| 22 May 2026
Convective activity within a tropical cyclone undergoing extratropical transition over a warmer ocean
Abstract. The Northeastern Atlantic basin is a region where the number of cyclones with tropical features can increase in the future due to anthropogenic climate change, a particularly important concern given the region’s vulnerability to such systems. This work analyses the influence of warmer sea surface temperatures (SSTs), expected in future climates, in the convective activity of Tropical Storm Delta. Delta, which caused strong damage over the Canary Islands (Spain) in November 2005, is representative of a tropical cyclone (TC) experiencing an extratropical transition (ET) on its path to western Europe. Two simulations of the storm were performed with the high-resolution atmospheric numerical model HARMONIE-AROME: a control simulation with initial and boundary conditions from the ERA5 reanalysis, and a warm simulation where a uniform perturbation of +2 °C was added to the SSTs surrounding the cyclone. The convective activity was analysed only in the convective cells near the cyclone's centre, employing the cloud tracking package Tobac, based on brightness temperature. Results show that increases in low-level water vapor flux, together with lower LCL and LFC levels and increased CAPE, create an environment more favourable for the development of deep moist convection in the warmer ocean simulation. These thermodynamic changes lead to more frequent intense moist updrafts and a larger number of convective cells associated with the cyclone, with greater vertical extent and higher precipitation rates. Consequently, Delta becomes a more intense and deeper TC, driven by latent heat release, reaching hurricane status. Later, Delta’s ET starts earlier and gets extended over time, while turning notably more severe too. These results may contribute to a better understanding of the behaviour of convection within cyclones with tropical characteristics affecting the Macaronesia and Western Europe under future climates.
The paper studies an important and useful question. Tropical Storm Delta is a good case because it affected the Canary Islands and because it had tropical and extratropical characteristics. The use of HARMONIE AROME, Cyclone Phase Space, and Tobac gives a useful framework for studying convection near the cyclone centre.
However, the manuscript is not ready for publication in its present form. The main result is physically possible, but the evidence is not strong enough for the strength of the conclusions. The experiment changes only SST by plus 2°C and keeps the atmosphere unchanged. This is useful as a sensitivity test, but it is not a future climate simulation. The paper often writes as if the result is a future climate projection. This is too strong. In addition, there is also no ensemble approach, and the control simulation is not validated enough. The interpretation of latent heating, earlier extratropical transition, and warm seclusion also needs more direct diagnostics.
The paper can become a useful contribution, but it needs major revision. The authors should reduce the claims, improve the physical diagnostics, and better validate the control simulation.
General comments
1) The authors add +2°C to SST but keep the atmospheric initial and boundary conditions unchanged. In a real warmer climate, the atmosphere would also be warmer and more humid, and the vertical stability, humidity profile, tropopause height, pressure field, and vertical wind shear could change. With only SST changed, the model starts with a strong ocean atmosphere imbalance. This can artificially increase surface fluxes, CAPE, and convection. This design is acceptable for an idealized SST sensitivity experiment. But it is not enough to claim that the results represent a future climate response. I would either add a consistent atmospheric perturbation following a Pseudo Global Warming method, including temperature, humidity, pressure, and lateral boundary changes, or reframe the whole study as a pure SST sensitivity experiment. In that case, it would be better to remove or weaken all sentences that present the results as a future climate projection. The abstract, discussion, and conclusion must clearly say that only SST was changed and that the results show a conditional sensitivity, not a full future scenario.
2) The study uses one control run and one +2°C run. This is not enough to support strong statements about future cyclone risk, storm severity, or impacts over western Europe. The warmer simulation also changes the storm track and the timing of the interaction with the upper trough. Therefore, changes in wind and rainfall are not only due to thermodynamics. They also come from different track, different shear, different dry air interaction, and different phase of the cyclone. I think the best solution is to add more simulations. For example, +1°C, +3°C, different initial times, or a small ensemble with perturbed initial conditions. If this is not possible, the authors must clearly say that the results are based on one deterministic sensitivity experiment and should not be generalized. The authors should also compare the simulations by cyclone phase, not only by clock time. For example: tropical phase, start of ET, middle of ET, end of ET. They should show SST below the storm, vertical wind shear, dry air position, surface fluxes, and upper level trough interaction along both tracks.
