In their article, [Katharina Hartmuth et al.] present a novel method to assess the forecast skill of the ECMWF ensemble in predicting extreme weather events associated with Mediterranean cyclones. The first part of their study explains the methodology and illustrates it with three case studies of impactful Mediterranean cyclones. The forecast skill is evaluated based on the ability to predict the occurrence of extreme precipitation and extreme surface winds (both defined by exceedance of the local 99th percentile).

The paper is overall clear and well-structured. Particular attention has been given to the description of figures, which makes them especially pleasant to read. You will find attached some suggestions, mostly minor, that may help improve the paper.

The paper presents a new object-based framework to quantify ensemble forecast skill of extreme surface weather events linked to Mediterranean cyclones, demonstrated through three case studies (Denise, Jan, Daniel). The study finds good predictability at short lead times but high case-to-case variability at longer ranges. This contribution is clearly within WCD’s scope and addresses questions of societal relevance.

General assessment

The manuscript is strong and well-prepared. It provides a detailed and systematic methodology, and the case studies are relevant and timely. The figures are carefully prepared and generally of high quality. The paper also addresses societally important questions, given the devastating impacts of recent Mediterranean storms, which underlines its relevance for early warning and risk preparedness. One central issue, however, is that the role of this Part 1 relative to the forthcoming Part 2 is not entirely clear; it would be important to ensure that Part 1 stands on its own with distinct take-home messages.

Where the other reviewer emphasized technical aspects of resolution and thresholds, which I completely agree and I think would greatly improve the manuscript, my main suggestions will relate to:

• Framing and motivation (why this method, and how it complements existing approaches),
• Balance between method and science (it currently reads primarily as a methodological paper with case studies as illustrations and could be strengthened by drawing out more physical insight),
• Integration of results across cases (a more comparative discussion would make the findings stronger),
• Positioning relative to Part 2 (clarify what this paper stands alone for, and what will only be delivered later).

Overall, the paper is a solid methodological contribution and of clear relevance to WCD but I recommend major revisions to sharpen the message. To maximize impact the manuscript should: (i) better highlight the scientific insight gained from the case studies, (ii) improve the comparative and interpretive discussion, and (iii) clarify the role of this Part 1 relative to Part 2.

Specific comments

• Motivation and novelty: The paper could better explain how the proposed framework advances beyond existing object-based verification methods, and why the cyclone-centered perspective is particularly valuable compared to more classical metrics.
• Balance between method and science: The manuscript leans heavily on methodological description, with the case studies serving mainly as illustrations. The paper would be strengthened by drawing out more physical insight from these examples. The scientific insights could be expanded by linking forecast skill more explicitly to storm characteristics (e.g., intensity, lifecycle, baroclinic vs. diabatic influences).
• Comparative discussion of cases: Each storm is described in detail, but the synthesis across cases is limited. A stronger side-by-side comparison would highlight important contrasts (e.g., Storm Jan’s poor predictability vs. Denise’s higher ensemble detection). A summary table or concise figure could help.
• Scope of Part 1 vs. Part 2: The division between what this paper achieves and what will follow in Part 2 should be clearer, to ensure Part 1 stands on its own as more than a methodological note.
• Assumptions and parameters: Some methodological choices need fuller justification (e.g., attribution radius of 400 km, seasonal percentiles, coarse temporal/spatial resolution). Even if not modified, sensitivity or limitations should be discussed more explicitly. This point has already been noted by another reviewer, and I also consider it essential to address explicitly.
• Figures and presentation: Figures are of high quality but often too dense (e.g., Figs. 9-13). Simplification, clearer legends, or moving some panels to supplementary material would improve readability. In addition, some legends (e.g., “number of overlapping objects”) could be clarified, and color/hatching choices made more distinct.
• Terminology and clarity: As the other reviewer noted, the use of “probability” for “fraction of ensemble members” may mislead readers; either define this clearly or use alternative wording such as “ensemble fraction.” Likewise, conditional/unconditional terminology could be simplified.
• Connection to impacts: The discussion could more strongly link the findings to societal relevance, such as implications for early warning systems and the realistic lead times at which reliable forecasts of Mediterranean extremes can be expected.

With these revisions, the paper will not only present a valuable methodology but also provide clearer scientific lessons from the case studies, making it an excellent contribution to WCD. I therefore recommend major revisions, with the expectation that a revised version could be suitable for publication.

Evaluation according to WCD criteria

In terms of scientific significance, I consider the contribution to be good. The object-based ensemble method represents a useful innovation and is clearly relevant to the study of Mediterranean cyclone predictability. However, as this paper mainly introduces the methodology and demonstrates it through three case studies, its broader scientific impact is currently modest but could be increased with greater emphasis on the insights gained from the case studies.

Regarding scientific quality, the work is carefully carried out and methodologically sound. The authors describe their approach in detail and apply it systematically. Nevertheless, several assumptions—such as the choice of attribution radius, the use of seasonal percentiles, and the reliance on relatively coarse temporal and spatial resolution—require more justification. These issues do not invalidate the results but leave some uncertainty about their generalizability.

The presentation quality of the manuscript is overall good. The paper is well structured, written in clear English, and supported by a rich set of figures. At the same time, the manuscript would benefit from some streamlining: certain sections are verbose, some figures are overly complex, and the main messages could be highlighted more strongly. Simplifying figure layouts and focusing the discussion on key findings would enhance readability.

This paper evaluates a new method for detecting and characterising high-impact weather systems using objective diagnostics. The approach is applied to reanalysis and ensemble datasets, with the aim of improving the understanding and detection of extreme events related to Mediterranean cyclones. The work is very relevant, as it supports the operational need for new and user-friendly tools to identify hazardous weather in a timely way.

Major comments

1. The analysis currently relies on quite low spatial and temporal resolution. While this is a reasonable starting point, it may not adequately capture small and rapidly evolving systems such as medicanes, particularly in the Mediterranean. If feasible, the authors could consider exploiting higher spatial/temporal resolution data to better resolve extremes. It may be a lot of work and not feasible for this paper, but one good option would be to use the CERRA reanalysis instead of ERA5, as it has much higher temporal resolution and closer to the resolution of the ENS IFS. Also, different IFS model versions are used, and this may affect the final results. You should check if, at least, the ENS horizontal resolution didn´t change (from 18 km to 9 km) between your case studies, as it may give different results. In that case, it may be better to use re-forecasts (hindcasts) to always have a similar model configuration.

2. The framing of the study objectives is somewhat unclear. As a reader, it is difficult to distinguish which aims apply to this specific paper and which are intended for the companion study.