in that this basin starts regularly seeing named-intensity TCs, versus not at all in the other simulations we do, and only once ever in the entire historical record. I've added this in the phrase "regular activity."
"L446: Please provide a reference for the lack of TCs forming in the South Atlantic. "
L446: The Pezza reference I've used again, also includes all the reasons for this.
"L452: Why are you generalising increasing TC activity in the South Atlantic to other ‘regions of the world’ without evidence to support this statement?"
L452: Got rid of this for simplicity.
"L457: Please specify what you mean by thermal energy. Are you referring to higher SSTs? Larger moist static energy? Or larger enthalpy?"
L457: Higher SSTs.
"L459: I do not agree with the statement that very few places experience an increase in TC frequency. What about the increase in the Pacific evident from the red contours there?"
L459: Now noted there.
"L463: What is the BRACE signal? What does BRACE stand for? Can you quantify the degree of correlation."
L463: BRACE is the "Benefits of Reducing Anthropogenic Climate changE" paradigm, and it studies how climate effects differ between RCP 8.5 and RCP 4.5. This is about 5-6 degrees of warming versus like 2-3 degrees at century end, but I've decided to get rid of this whole paragraph for space and because, as I said there before, the situations are not in fact directly comparable because of SAI specific effects.
"L469: How do you define ‘significant density’?"
L469: Changed to everywhere they typically occur each year presently, which excludes our South Atlantic basin and tracks in SSP5-8.5 that manage to travel southeast of Australia.
"L470: What does ‘69 over the decade’ mean? Is this an increase or a decrease? Also, this number is meaningless without stating how this change relates to the absolute number of TCs in the historical scenario."
L470: Added Control Run number here.
"L473: Is the result robust? If not, please remove baseless speculation about the result being robust."
L473: Removed this clause.
"L476: What is a consistent correlation across scenarios? Do you mean that the ENSO index is very different between the three simulations, which makes it hard to attribute any differences in TC numbers to ENSO?"
L476: It's that supressed Atlantic activity and enhanced east Pacific activity both occur very strongly with the very strong EL Nino response in Intervention, but east Pacific activity is not really enhanced with just advanced warming, while Atlantic activity is still quite suppressed, means something more complicated is going on here.
"Figure 6: Font size is too small."
Figure 6: Fixed this.
"L478: Please define ‘relative global storm distribution patterns’. ’How can simulations share something that is defined as being relative?"
L478: Let's just get rid of this whole paragraph since we added length elsewhere and the same trends are captured on a basin-by-basin scale elsewhere.
"L482: Please quantify what you mean by ‘matching’."
L482: Removed
"L485-488: You say that translational speeds are largely unchanged. But isn’t largely unchanged the same as saying a small change? ‘Largely unchanged’ is not sufficiently precise a term here."
L485-488: Now indicated more specifically here.
"L498-501: Why do refer to ‘polar amplification’ and ‘mid-level shear’ when you have not shown figures of polar amplification and mid-level shear in your results? "
L498-501: Polar amplification is one of the three degrees of freedom the SAI control algorithm used by Tilmes et al specifically tries to limit, and a major difference between the between the SAI climate they produce and the baseline SSP5-8.5, although we don't quantify here the amount of change in this. And in the present mid-level shear is what breaks storms apart as they north and south, but I don't show that's also what happens in our models. So, I've removed both these ways of describing what's happening to storms. Now it's just higher SSTs at mid latitudes, and whatever it is that kills storms as they travel poleward that isn't quite overcome with warming, seemingly.
"L512: Again, I do not know what heat-engine effects are."
L512: Changed to just ocean heat here.
"Figure 8: Why are you plotting weekly TC frequency? Monthly frequency is standard and would be easier to compare with other studies?"
Figure 8: Changed to monthly.
"L516: TC formation is not only dependent on SST, but a range of environmental factors."
L516: I'll just remove this small paragraph, it kind of just restates something you can see in the figure.
"L521: You refer to ‘overall hemispheric trends’, but it looks like you are referring to differences not trends."
L521: Let's say hemisphere-level results here.
"L525: Why would a change in storm lifetimes affect the monthly distribution of TC frequency? "
L525: If storms were forming at the same time, but each lived a few days longer, say, median storm occurrence dates would also stretch back that far. I've added a fuller explanation of this possibility.
"L526: Change ‘Chapter 5’ to section 5."
L526: Changed to "Bulk Variable Analysis."
"L529: What do you mean by ‘dynamically calculated land temperature patterns’?"
L529: Basically that these came out of EarthWorks, changed to that.
"L532: What do you mean by ‘global storm activity’? Are you referring to TC frequency? Or accumulated cyclone energy (ACE)? And how can you be sure that ‘significant’ increases are likely if only the global surface temperature changes?"