3) The paper says that warmer SST increases moisture, lowers LCL and LFC, increases CAPE, strengthens updrafts, increases latent heat release, and then deepens the cyclone. This chain is plausible, but the paper does not directly diagnose latent heating, diabatic heating, PV changes, or a heat budget. The claim that latent heat release drives the stronger cyclone is too strong without these diagnostics. The same is true for the claim that earlier ET is caused by a more intense and vertically extended cyclone. The paper needs to show the upper level flow, shear, and PV evolution more clearly. The authors can add direct diagnostics of diabatic heating or latent heating, vertical sections through the cyclone, surface fluxes, and potential vorticity fields. The warm seclusion idea is interesting, but it needs direct evidence. It should not be based only on CPS.
Specific comments
Abstract: The abstract should clearly state that only SST was changed. It should not sound like a full future climate simulation. The sentence saying that Delta becomes a hurricane must clarify the wind averaging issue. The Saffir Simpson scale uses 1 minute wind, while the model gives 10 minute wind. The final sentence about future climates in western Europe should be weaker. I would say that the experiment suggests a possible sensitivity to warmer SSTs, not that it proves a future response.
Introduction: The introduction gives good motivation, but it moves too fast from one Delta case to the whole western Europe region. I would keep the scope narrower. The paper should better explain why Delta is a good representative case. Is it representative of many future systems, or only a useful example?
Data and Methodology
The +2°C SST perturbation needs more justification. The authors say it comes from a 30 year trend, but the trend calculation, uncertainty, and method are not shown in the main paper. A linear extrapolation for many decades is uncertain. The authors should show the trend analysis or move the claim to a weaker form.
Could you please explain whether the SST perturbation is uniform in space and time. In other words, if SST is updated during the run or only at the initial time and boundaries.
Please give more information about HARMONIE AROME cy43h1. State if this version differs from the standard version and which physics options are used. Also it would be better to include a map showing full model domains or you can simple overlay domains in Figure 1.
Could you please discuss spin up more clearly? The grey area in Figure 2 shows spin up, but it is not clear whether spin up hours are removed from all analyses.
The authors use all grid points inside all detected convective cells and apply Mann Whitney and Fisher tests. I couldn’t get the point here because grid points inside one cloud are not independent. Nearby time steps are also not independent. This means that the number of samples is artificially large and the p values are too optimistic. Therefore, the statement that the large sample size increases statistical robustness is not robust in this context. The authors should redo the statistical analysis using more independent samples such as cell mean values, hourly cyclone mean values, object based values, or block bootstrap by time. They can also bootstrap by convective object.
The Gaussian filter with sigma equal to 5 needs units. Is this 5 grid points, 5 km, or another scale? This matters because the filter can remove or modify real pressure structures.
The 5° by 5° tracking box may reduce the influence of other systems, but it does not fully avoid it. Other pressure systems can still affect the pressure field. The authors should explain this more carefully.
Results and Discussion
The earlier ET in the warmer run needs more support. I would be more diagnostic to add vertical wind shear, upper level PV, and outflow diagnostics.
The comparison with Ophelia is interesting but speculative. The paper has not shown that convective asymmetry is the key reason for the different ET response. This comparison should be weakened or supported with more analysis.
I couldn’t get the point by “instant warm seclusion” and how it differs from a classical warm seclusion, and what fields prove it in Delta. The warm seclusion discussion is a bit unclear. Please improve it.
The control run has a track bias at the beginning and does not reproduce the observed pressure fall very well. The paper says HURDAT may have uncertainty, but this does not remove the need for validation. I think the control simulation should be compared with independent data such as satellite brightness temperature, observed precipitation where possible, and ERA5. The fact that the +2°C run matches the best track better than the control run is important. The authors should discuss why this happens and what it means for the control experiment.
