This is a review for ESSD, von Matt et al., 2025: HYD-RESPONSES: daily hydro-meteorological catchment-level time series to analyse HYDrological drought dynamics in RESPONSE to (cumulative) water deficits in Swiss catchments 

The described dataset appears to be of potential use, albeit limited for Switzerland-focused usage. I think the manuscript requires major revisions in terms of structure, motivation for certain choices or clearer descriptions as summarized in the general and the specific comments below.

General/major comments:

Item 1. The abstract and the introduction, while motivating and describing the dataset, do not lay out the logic of the manuscript. For example, the use cases in sec 6 come a bit as a surprise to the reader. In general, the manuscript would benefit from an additional verification effort, showing that the aggregation procedures work as they are intended. For example, 4.2.1: FOEN uses M7(Q), too, so there would be scope for a comparison.

Item 2. Description of catchments and respective coverage by the datasets (section 2-3, figure 1, 2, 3): Here, I think that a bit more care could be given to a comprehensible presentation:

Section 2 would benefit from clarifications i) that the catchments lie partially outside of CH, ii) whether they are fully disjunct (i.e. there is or there is no hierarchy in the catchments), and iii) that any altitude discussed here is basin average.

Then, in section 3.1, there is an explanation why certain stations are not used for the HYD-RESPONSES dataset; some items are self-explanatory, but others are not. Suggest to explain.

Section 3.2 starts a bit sloppy, as meteorological variables are indeed provided from multiple sources; suggest to re-phrase or re-organise this description. Which data does MeteoSwiss use for RhiresD in areas outside CH?

The very last sentence of section 3.2 explains that ERA5-Land data are aggregated to daily values, but does not provide details. Is it done in a way that makes respective data comparable to the other sources (TabsD, TminD, TmaxD, RhiresD)?

Fig 2 is somewhat unmotivated and only referenced in section 6.2.

Section 3.3 does not provide information on coverage outside of CH. This should be clarified. If restricted to CH, is this a problem for catchments with significant portions in neighbouring countries? L344 onwards and the discussion and conclusion sections in principle suggest so.

Item 3. The information around homogeneity/breakpoints is taken from elsewhere. Still, I think it is important to clearly explain their concepts (L189), because these are not self-explanatory terms. Their usage is also not quite clear. L213: what makes a breakpoint “significant”? Is it an ad hoc definition?

Item 4. Guidance on how to use the many different quantities in the dataset; the authors have compiled many different indices, deficit manifestions, etc, and their genesis and differences among related ones are explained well enough. However, what is missing is guidance for users which to use for which purpose. Some examples:

L245: Can the authors give context on whether the deficit remaining uncompensated over several years is realistic? They provide an extra quantity of annually reset deficits which indicates that it might not be. An expert user will probably know what to make of it, but less experienced users might be a bit lost.

Section 4.4: Which application would require the usage of SPI/SPEI/SMRI, respectively?

Section 4.6: fixed and variable threshold definition.

Section 5.2, L335: two BSI variants.

Item 5. The manuscript remains sometimes short when in comes to motivating certain choices or methods. For example:

L295: Why are more percentiles provided only for M7Q?

L309: purpose of “minor pooling”

I suggest to screen the manuscript for these instances and include respective explanations.

Item 6. Figures and references to figures. I think these must be clearly improved:

Figure 2 was mentioned in my item 2 already.

I find no reference to Figure 6 (although it might help explain the “phases” question that I put under “specific comments”).

Figure 7 should probably specify subfigures as (a) and (b).

Figure 9: Shouldn’t the caption specify *streamflow* drought for the pink and green shadings? I think what is missing here is the time series that was used for defining the drought periods. It seems related to panel j but not identical according to the text from L418 onwards. The y-axes should be labeled properly (T2m: [K], SWE: [mm], ...).

Figure 10: not referenced in the main text; the text supposed to discuss the figure mentions “pluvial inferieur”, which is not shown in Figure 10, several times around L470. What does the “n” stand for, catchments?

Specific comments:

L23/L89: please introduce CAMELS-CH to the uninformed user not only in section 8 but already at first mention/in the introduction.

L39/41: do the authors really mean “anthropogenic” in the sense of “caused by humans”? Or is it more about these events impacting the human “sphere” differently? Might also be relevant elsewhere (L375, L390).

L79: “as a result of non-transformation”, is this simply replicating the statement of “non-standardisation”? If so, recommend to delete.

Caption of Table 1: refer to glossary table in the appendix for variable names?

