Klotz et al. have posted a rich set of meteorological, discharge and sediment yield data for 6 French alpine catchments containing badlands from the Draix-Bléone Observatory. Given the difficulty of conducting such monitoring in a mountainous environment, the temporal span of the proposed chronicles makes this dataset quite unique and therefore very valuable. The authors have extensively illustrated, via previously published studies, the potential of the proposed dataset to contribute to a variety of issues related to runoff and erosion in this specific mountainous environment.  

The article is well structured and clear. The data are readily accessible and usable in their current format and size, either by using the 5-year snapshot that is available for review or by using the BDOH web interface for the full dataset (requires registration, but has powerful features that make exploration of the dataset quite user-friendly). 

I very much appreciated the significant effort made by the authors to explain the complex procedures for collecting, processing and qualifying hydrosedimentary data and I think this manuscript is acceptable for publication with minor revision.

My main suggestions are intended to provide additional information to help those who would like to use these data sets or replicate similar collection and processing procedures. In addition to minor editorial suggestions, they focus on the following two points: (i) the description of the data processing and qualification procedures, which is sometimes too brief or a bit confusing, and (ii) how the quality codes assigned to the different data can or should be used.

My suggestions for improving the paper are detailed below in three comment sections (general/specific/editorial)

General Comments:

All the data proposed in this document have been qualified with a quality code -presented in Table 2- in the form of 4 levels (good - intermediate - uncertain - poor) if we exclude the codes for missing data and data without quality code. The effort made by the data producers to generate such detailed quality grades and to attempt to formalize the attribution rules is noteworthy. However, nothing is said about how to take these different levels into account when analysing these data. It would be greatly appreciated if the authors could provide guidelines to future users of the data on how to take these different levels into account. For example, do they advise to associate different levels of uncertainty to them? If yes, could the authors provide an idea of uncertainties associated to each level (at least as a range, i.e. [Min; Max]) ? Since uncertainties may depend on the configuration of each data monitoring system, it would probably be appropriate to include this information in the header of each data file. In addition, I would suggest including in paragraph 4 a small discussion of how existing studies that have used these data have accounted for these different levels of quality...

Specific Comments:

L41-42: I would not use "are only available in..." as other datasets combining records of suspended and bedload may exist (e.g., DOI: 10.2478/johh-2019-0003).     

L85-86. The sentence "Mean annual rainfall for the observatory... and mean annual temperature..." would require you to provide the method you used to estimate a average value from multiple measurement locations. Instead, I suggest that you provide an average value at a single observatory location inside the observatory and indicate the time period used to estimate the average.

L96-97, Table 1 and Legend of Figure 1: At the end of the introduction the authors explain that the data from the Galabre catchment area, which belongs to the Draix-Bléone observatory, will not be presented in this datapaper because they have already been published. Therefore, I suggest not to mention it again here, nor to present it in Table 1, nor to mention that it is not represented in Figure 1.

Table 1: please consider i) dropping the line for Galabre; ii) adding a column indicating the % of badland area for each catchment; iii) providing one sentence to describe how the average catchment slope was evaluated (from which data ? which spatial resolution if derived from DEM...); iv) specify in Line 91 that the vegetation cover was stable during the whole period of monitoring (if not, provide a range of value instead of a unique value)

Table 3: add the resolution for the Archail raingauge

L129-131 & 140: please provide more details on how the raingauge data were corrected because I didn't see in the data I consulted any adjustments of the bucket volume.

Figure 5: please consider using a ratio 1:2 between y1 and y2 (i.e. P = 2*T) as usually done to present this king of graph in the Mediterranean context (here you used P = 4T)

Table 5: add the acquisition frequency for the Bouinenc station

L229 & 233: Please specify the difference between noise suppression (L233) and oscillation suppression (L229) and/or consider merging the two steps.

L234-235: Please provide more detail on how the time series of recorded water levels are corrected for inconsistencies between some occasional scale readings (or manual bucket measurements) and the continuously recorded data. 

Section 3.3.1: Can you provide a table summarizing the data set for suspended load, as was done for all other data categories?  

L285-286: How many samples are enough to establish an event-specific relationship?

L285-295: the combination of procedures (and the priorities between each step) need to be clarified (see comments below). 

 -- Step 2: Consider merging step2 into step4. 

 -- Step 3: is is not clear if all sediment concentrations derived from the voltage/concentration (or the turbidity/concentration) relation are included in the reconstructed concentration time serie. If only part of them are included, please detail the criteria used. What happens if no sample is available ?  

 -- Step 4: Are concentrations derived from discharge/concentration relations included only when turbidimetric measurements are not available (i.e., shallow water depth, or concentration below 10 g/L, or device failure) ? are there other cases where they are included in the final reconstructed concentration dataset ? 

 -- Step 5: What is the decision criterion for moving from the Step 4 to the Step 5 ? 

L300: consider changing "volume" by "mass" in this sentence, as multiplying the discharge (l/s) and the concentraztion (g/l) provides a flux (g/s), and then integrating gives a mass.

L301: consider switching "volume" and "mass" in this sentence.

L304: "When enough samples are available"  is vague... please try to provide an idea on how many samples are required ?

L313-314 & legend of Figure 7: please specify whether instantaneous or event-scale average concentrations are concerned 

L352-354: please explicit the rule(s) for assigning a volume of sediment deposits to a series of floods.

L364: Would it be possible to mention the long-term bedload's contribution to total exports (in addition of the value for 2013) ?