The use of Tobac is interesting and could be novel for a tropical cyclone case. But the method needs stronger validation. Brightness temperature alone can include anvils and cirrus. I guess it does not always identify active convection. The authors should show examples where Tobac objects are overlaid on BT, reflectivity, vertical velocity, and precipitation.
The sensitivity tests should be shown more clearly in the main text. A table with main metrics for all 16 threshold and step choices would help.
The area of convective cells is highly threshold dependent. Therefore, conclusions about area, fragmentation, and organization should be more cautious.
The phrase “convective initiation” is not fully correct if a cell is selected because it exceeds 35 dBZ at any time. It is better to call them “convective cells with strong reflectivity”. and why was 35 dBZ chosen for this storm and this model?
The paper says lower LCL enhances latent heat release. This is too simple. Lower LCL means condensation begins lower. It can help parcels reach saturation, but it does not by itself prove larger total latent heat release. As I mentioned in my general comments, the authors should use more diagnostic analysis to show the linkage more clearly.
I couldn’t understand the units of WVFlux. If q is included, the unit cannot be simply m s⁻¹ without explanation. Am I missing something? In a similar manner, the unit of wq also needs checking. If it is vertical velocity times specific humidity, the unit should be explained.
The paper reports precipitation increases mostly as percentage changes. It would be useful if the authors can also give absolute values. What are the total accumulated precipitation in CSIM and plus 2SIM?
The highest precipitation occurs in the second half of the track during ET. This is important and needs more discussion. Why is the ET phase wetter than the tropical phase? Is it due to stronger baroclinic lifting, track change, moisture transport, or slower motion?
I think the impact discussion is too strong. One idealized case cannot prove that western Europe will face more violent and damaging cyclones. The conclusion should present this as a possible hazard signal that needs more study.
I would be careful with hurricane categories. The model gives 10 minute sustained winds, while the Saffir Simpson scale uses 1 minute sustained winds. The authors convert HURDAT winds for comparison, but they also need to be careful when assigning categories to model winds. If model winds are converted to 1 minute values, please state the conversion. If not, avoid category labels or use them only as approximate.
Conclusions:
The conclusion should be shorter and more careful. The last sentence about urgent political action is not appropriate for this paper. The study is one idealized sensitivity test. It can motivate further work and risk assessment, but it cannot support such a broad policy statement. A better conclusion would say that warmer SSTs can strongly modify Delta in this model experiment, but more cases, ensembles, and full climate perturbations are needed before general statements about future western Europe risk. The authors should also clearly state the main limitations in the conclusion: one case, one model, no ensemble, SST only perturbation, and object tracking uncertainty.
Figures
Figure 1:
Panel a legends and labels are too small.
Panel c is described as the initial synoptic situation on 25 November, but the simulation starts on 26 November. Is there any specific reason to start on 26 November?
Figure 3:
If the figure is used to support asymmetry, add a metric or annotation showing the asymmetric convective region.
Figure 4:
I think the figure is overloaded. I would split it into two figures: one for thermodynamic variables and one for convective response variables.
Figure 5:
It would be useful to add cyclone tracks.
Figure 6:
The white stars marking ET onset are not easy to see. Better to use a clearer symbol or outline.
Table 1:
“Cloud Top” to “Cloud top height”.
You mean “precipitation intensity” for TP? Or total precipitation of 10-minutes?
“1 hour Total Precipitation” to “1 hour precipitation”.
I would check the units of wq and WVFlux.
Editorial issues:
Line 48: “up and coming years” to “coming decades”.
Line 112: “synoptical systems” to “synoptic systems”.
Line 120: “start the ET” to “start of ET”.
Line 442: “should be had in consideration” to “should be considered”.
Line 455: “a warmer SSTs” to “warmer SSTs”.
Line 457: “impactful” to “damaging” or “important”.
Line 467: “perturbate” to “perturb”.
Please check the numbering of sections. The paper has “3 Summary and Conclusions” after Section 3 Results and Discussion. It should be Section 4.