L230-233: The causality here is not entirely clear. I am guessing: The reset at sept 1st is a feature of SPASS SWE in order to avoid the “snow tower” feature. This feature leads to large snowmelt values in delta SWE. These large values are mitigated by the described interpolation. If so (or even if not), the chain could be spelled out more clearly.

L274: 50 implausible/missing values: Is this referring to the time series, or the distribution (i.e., before or after binning)?

L282: This means that values are capped at +-3 STD, which could be spelled out here. Can the authors briefly repeat the reasoning for this by Stagge et al. (2015) in this context?

L284: Please explain “standard time series”.

L299: “Q347” is only introduced in section 5.2; an explanation is necessary at the first instance, I think.

L305: What constitutes a “phase” as used throughout 4.7? Probably related to Figure 6 (unreferenced, see above). Unless this is clarified, it remains unclear what an “event” is.

Sub-Headings in section 5: “extraction of catchment descriptors (CD)”, “derivation of other CD”; these seem not very precise. What exactly is the difference between CD in 5.1 and 5.2? Can the header name or the distinction be specified?

L331: Is competing areal extent of several categories in a catchment the only possible constraint to representativeness? I could imagine that for example the spatial distribution in terms of upstream/downstream/margins could also be a factor.

L365: “partly on both monthly and yearly time scales”, this should be specified and motivated.

L383: The authors have provided a short interpretation of Fig 8; I think they should also do this for Figure A1 (which could also be moved from the appendix to the main part?).

L413: “lacking snowmelt”: wouldn’t there be more concrete evidence for this somewhere in the HYD-RESPONSES dataset?

L424: “Larger CWDs during...” is this a general (climatological) statement? The term “seasonal climatology” and the “(not shown)” addition might indicate this, but I recommend to stress this more clearly.

Starting sentence of Section 7: something is not quite right here with the references/model names.

Technical/editorial comments:

L9 onwards: inconsistent in naming all MeteoSwiss/SLF parameters, but only ERA5-Land as a whole.

L12, “information on precipitation, evaporation-driven and streamflow deficits”, can the authors please re-phrase; something seems not quite right here.

L93/94: “n=18/94” I think it is misleading as the n refers to streamflow regime types first and then to individual catchments; why not simply include the numbers in prose?

L203, “variables” should go after “flux”?

L264: suggest to unify reference to R packages across the manuscript. E.g., elsewhere it is “cwd R-package”, here it is “SCI-package”.

L284: There is a left-over bracket, probably to be closed after “3.2” in L285.

L293: I suggest to spell out the z score more clearly instead of the “(value-mu)/sigma terminology.

L308: “event definition”, drop “definition”?

L362: MAM = march april may as elsewhere? In general, it is not clear what the acronym(s) is/are supposed to say.

L383 An other > Another

L424: “actual” in the sense of “non-anomaly” or in the sense of “non-potential”? Same for L431.

L563 and elsewhere: data set or dataset?

L563: “alsop” typo

This manuscript presents a comprehensive hydrometeorological dataset that is developed from a range of other, existing, datasets from various sources. This dataset is developed specifically to support assessment of drought and low flow conditions across catchments in Switzerland, and includes time series of a wide range of essential climate variables as well as derived drought indicators (e.g. SPI, SPEI, etc). As this dataset, or rather what I would consider a data collection (See scope of ESSD, https://www.earth-system-science-data.net/about/aims_and_scope.html) can support detailed assessment of drought and the drivers of drought (see also use cases presented in the paper), I would think this in line with the scope of the journal.

Overall, the paper is well structured (with comments, see below) and the datasets that have been developed are outlined clearly, including the original data sources and general comments on data quality.

One of the main concerns I have is that the dataset combines various underlying sources, in particular observational datasets and re-analysis datasets and datasets from models. Some attention is given at the start of the discussion on where care should be taken, but this is only discussed for selected variables, particularly related to snow (e.g. SWE), but much less discussed for other variables, for example E and PET (see also comments below). In this sense the discussion is somewhat poorly developed. The first part addresses some of the limitations, but a broader reflection on the quality of the dataset, including derived variables, would add to the depth of assessment of what is presented. Perhaps some indicative confidence level on the different datasets and combinations would be useful. I think this should also be added to the metadata provided with the dataset in Zenodo. 

The dataset has been developed specifically for Switzerland, and for application in the Swiss context. It would be interesting to comment on how applicable the methods used/presented here would be applicable in other contexts. Some reflection on how applicable this could be in other contexts/countries, what the requirement are of underlying datasets would be etc. For example, use is made of ERA5-Land, which is of course available globally, but on the other hand the availability of observational data in CH is excellent. In other settings where there is less observational data, more use would then perhaps be made of re-analysis data. However, would this still make “sense”?

