The dataset description is excellent, and the dataset will prove to be very useful. The collection is in principle redundant, as all data are already available on ESGF. I do, however, find it very useful to collect this processed dataset in one place. Annual maxima of varying duration should be a very useful intermediate for plenty of potential studies.

Just a few very minor comments: 

Would it be possible in the future to supplement the current data collection with relevant files from other EURO-CORDEX5 simulations with sub-daily precipitation data? Some simulations have published 3-hourly data, which may complete any analysis of temporal resolution below this period. In case you have plans to do this, please mention it in the text. -Otherwise you may ignore this comment.

Please employ the alternative method mentioned in l155-157 instead of the one currently used. It is good practice to average additive quantities before taking ratios, in order to avoid undue weight for points with very small numbers in the denominator. Please revise the text accordingly.

Please mention exactly how you find the 1.5 GWL and 3 GWL periods (probably trivial, but nice for the documentation of methods). Your method for calculating change per degree GWL is a bit unconventional; please discuss how big an effect this has relative to the more frequently used one of taking changes from historical to end-of-century and dividing by whatever global warming happens between those two periods. I do see advantages in you method related to avoidance of extreme periods in very sensitive simulations, but please add some discussion.

Please consider using a more diverse colour palette in figs 1 and 2. It is currently very hard to distinguish levels.

Referee comment to the ESSD "EURO-SUPREME: sub-daily precipitation extremes in the EURO-CORDEX ensemble" by Anouk Dierickx et al. (https://essd.copernicus.org/preprints/essd-2025-30) preprint, and the corresponding WDCC dataset "Subdaily Precipitation Extremes in EURO-CORDEX 0.11°" by Vyver et al. (https://www.wdc-climate.de/ui/entry?acronym=EURCORDEX_prec)

All line numbers hereafter refer to the https://essd.copernicus.org/preprints/essd-2025-30/essd-2025-30.pdf preprint and the https://essd.copernicus.org/preprints/essd-2025-30/essd-2025-30-supplement.pdf supplement file.

(P: preprint PDF; S: supplement PDF; P: page; L: line or lines)

*** Summary ***

First of all thanks to the authors for processing and compiling this useful dataset on annual maximum precipitation amounts for a subset of the EURO-CORDEX RCM ensemble and the comprehensive and extensive accompagnying study, which demonstrates the potential usefulness of the dataset. Despite the fact that the data is basically available through the Earth System Grid Federation data nodes, the data processing and compilation and sharing as FAIR open access research data makes total sense. The manuscript is well written, the dataset is well prepared and fits the scope of the journal. Some open issues as to the construction of the ensemble, the processing of the dataset and the presentation through the data descriptor paper albeit remain.

*** General comments ***

The dataset, or rather the data product, is novel and useful to and usable by the community. Based on information provided the data product could be reproduced, if needed; see my comments below, a little bit more detail would be desirable. The data product is presented with enough context to existing literature; with some sections though, more references to existing CORDEX analysis may be useful. The manuscript supports the dataset well with very useful examples.

The dataset quality is fine, the dataset DOI works well, data meet FAIR principles. The data is findable and accessible (after free user registration) through the long-term WDCC storage and dissemination infrastructure, uses compressed netCDF-4 as an interoperable data format, complete with meta data and provenance information as well as version control. Common standards are met. Some notes and recommendations on dataset processing and refinements are given below. The dataset itself is of high quality. There does not seem to be any inconsistency between the paper manuscript and the dataset. 

The dataset is useful and usable; again some proposition is made below to increase this further by regridding the data to a common grid. The manuscript is properly structured and clearly written. Methods are described in more detail in a useful appendix. Visual material are OK, some minor comments are given below.

*** Specific comments ***

- P/P2/L44ff: Given the validation and usage examples you provide later, perhaps it makes sense to emphasize that the data is especially useful for extreme value statistics.

- P/P3/L70ff: Perhaps mention that there are many more ensemble members available, but you were after sub-daily data, which are provided by much less ensemble members of the CORDEX-CMIP5 simulations. Maybe also address here, that the CORDEX-CMIP5, other than CORDEX-CMIP6, was not based on some balanced matrix design as described in Katragkou et al. (2024, BAMS), who you cite. So I wonder what motivates eventually the ensemble subset the dataset covers? The EURO-SUPREME dataset does not contain evaluation run results, but especially for using the data for benchmarking, e.g., CPRCM simulations or other validation purposes. Especially for the Section 3 evaluation, this evaluation run dataset might also be helpful. Would it make sense and could this still be added? For completeness also the land-sea masks ("sftlf" variable) and the orography ("orog" variable) data should be provided.

- P/P3/L75: You mention the data is on different grids, which the reviewer can confirm from personal experience. However, given you want to provide a dataset most easy to use, would it not be better to provide the data ona  common grid, e.g., a curvilinear EUR-11 grid, based on the original grid specification of EUR-11 (now EUR-12)? You may regrid using nearest neighbour resampling. This seems to me a weakness of the dataset provided.

