Review of the article "Mediterranean Tropical-Like Cyclones forecasts and analysis using the ECMWF Ensemble Forecasting System (IFS) with physical parameterizations perturbations" by Saraceni et al.

I read the manuscript with great interest. I appreciated the methods and richness of the modeling framework. However, my opinion is that the paper is not ready for publication. In terms of presentation, English and technical language need to be improved. I strongly suggest careful editing throughout the text. Especially awkward use of technical language leads to uncertainty about several concluding remarks. Please also refer to several specific comments up to line 385. In addition to the presentation, the introduction is not deepening into the subject and it seems that this extends to the interpretation and discussion of the results. In fact, important late advances in the state of the art are not taken into account and relationship between cyclones development and underlying processes is not adequately demonstrated. Finally, I believe that the scope of the study is not specific enough (see also specific comments). Actually the experiments might tell us more about the sensitivity of the cyclones to diabatic processes and less about the role of SPP in forecasts. To complete the latter, probably you should also use SPP+initial condition perturbations as it would be done in actual operational forecast conditions. 

Given the important methods my suggestion to the authors is to withdraw and resubmit. Therefore, they could reorganise and be more specific on their objectives. For instance, they could break down cyclones into processes, discuss accordingly which ones are important for the different cases and make a more efficient linking with cyclone tracks and intensity.

Specific comments 

Line 16: "dynamically driven" is meant advected?

Lines 16-17: "that is driven by surface heating and stratiform and convective condensational heating" I suggest you replace with "diabatically produced by latent heat".

Line 21: please modify as "visual similarity" or "phenomenological". The way it is formulated, the phrase suggests that the two kinds of systems are governed by similar dynamics.

Lines 39-40: The two phrases do not articulate. The first one compares Medicanes to othe Mediterranean cyclones. The second starts with "indeed" but compares tropical cyclones to Medicanes.

Line 39: Actually, this issue is still debatable. In fact there are more evidence that cyclones already identified as Medicanes do not differ much in terms of processes from any other extratropical cyclone. Please revise or remove this phrase.

Lines 41-43: This part is not very accurate. Almost no intense cyclone may develop in the Mediterranean without baroclinic forcing. In fact, one of the main community questions about the formation of medicanes is whether an already formed cyclone may be sustained uniquely by diabatic forcing, i.e. whther the so called tropical transition applies to medicanes (please refer to the review paper by Flaounas et al. 2022)

Line 43: I am not sure I understand this phrase. How can warm sea interact with a trough? Is it meant that stability is reduced if colder air in a trough is advected right above warmer air masses at the sea surface? If so, I would argue that convection in this case might be favoured due to forced large scale ascent. Furthermore, in the cited papers, it is discussed the development of medicanes in conjuction with the WISHE mechanism but WISHE supposes no "upper-level troughs" to develop a warm core.

Lines 45-47: Please be more thorough to explain what is referred as "mechanics". In connection with my previous comment, please note that McTaggart-Cowan et al. (2015) also refer to the concept of tropical transition.

Lines 49-51: Large scale and diabatic forcing might indeed have an important role in medicanes development however this is true for any extratropical cyclone. What makes medicanes rather different is the relative contribution of these two forcings in different stages of their lifecycle. You might want to refer to the three paradigms in the coclusions of Miglietta and Rotunno (2019).

Line 51: what is meant by "limited data availability"? Also, why the low frequency of occurrence is related to poor predictability?

Line 55: Not sure I understand this argument. In this paragraph it is argued that large scale forcing and convection are the two main processes that develop cyclones into medicanes. Both might be explicitely or well reproduced even in relatively coearse resolutions (e.g. 10-25 km). Could you please elaborate more on the added value of convection-permitting simulations. 

Line 57: It is not clear to me what is meant by "rather than observationsl aspects"?

Line 65: If I am not mistaken, Di Muzio et al. (2019) suggest that the prediction skill is highly variable and most of the times predictability is poor in lead times of more than 4 days. Indeed, this comes in contrast with the following phrases. 

Line 73-75: Is it meant that in the studies of Di Muzio et al (2019) and Portmann et al (2020) the ensemble members did not use SPPT? If yes, please revise accordingly.