L532: Changed to storm frequency, cited our SST threshold paper again (Tory and Dare), and removed non-statistical use of word "significant." But, increasing SSTs, all else equal, should enable more TC seeds to form into storms.
"L532-534: I do not understand what you are trying to say here."
L532-534: Let's just get rid of this.
"L535-545: You talk about the effect of changes in SST on the distribution of TCs without any Figure or Table to back this up. Delete this text if you cannot back up these claims."
L535-545: Added global temperature change maps (which I previously got rid of because I thought my paper was running too long). I've also changed the phrasing here, just to say that SST changes must be acting counter to some other inhibition on TC development.
"L549: What is a history-agnostic ENSO index?"
L549: Doesn't depend on a long-term climatology of, say, the Nino 3.4 region.
"L550: Above what SST value is convection permitted?"
L550: As described by Williams, and now also here, they decided that because deep convection in one longitude has to be balanced by downwelling elsewhere, the actual global mean surface temperature in a given latitude is a good proxy for a convective threshold.
"L553: What do you mean by ‘important new modalities’? "
L553: I've deleted that, but basically if we were just judging ENSO strength by temperature in the ENSO region, we'd have trouble telling what's, say, a strong El Nino, and what's just a result of global temperature changes manifest in the same region.
"L556: Please provide references for the ocean upwelling argument and the permanent El-Niño arguments."
L556: The mechanisms behind permanent El Nino don't really matter here, but Pacific temeperature patterns like this are what other research describes as a permanent El Nino, see the forcings used in the Duque-Villegas et al paper.
"L559: What exactly is driven by a shift toward much more powerful positive-ENSO events? How can I assert from Figure 9 and 10 that the positive-ENSO events are ‘much more powerful’?"
L559: The ELI index value mean is much higher for the whole decade. But this is driven by very large temporary states where the ELI index moves east. ELI is a fairly good proxy for El Nino as measured by SSTs in the present climate. And so using it here, when we get those big positive monthly values in what's now figure 11, those are powerful El Nino events.
"Figure 9: Caption makes no mention that SST anomaly is plotted. Which two simulations are used to calculate the SST anomaly? What does ONI stand for?"
Figure 9: Done.
"L562: What does ‘perhaps entirely’ mean? Please leave out baseless speculation."
L562: Removed this phrasing.
"L565: Do not ask the reader to check the long-term historical index values calculated in another study by themselves. Provide the comparison in the text."
L565: Added comparison values.
"L569: What does ‘provide an opportunity for this difference in scenario responses’ mean?"
L569: Basically just that these intermediate things are how ENSO affects TC activity in the region, but we don't need that clause.
"L571-576: How is this information about Atlantic TC activity relevant to your results?"
L571-576: This is important because it shows that, while El Nino events suppress Atlantic TC activity, this also seems to hold for permanent El Nino/El Nino like warming too. This isn't dependent on any mechanisms that can only be temporary.
"L580: Where are these maps that you have produced? Presumably in Figure 11, 12 and 13?"
L580: Yup, now 12 13 14, now indicated.
"L585: How are the changes ‘profound’?"
L585: They just appear big, got rid of that adjective.
"L593: What is strong local forcing? Forcing of what?"
L593: Basically local changes in shear between SSP5-8.5 and control, changed to just no strong local increase.
"L595-598: Why do you talk about the ‘Australasia storm development region’ and not the other regions. Stick with the global picture or discuss all regions. Why do you use ‘significant’ here if it is not used in the statistical sense?"
L595-598: Got rid of this altogether.
"Figure 13: I find this Figure with a different colorbar scale very misleading. The CIN values are much smaller than the values with the corresponding shades in Figure 11 and 12."
Figure 13: I've noted this change in colorbar more directly, and also changed the color here to orange instead of red. But this is world's most active storm area, I feel like I do need this third focus graph.
"L606: Please be specific what you mean by ‘extreme’ degree of suppression."
L606: Let's not call it "extreme," although it is our biggest basin-level change.
"L610: You assert that substantial increases in vertical shear in the southern hemisphere ‘can more than explain substantial reduction in SH storm activity’. Yet you cite this result from another study and you do not consider other environmental factors, such as relative humidity, vorticity etc… Please remove the word ‘explain’. You can assert that your results are consistent, but your results do not support the claim that vertical shear changes explain reductions in SH TC activity."
L610: This can be, "well in line with."
"L613: This is another example of a juxtaposition of two words that are meaningless together. What do trends that ‘potentially most exemplify the complexity of cyclogenesis controlling factors’ mean. Either they exemplify the complexity or they do not."