- P/P3/L76ff: Perhaps structure the Section 2.2 a bit more, separating the variables desciption, data formats, data access, etc. from each other. In L92ff your systematic pre-processing / quality checking steps may be described in more detail; right now it reads as if on accident outliers, grid point storms, etc. were detected. Please indicate whether you applied a systematic check.

- P/P3/L77: Maybe introduce the block maxima method, later on discussed in the Sect.8.

- P/P4/L99: In addition to Fig.1 and Fig.2 perhaps provide an example dataset visualisation, like the 24h maximum annual precipitation for any given year, so the reader gets an idea of the nature of the dataset. Starting off with the return levels, i.e., a drived quantity based on the originally provided data could come after that.

- P/P5/L198ff/Sect.3: Would it also make make sense to cite some of the many studies which have validated and evaluated the EURO-CORDEX CMIP5 RCM ensemble precipitation dataset? In light of the previous Fig.1/2 it makes sense to evaluate the return levels. However, given the seemingly sketchy reference data and the spatial aggregation, would it not make more sense to compare to the very same diagnostic as you provide, annual maximum precipitation per grid element per duration class, and assess typical bias measures? For example for PRUDENCE regions, which you use in Section 5 anyhow? Because given your return level evaluation yields results with larger biases, the reader might question the dataset usefulness. I do think however the dataset is very useful, and it is know that RCMs exhibit biases with precipitation. At the same time, the return level comparison allows you to compare the statistical properties of the historical run vs observations. So without using an evaluation (ERA-Interim driven) simulation based precipitation diagnostic a direct comparison is difficult as well.

- P/P7/L129f: You should make clear you still refer to the return level biases. 

- P/P9/L135ff/Sect.4: Would this section perhaps fit better under the Section 5 with application examples as the goal of the paper is primarily on the dataset you present and not the derived climate change analysis?

- P/P12/L171ff: This is an interesting application example. Only I would not consider here the CORDEX-SUPREME the benchmark dataset, it is used in a benchmarking experiment, but the benchmark dataset is the observational reference data from the synop stations.

-P/P14/L8f: It may not be required by the journal, but as you put a lot of emphasis on the analysis through the return levels, it would be interesting to the reader to also have the code available through a long-term public git repository.

*** Technical corrections ***

Here are some detailed suggestions and remarks, which may help to improve the text:

- P: Out of curiosity, what does "SUPREME" stand for? Does it have a meaning? 

- P/P1/L6: replace: "0.11" (redundant with following and information is missing) with "regional climate model (RCM)"

- P/P1/L6: rephrase: "(coupled to CMIP5)" to "(downscaling CMIP5 GCMs)"

- P/P1/L6: rephrase: "precipitation depths" to "precipitation amounts"

- P/P1/L7: rephrase: "More specifically ... by a" to "Specifically, data are based on a 35-member RCM ensemble ..."

- P/P1/L8: add: "EURO-CORDEX EUR-11 (0.11°) domain"

- P/P1/L20: missing: reference to last sentence

- P/P2/L22: remove: "urban water management", just "water management"

- P/P2/L32: rephrase: call the models CPRCMs, as you refer to RCMs at km-scale resolution

- P/P2/L33f: rephrase: the computational resources needed are bigger, this is presumably what you mean; also cite Schaer et al. (2020, https://doi.org/10.1175/BAMS-D-18-0167.1), with an overview of computational demans of CPRCM simulations

- P/P2/L34f: add: also the length of the simulations (so far) is rather short, aside from ensemble and domain size

- P/P2/L35ff: there are also other ways to demonstrate the CPRCM added value, I think this is a misleading motivation; the fact is that the CPRCM may simulate extreme precipitation more accurately, albeit they may be of limited use due to the constraints you mention

- P/P2/L39f: mention the CPRCM benchmark objective later on, in line 44f

- P/P2/L44ff: align the listing of purposes with the overview in the abstract, to make the motivation for the dataset very clear 

- P/P3/L63f: explain the CORDEX acronym, also it is not a "framework" but a WCRP "project", cite Gutowski et al. (2016, https://doi.org/10.5194/gmd-9-4087-2016) 

- P/P3/L84f: It seems you follow the data reference syntax and filename nomenclature as used by CORDEX, which I think is very good. Perhaps you want to indicate this.

- P/P6,7/Fig1,2: Please use a more differentiating color-scale. Despite adjusting the colorbar range between 1h and 24h annual precipitation maxima return levels, the colormap should be the same for the same kind of variable.

- P/P6,7/Fig1,2: Please indicate that you do not show the entire EUR-11 CORDEX domain for readers not so familiar with the CORDEX RCM ensemble.

- P/P10/Fig.5 caption: return level of what?

- P: partly there is a mix of British and American English spelling. Please double-check this throughout.

- P/P12/L190: remove: "over the (southern)" -> "over (southern)"