The use cases that are presented in the paper are interesting and indeed demonstrate the utility of the dataset. On the other hand, they may not be the core of the paper, and the paper is very long One could consider including these in the supplementary material. What could then be interesting is to provide some general reflection on the application of these use cases, and how the HYD-RESPONSES dataset and methods have enabled these analyses, and why that was difficult or not possible prior to the development of this dataset.

Detailed comments:

Line 98-99: Catchments are described as being at a certain elevation. This seems to be the mean elevation (this is mentioned later). May be useful for comprehension to name that here.

Line 116: It may be useful to explain a bit better what “meaningful” means. I understand this is obtained through personal communication, but it is somewhat vague.

Line 118: Stations where Q is measure at a water level station. I find this somewhat confusing, as I would presume (and my experience working with FOEN would confirm) that most river stations measure water levels and derive the discharge through a rating curve. Perhaps something else is meant. Please clarify.

Line 118: Mention is made of NADUF stations. I am not familiar with what these are. Please provide some explanation.

Line 121: list of stations included

Line 124: May be useful to mention what is meant by assembled – I assume that this is compiling the catchment average for these variables.

Line 126: I appreciate that the authors use the original names/ids of data depending on the source (e.g. RhiresD, tp). Table 1A provides some explanation which is useful. Perhaps it would be useful to provide in that table something like the WMO standard naming conventions. Would also be clear if the authors mention this strategy in the text, as it may otherwise become somewhat confusing.

Line 139: Mentions is made of the quantile mapping approach being used. It is not so clear to me what the reference is for this bias correction through quantile mapping. Please clarify.

Line 167: I think the word “The” at the start of the sentence needs to be dropped as it is otherwise not clear which digital soil suitability maps are intended as these have not been introduced.

Line 192: Does mean height here imply the mean elevation? Please be consistent.

Line 205: Check the sentence starting with “For accumulation… “. It is somewhat confusing and may need to be rephrased or elaborated to be clear.

Line 215-216: Limnographs are often mentioned. The word is correct but to my mind not in such common use. It is somewhat a Germanism to my mind. Perhaps use Water level sensor or something similar

Line 233: Mention is made of interpolation between the day before the end of September and the day after, when SWE is det to zero. Surely the amount of water that melts is the same whether this happens over one or two days – and still unrealistically high, given that when resetting I assume all snow is considered as being melted. Perhaps I am misunderstanding the concept.

Line 239: Also related to the general comment. Here E and PET are derived from observed P (interpolated) and E and PET calculated in ERA5-Land. I am not clear how biases are dealt with, especially in E. If in ERA5-Land the precipitation is strongly biased, then surely E will (climatologically) tend to be too low in catchments that are water limited.

Line 246: Here it seems to be suggested that PCWD is calculated using ERA5-Variables. Is that both for P and PET? Please clarify.

Line 255: The word period is used here to indicate the accumulation window for SPI and SPEI. In other sections the word period is used to denote a period in time (e.g. 10 years). Please use words consistently with a defined meaning, as it is otherwise somewhat confusing.

Line 273: mention is made of fewer than five DOYs flagged. I am not sure how these are flagged! Perhaps I missed it.

Line 282: I was curious in the derivation of the distributional parameters of the distributions applied in SPI, SPEI and SMRI, if the same period of data was used to derive the parameters of the distribution, or if for each case the whole available time series was used. That could make comparison more difficult.

Line 292: Mai à May

Line 336: Length – I guess of the main drainage path -please clarify.

Line 421: What does HRSg mean – often used but not clarified.

Line 441: Why the 15^th percentile – if this is just to illustrate the please state is an arbitrary threshold.

Line 490: It may be good to note that ERA5-Reanalysis and ERA5-Land are (to the best of my knowledge) not independent, with ERA5-Land derived from the former by downscaling using features such as elevation etc.

Line 499: The discussion that ends here is relevant, as in the dataset several indicators are developed that combine data from different sources - such as the cumulative water deficit, and the snowmelt corrected precipitation datasets. Given that these combine observational and reanalysis data, this may result in different levels of reliability of the derived datasets. I would be curious as to how is this flagged in these derived datasets. In other words, is some flag of degree of confidence set in the meta-data?

Table 3: I am not sure if Zenodo can be considered a provider – is this not more a repository?

Line 563: also

Figure 4: “complete” is mentioned – I guess this is the same as full. Nice to be consistent.

Figure 5: The label NAs/Implausible should be described as to what it means (one can guess of course – but best to be clear).