L613: I've changed this to, "illustrate competing basin-level influences."
" L615: What do you mean by ‘powerfully’?"
L615: We'll change this to, high ENSO indices.
"L618: Define ‘main-season SSTs’"
L618: Aug-Oct
"L621: We already know that higher SSTs are more conducive to TC formation. Is it necessary to repeat this again?"
L621: Removed.
"L623-624: I don’t see a substantial expansion of a high-CIN region in the western North Pacific. Figure 11 does not display high CIN values here and Figure 13 is misleading with a different colorbar scale. Why do you not apply a consistent definition of what constitutes a high CIN value across all ocean basins?"
L623-624: There's definitely an expansion visible in figure 12, though I've removed substantial.
"L626: What do you mean ‘by NH standards’. Please quantify this statement."
L626: There's bands of very high shear in the southern ocean, but I've removed this qualifier.
"L268: Again, ‘could more than …’ is meaningless."
L628: Removed.
"L629: What is the difference between the being ‘likely the primary explanation’ and ‘one possible explanation’. Your wording is overstating the confidence of your results."
L629: Removed.
"L630: What is a dominant explanation? How do you define dominant? How do your results show that this is the case?"
L630: Changed to overarching, but big positive ENSO shifts are consistent with lots of the regional condition changes that end up affecting TCs globally, and we do show that above.
"L635: I do not understand this sentence at all."
L635: Removed the last sentence, big idea of this conclusion is that ENSO effects are the main things changing TC distributions globally between our Geoengineering and Control scenarios, but not the only thing, especially for the east Pacific.
"L637-640: This paragraph is not relevant to your results."
L637-640: Removed.
"L641: You have not examined the results with different paradigms. Instead, you have looked at individual relationships between various variables and processes, such as ENSO, and TC distribution and frequency.
L643: ‘Novel’ here is redundant as results are, by definition, always novel. Please articulate what is the novel contribution of your study to the field.
L643: What are resolved TC physics? And how can a scenario be realistic?"
L641-643: I think this is better. Maybe we haven't used any real new analytical paradigms, but certainly this is where geoengineering research has advanced to now, and this is the first study combining such with TC tracking.
"L645-650: I have serious reservations about the novelty of this tracking algorithm. It is not clear why the tracking algorithm features in the conclusions, as it has brought clear benefit to the study."L645-650: We've created a tracking algorithm based on really simple, easily describable rules like TSTORMS, but which also has features to let storms be detected much closer to each other in space without interference, exclude subtropical systems, and have intensity measured based on pressure instead of wind speed, as called for in the Davis (2018) paper. It's about as different from TSTORMS as other tracking algorithms are from each other, though that was our basis to start. I stand by the importance of the work I did looking through the literature, talking with folks about problems with TSTORMS and similar algorithms, and I feel this is in fact an important contribution.
"L651-654: This is not new information and does not belong in the conclusions. Also, you are grossly overstating the agreement between model and observations with the term ‘high fidelity’."
L651-654: I've changed high fidelity to good agreement. This does belong in the conclusion because this is the first paper using EarthWorks for research, though others at higher resolution are on the way. The fact that this suite of components, together, has done so well representing the world's storm basin geometries and overall activity in our Control Run is a very good sign for future experiments trying to study the intersection of weather and climate with EarthWorks, I stand by this paragraph.
"L656: I still do not understand what you mean by a non-linear here. Why would you expect TC frequency to scale linearly global surface temperature anyway?"
L656: Here, basically the big thing is that SAI doesn't return us to the global TC profile we had before warming. It's really this third, other tempestology, perhaps because of side effects of SAI or stuff it fails to prevent, like permanent El Nino.
"L661: I am not convinced that computational power was the main limitation of this study. The main limitations stem from the limited analysis, single model, horizontal resolution and any model biases in the EarthWorks model. Please include a more comprehensive and candid discussion of limitations in the main text."
L661: Computing power is what's limited our ensemble size, simulation run time, and thus many forms of analysis that would depend on having more data. I've expressed this relationship more directly in this paragraph now. And also, a new paragraph in Section 2 about our lack of an ensemble here.
"L673-679: I do not see the relevance to your work here."
L673-679: Replaced our last paragraph:
'Thus, in both the specific questions we have sought to address here, and the tools and resources we have marshaled to do so, this work can only be understood as the very beginning of efforts to understand how different forms of targeted SAI implementation could impact TC and other mesoscale event risk. As other researchers pursue impacts research in this area, perhaps with new, standardized protocols \citep{GeoMIP7}, they will face similar methodological and other challenges, and we hope that this could serve as a starting point for developing techniques and research questions in this space.'