Line 82: This is awkward phrasing. Please specify how are forecasts improved, and what is meant by "specifically with ECMWF ensembles and tropical cyclones". Also please comment on the added value of SPP. Especially concerning the improvement of the forecast, could you please be more specific on the added value of SPP. I think that forecast reliability with SPP is comparable to the one when using SPPT. 

Lines 82-84: Awkward phrasing. Please be more specific that SPP is perturbing physical parameters within the parametrisations, instead of parametrisation outputs (tendencies) as in SPPT.

Line 93: should physical be changed to diabatic?

Line 117: "are run". Familiar language

Line 117: "forecasted period" please change to "duration of simulations"?

Lines 119: Please provide motivation for the chosen initalisation dates.

Lines 122-123: Please provide more information about the coupling, e.g. resolution of the models, frequency of exchanges etc.

Line 130: "the whole parametrisation...". This phrase is unclear please improve technical language. What is "sub-grid orography"? What kind of parametrisation is meant by "large-scale precipitation"? 

Line 138: what is meant by "poorly constrained"?

Lines 139-142: Without being an expert on SPP, I have the impression that here you describe SPPT. Does SPP include "in-space varying noise derived from in-time evolving 2D random number fields"? 

Lines 143-145: Please improve technical language in the description of parametrisations.

Lines 173-177: What was the input dataset for the calculation of cyclone phase space (CPS) diagrams? Please be more specific. It is unclear to which life stage of the cyclone refer to the values in Table 2. 

Line 185: What is meant by "a clear mid-level warm core"? Presumably CPS diagrams provide lower or upper level diagnostics. 

Lines 203-208: In the following, I copy text lines and provide comments to highlight uncertainty in writing style.

"Medicane Trixie formed on the 28th of October, as the consequence of a deep cut-off low which emerged on 26–27 October and moved from northern to southern Italy in the following days, triggering deep convective storms along the Italian west coast." 

Here I guess that it is meant that Trixie was formed due to baroclinic forcing from a cut-off low. It is still not clear what is meant by "emerged". Usually a cut-off low is the remnants of an intruding trough. With "storms" is it meant other cyclonic systems or just convective events? It is also unclear how these storms are related to Trixie.

"This PV anomaly crossed the Adriatic Sea on 28 October at O4:00 UTC (EUMETSAT analysis by Scott Bachmeier, Jochen Kerkmann, and Djordje Gencic) and then quickly moved to Sicily and approached Tunisia and Algeria."

Here probably PV anomaly refers to the cut-off low. It could also however refer to the convective storms as an area of concise diabatiocally-produced PV. Still it is not clear what the described displacement of this PV anomaly tells us about Trixie and how/if it contributes to its development. The citation here seems to be provided in plain language.

"Then between the 26 and the 28 of October, the medicane started to develop in the area of the old PV anomaly. On the 29th of October, it deepened and moved to the east of Malta, then on the 30th of October, it moved eastward towards Greece (Figure 1g)."

What "old anomaly" refers to is unclear. In total, I am not sure what we learn about Trixie in terms of processes and how this is useful for the subsequent analysis.

Line 225: If it is "largely influenced by large-scale processes" then how it is a "reflection of the model’s capability to reproduce multi-scale processes"? 

Line 230: Hart (2003) developed a diagnostic not a "theory". Please revise.  

Lines 228-231: These are motivational phrases that mostly belong to the introduction. Also they do not really align with the objective in line 90. Please be more precise in the objectives.

Line 239: I do not understand what "sensibly" means. Please quantify your errors or just make reference that Trixie is following an opposite direction. 

Line 265: it is stated that INI experiments present largest spread. On the other hand, line 286 suggests that the spreads between the SPP and INI experiments are comparable. Actually, the conclusion of 4.1 is not clear to me. I guess your results suggest that the cyclone track (not the development as mentioned) is sensitive to both initial conditions and parametrized processes. 

Lines 357-358: Here there is an analysis on the usefulness of precipitation forecasts. But in an actual forecast there would be a use of both initial condition perturbations and SPP(T). Please revise your concluding remarks.

Lines 360-385: Please move this part to the methods section.