Section 4: as already said, the author have extensively illustrated, via previously published studies, the potential of the proposed dataset to contribute to a variety of issues related to runoff and erosion in this specific mountainous environment. Surprisingly, the authors did not provide perspectives on potential future uses of their datasets. Please could you discuss a bit more potenital future works.  I also suggest to ad one or two sentences on the future of data collection.

Section 5: I haven't found a way to access GIS files, such as catchment boundaries or device locations, either in BDOH or in the 5 year snapshot provided for the review process. If these data already exist somewhere in access, I suggest mentioning the links to access them, otherwise I suggest adding them to the datasets proposed in this datapaper. 

Editorial Comments:

L64: Drop “^” from “Legoût”

L333: Change "Fig.9" into "Fig.8" 

References: please carefully check the list of references as I identified the following issues: i) (Le Bouteiller et al., 2019) and (Gras et al., 2007) to be added ; ii) alphabetical sorting problem for Klotz at al.; iii) reference (Olivier et al., 1995)  is presented in two parts (L534 & 553-554)

This is a nice paper giving a comprehensive overview on the infrastructure, data and measurements of the Draix-Bléone observatory. I have two main comments on potential additions, which the authors may want to think about.

First, I think it would be helpful to represent some of the information graphically, for example, some of the more complicated workflows (suspended load), and the history of infrastructure and instrumentation. The latter could be done as a Gantt chart, showing timelines of deployment for different stations and their sensors, and highlighting changes in instruments and procedures.

Second, the topic of errors and uncertainties is hardly treated at the moment. The authors explain the quality assessment into categories; yet this is something different to giving quantitative uncertainties. How do the different classes convert to errors that could be added to plots? Have some of the uncertainties been systematically assessed, for example through repeat measurements or by comparing to high-quality benchmarks? For example, the suspended load measurements using the turbidity probes can be compared to the benchmark data from manual sampling to provide error estimates.

Some more comments can be found below.

21 here and elsewhere: dot missing after et al. (et al. is an abbreviation for the latin ‘et alia’, which means ‘and others’)

33 an ‘and’ missing before the final item of the list.

39 Could refer to Turowski et al. Sedimentology 2010 here.

51 I agree with the statement here. Yet, the cited literature does not give justice to the important scientific results coming out of badland studies. There are key papers for example of Alan Howard, or from Chinese and Taiwanese research sites that might be added.

61 …consists of…

178 …while others…

186 maybe show the different sensors graphically, in a timeline or as a Gantt chart?

191 What is the threshold?

197 …during each field visit.

198 Again, it might be helpful to have a graphical representation of the activities / data available for the various gauging stations.

200 please add references for the Parshall standard relation and the depth-discharge relation of the trapezoidal flume or provide more information.

205 How is discharge obtained for the Roubine?

215 …based on manual measurements of flow velocity using the salt dilution method.

215 please give the date of the flood

217 change to “As a result, no discharge data is available after this date” or similar.

258 Can you give details on the thresholds?

262 turbidity

271 suction pipe

273 a graphical representation of the work flow may be helpful

286 …is used.

314 An implication from the statement here is that, at high concentrations, viscosity and density of the fluid significantly change. Is this considered when converting stage to discharge? If yes, how? If no, what is the potential error?

330 maybe add a brief discussion on how the traps affect discharge measurements and the relationship between discharge and bedload transport (they probably act as a buffer for high flood peaks).

342 How is the total volume measured?

344 Can you give typical values?

“A high-frequency, long-term data set of hydrology and sediment yield: The alpine badland catchments of Draix-Bléone Observatory” proposed by Klotz et al. is a paper that presents a quasi-unique dataset consisting of long-term measurements of rainfall, meteorology, runoff, suspended- and bed-load in the Draix-Bléone Observatory. The materials and methods used to realize the monitoring program were described in detail, also making available a 5-year dataset via the BDOH web interface. As stated by the authors, long-term datasets about climate, hydrological, and sediment dynamics are rare, especially in the alpine range. Therefore, efforts such as this to present in detail the processes of data acquisition and elaboration are very valuable, especially if the data are then made available to the scientific community, as in this case. That said, I have some comments. First, it is unclear (to me) what message the authors want to transmit with this article. I am not sure that the presentation of a dataset and the relative acquisition procedures are sufficient for publication. Maybe, an interesting aspect would be how the uncertainties and the quality code presented may affect the results obtained. This point could be discussed in section 4, which sounds highly site-specific.

My specific observations, comments, and suggestions are:

Table 2: I found the code a bit misleading. Maybe, it would be better to assign 0 to missing data, 1 to no quality, 2 to poor quality…..5 to good quality. So, it might be easier to follow?

L129-130: Was this correction made by spreading the difference over the entire period or by considering only the rainfall events? Please, specify.

Table 4: Mean wind speed and Mean wind direction are mean values as Temperature, Humidity, and Radiation? If yes, please add such information in the table as made for the other variables.

L186-189: Could the use of different devices have caused a certain uncertainty in the data gathered? If yes, please discuss this point.

L194: So, you visited the study areas every 2-4 days?! Impressive! Maybe, it could be interesting to spend some words about the pros and cons of this high sampling resolution.

L199-200: “Note that…measurements” could be removed. Unnecessary.

L227: Nilomètre should be Nilometer? Please, check over the entire manuscript.

L313-314: Please, rephrase. Unclear.

L435: Was this emphasis given in the dataset or in the article? In both cases, the sentence sounds quite cryptic. Please, rephrase.