Citation: https://doi.org/10.5194/egusphere-2024-3526-AC1
-
AC1: 'Reply on RC1', Andrew Feder, 26 Feb 2025
-
RC2: 'Comment on egusphere-2024-3526', Anonymous Referee #2, 17 Jan 2025
In this paper, the authors "describe" (for the reason why this is in quote, see the rest of the review) the results of a set of simulations at high resolution (30km) and investigate Tropical Cyclones changes under global warming and under SAI. I am not a TC expert, therefore there are some part of the manuscript I am less likely to have good insight into, so I hope other reviewers can make up for my lack there (such as on Section 2.3).
I think the paper has the potential eventually to be very good, but there is a rather large problem that needs addressing before anything else, and that makes this paper unpublishable at this stage: EarthWorks is announced and praised, and TC results are validated, but there are no references in the Methods about the model itself. No effort is done to show results for the surface climate (what's the warming in EarthWorks under the GHG scenarios? The first phrase where warming is discussed is "In line with the bulk of prior modeling/earlier literature, severe warming in EarthWorks yields a world with fewer TCs, but a greater number of CAT5 and outlier/extreme intensity storms" but... there's no discussion of how much warming there is? How does the ITCZ look like? Do clouds make sense? Does the model has a stratospheric circulation, or if the aerosols are prescribed, what radiative code is used? What's the resulting TOA imbalance?) If this is a new model, then there needs to be a serious validation behind that can't jump straight to the TC results. The only "paper" cited about EarthWorks is a proposal to NSF as a PDF: to say this is not best practice is definitely not enough. I can't judge the maps I see about shear if I can't see what the surface temperature in this model is. The authors also don't explain how they use the forcing from CESM2: sometimes they talk about SO2, but does it means you're using the SO2 injection locations from the Tilmes et al. (2020) paper, or the aerosols forcing in 3D? How is it assimilated, and have you verified it works? If so, why are you not showing this?
In no way the way the authors present their data about a new model (also considering the lack of a data availability statement about the models result, not to mention the code) is in line with the practices outlined by ESD (or by any other reputable journal in 2025): https://www.earth-system-dynamics.net/policies/data_policy.html Their description of EarthWorks is non-existent.
With all this said, I really can't recommend anything more than a sound rejection at this point. I have other comments about the manuscript and some of the analyses (most of them positive!) but I don't think I can engage with a paper that ignores any guideline set in the last decade about sharing underlying data, model validation and transparency. If the authors have a manuscript of validation of EarthWorks (which I didn't find also by Googling, at most I found a GitHub page the authors didn't even bother adding here), they should cite it, and they should explain how the forcing from SAI works in their model (if this is a new model, even if based on CESM, they should show it in depth). Only after all of this can this manuscript, which is downstream of those, be assessed for its scientific merit.
Citation: https://doi.org/10.5194/egusphere-2024-3526-RC2 -
AC2: 'Reply on RC2', Andrew Feder, 26 Feb 2025
So, this is a little difficult. This is the first paper describing EarthWorks results because it's a preliminary set of experiments before we get to our target resolution. But, all the components used here (MPAS land, ice/ocean, and atmosphere, the community land model) exist independently, and could be used all together without EarthWorks. They would just run very slowly, largely on CPUs. The important thing noted here now, in my latest set of edits to the paper, is that in terms of outputs, EarthWorks is identical to the sum of these existing components. This may not be enough for you, understandably, in which case the only remedy would be to only publish this paper later once there's a big methods paper for the EarthWorks configuration describing a lot more of the relevant model/climate properties. But as you say, that can't really be what this paper is, short of adding really a lot more stuff with a different focus.
In terms of stuff that is immediately fixable, I've added temperature delta maps (that I previously removed for space reasons). I try to explicitly state that we are forcing SO2 fluxes, these undergo chemical changes, and then do eventually fall out. EarthWorks model code is in the EarthWorks Github, now in the code availability section alongside the tracking algorithm code.
Citation: https://doi.org/10.5194/egusphere-2024-3526-AC2
-
AC2: 'Reply on RC2', Andrew Feder, 26 Feb 2025
Interactive computing environment
Code and Sample Data for PHAST Algorithm Andrew Feder https://github.com/afeder17/PHAST
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
299 | 71 | 17 | 387 | 16 | 13 |
- HTML: 299
- PDF: 71
- XML: 17
- Total: 387
- BibTeX: 16
- EndNote: 13
Viewed (geographical distribution)
Country | # | Views | % |
---|---|---|---|
United States of America | 1 | 178 | 47 |
France | 2 | 20 | 5 |
China | 3 | 17 | 4 |
Germany | 4 | 17 | 4 |
United Kingdom | 5 | 17 | 4 |
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1
- 178