Review of "Mediterranean Tropical-Like Cyclones forecasts and analysis using the ECMWF Ensemble Forecasting System (IFS) with physical parameterizations perturbations" by Miriam Saraceni et al.

General comments

In this new study, Saraceni et al. discuss ensemble simulations of tropical-like cyclones over the Mediterranean using ECMWF's IFS ensemble system. The ensemble simulations are conducted using perturbed initial conditions from the ensemble data assimilation (INI), the recently developed Stochastically Perturbed Parameterizations representing uncertainty due to model physics with perturbed parameters (SPP), and a reduced set of the SPP focusing on convection (SPP-Conv). The study is comparing simulations for three medicanes, Ianos, Zorbas, and Trixie in 2020, 2018, and 2016, respectively. The three cases considered here represent a reasonable selection of cases with different mechanisms responsible for generating and intensifying the medicanes and showing varying prediction skills of the ECMWF IFS ensemble system. The study evaluates the ensemble simulations of the medicanes based on different observations and metrics, namely track and intensity, precipitation, thermal structure and asymmetry, and tropical-like phase characteristics. Benefits and deficits of the modeling system and the different ensemble sets (INI, SPP, SPP-Conv) are assessed.

Overall, I think the study was carefully and designed and conducted, with plausible results. Except for being a bit long, the manuscripts is mostly well-written and clear. I would like to recommend considering the paper for publication in ACP subject to a few minor comments and suggestions as listed below.

Specific comments

l5: All simulations were conducted with 9 km horizontal resolution. While this seems to work fine here, tropical cyclone simulations are often conducted with higher horizontal resolution, in order to better cover their characteristics and dynamics like the sharp gradients in their structure near the center. It would be interesting if the authors could elaborate on whether the results of their study are significantly affected by the choice of resolution of the forecast model or not?

l210: Is there a reference/DOI for the EUMETSAT report?

Fig. 1: Perhaps showing the reference data (operational analysis) in front of the ensemble tracks would be more clear? In the caption, please fix "operation analysis".

l243-245: Is this something you would actually like to demonstrate, that the spread from INI, SPP, and SPP-Conv is comparable? It might be helpful for the reader to clarify whether this would be a goal of the study?

l301: At this point, I was wondering what is actually defining the size of the initial perturbations. I realized only later, the size of the initial perturbations would be guided by their related uncertainties/perturbations found during the data assimilation? It might be helpful to add 1-2 sentences clarifying this.

l315: Maybe rephrase "... thus a point-by-point verification presents several problems and is guaranteed by using satellite data."? I found it a bit unclear.

l333-334: It would be good to add 1-2 sentences explaining why the secondary maxima are not well captured by the simulations. Later in the text it is suggested that this might be related to the resolution of the simulations, it seems?

Figs. 11 and 12: It might be easier to disentangle the contour lines of sea level pressure, if the coast lines would be plotted in (light) gray color.

l537: Maybe rephrase to "reduced convective activity _near the center_" to stress this, as the Trixie maps show much larger CAPE south of the medicane than the Zorbas maps, which might be confusing.

Technical corrections

l103: please check throughout the paper the proper use of abbreviations (Sect., Fig., ...) according to the Copernicus manuscript preparation guidelines

l105: In Sects. 5 and 6, the results... (?)

l129: SSP -> SPP

l205: O4:00 -> 04:00

l206: please check proper use of time/date formats throughout the paper (e.g., 26_th_ of October, ...)

l270: ...)_,_ a great uncertainty (?)

Fig. 9: fix october -> October in the caption

Fig. 13: fix Potential -> potential in the caption; _at_ altitudes from 25° to 50°_N_; fix date/time format (16_th_, 27_th_,...)

l521 and 524: Emanuel (2005) -> (Emanuel, 2005)

l550: fix 2°_C_

l573: do not use contractions, "isn't" -> "is not"

l580: Indeed, _it_ is ... (?)

l612: do you mean "underlining" or "stressing" instead of "underlying"?

l624: remove comma, "... dataset tends..."

l661 and 666: fix references to the figures

l697-698: remove some of the "mores" in this sentence

l707: _of_ medicanes