General comments

Shen et al. presented a very comprehensive reservoir dataset for China, Res-CN. The new dataset includes water area, water level, storage variations, and corresponding catchment characteristics that derived from multiple sources (i.e., satellite, reanalysis, and observation, etc). The authors also validated Res-CN with in-situ observations at selected reservoirs to demonstrate the accuracy of the dataset. It provides valuable information for hydrological modelers to investigate water managements and the impacts on (eco)hydrological cycle. Although I think Res-CN represents a significant contribution to improve our understanding of reservoir dynamics and water management in hydrological modeling, some parts were not clearly presented /explained in the main text because Res-CN contains extensive information. Additionally, some figures were missing in the supplementary materials. So, I recommend revision before publication. Please find my comments in the following. 

Major Comments

The authors mentioned in the introduction Line 99 that in-situ data of 138 reservoirs were used to validate the Res-CN, but the validations at a few reservoirs are shown in the result section, with summary in the main text. It is necessary to show the validations explicitly for all the 138 reservoirs that demonstrate the accuracy of Res-CN.

There are a lot of information provided by Res-CN, but some are not clearly explained. It mentioned in the introduction that 3,254 reservoirs were presented in this dataset, but in Table 2, the topography are available for 3,689 reservoirs. Table S10 summarized 18 attributes of topography, but it listed 19 attributes in Table 2. I can find 23 attributes in Table S13 for land cover. In addition, please clarify how the 173 is estimated from Table S14-S15 for the Soil & Geology. And how the 288 attributes are identified from Table S16 for Anthropogenic activity? Please clarify Table 2 and clearly link to the supplementary materials.

It will be useful to add more details for the water areas at line 88. For example, the range of 0.004-1373.77 [km²] is very wide. A histogram of the water areas will be useful for the end-users because researchers have different focuses. For example, a watershed hydrologist may be interested in relatively small reservoirs, but an Earth system modeler may only need large reservoirs. Also, it will be helpful to list some major reservoirs based on the water areas (e.g., first ten or twenty?). As argued by the author, the largest reservoir area is 1,373.77 [km²], but this number is not consistent with my source.

Minor Comments

Line 42, Please capitalize Earth.

Line 76-78: “In addition, there is no systematic assessment of whether reservoir water levels or water areas from previous studies and databases agree with one another, as shown in this study by many reservoirs whose in situ measurements are available.”. I don’t understand this statement, are you trying to argue your results suggests the results from previous studies are biased when compared to in-situ measurements?

Line 80: “there are approximately 30 Chinese” Do you mean there are approximately 30 reservoirs from China?

Line 106: Please provide the source or reference for the number of 98,000.

Line 109: Are the 3,254 reservoirs from GeoDAR?

Line 135: What is your criteria for reservoirs with large variations.

Line 165: I am confused about this statement. Is this “768 reservoirs” from this study? If so, please clarify it. If not, please cite reference to support it.

Line 243-244: Add reference or results to show the validation of delineation for the 1,398 catchments.

Line 305-306: The authors explain the large errors occurs in 55 catchments are because the size of the catchments is small. But Figure 3d and f show the large errors also occur in large reservoirs. The spatial map is not very clear to show where the errors from. Consider plotting the comparison with the reference dataset using the scatter plot.

Line 326-328: I am not sure if RMSE is a good metric to indicate error for the water level. The magnitude of water level varies with reservoir size. So, RMSE = 0.3m is considered as small error for a large reservoir, but it can be significant for a small reservoir. Since the time series of water levels are compared, some evaluation metric like NSE or KGE can provide more information about the evaluation.

Line 330: There is no Fig. S7 in supplementary materials.

Line 335: There is no Fig. S8 in supplementary materials.

Line 372-373: Fig.6a and b plot the water areas comparisons from all the reservoirs and months, then what does the median CC mean?  Did you also estimate the CC for each reservoir? Please clarify what does the median CC mean. Also, it is critical to show the evaluation at site level to demonstrate the accuracy of Res-CN.

Line 384: NRMSE, CC and RMSE have median values of 21%, 0.53, and 0.03 km³, respectively.

Line 391: Please specify the number of available reservoirs when using the water areas and water levels to derive the storage variations. Are they the same 119 reservoirs that used the DEM’s area-storage model?

Line 412: Please clarify this sentence:” Long-term mean meteorological variables calculated the evaporation rates are available in Fig.S9.”.

Line 424: Consider changing the colormap for Figure 8b, because the map doesn’t show any variation of water areas (e.g., only blue shows up).

Line 533: Were machine learning methods used in this study to derive the soil properties at different depths? If so, please specify what algorithm was used and how it was applied in this study. If machine learning methods were used in existing dataset to derive the soil properties, please clarify it.

Line 593: “Earth”.

Summary:

This study presents time series data on hydrometeorological, topographic, and catchment attributes for over 3000 Chinese reservoirs.  The authors have brought together datasets from many disparate sources, including in-situ data/information and satellite products, which is a commendable effort. The methods used are technically sound and the final product derived could be of great value for many purposes including hydrological modeling, water resource management, and ecosystem studies. The results presented provide many insights on reservoir attributes with a large spatial and temporal coverage. Therefore, this study is worthy of publication; however, there are certain issues that require further attention. In terms of presentation quality, the paper is generally well written but is not devoid of certain typos, grammatical errors, unclear statements. The authors should very carefully proofread the entire manuscript before submitting it again. My overall assessment is that the paper can be published after major revisions. I provided my detailed comments below.

Major comments:

L66: I suggest rephrasing the statement, especially for “failed”. The many studies noted by the authors have substantially advanced our ability to better monitor and model reservoirs globally. Perhaps, the datasets could be incomplete and there are more opportunities to develop relatively more comprehensive datasets, but I suggest giving a bit more positive bend to this statement; “failed” seems a bit unfair!

L89-90: some modeling studies that have dealt with such challenges could be cited here including (Dang et al., 2022; Dang et al., 2020; Galelli et al., 2022; Shin et al., 2020)

L84: what does “states” mean here?

L85 and elsewhere: I don’t think a “catchment shapefile” is a “catchment attribute”; file is a file. There are many other such instances where certain terminologies are not properly used. Also, what the “anthropogenic activity” – used in a singular form implies there is one such activity that is being considered.

Section 2.1: Why was 10% threshold used for the GSW data? The same question applies for 20 and 40 meters. Please provide justification. Further, I could imagine all of the many products used in these methods contain substantial uncertainties (being primarily remote sensing based). How would those uncertainties affect the outcomes derived here and how did the authors deal with these issues?

The comment above regarding uncertainty applies to Sections 2.2 and 2.3 as well. I suggest that the authors discuss various uncertainty sources and their impacts.

Figure 3 caption: please add unit to the x-axis of the histograms or provide a note in the caption. I wondered why the panels are organized in this specific order – why not swap (e) and (f) so that the same categories sit adjacent to each other.

Figure 4 and others: The Zenodo link was not active, so I couldn’t make sure if all the datasets were shared. Are all in-situ datasets included in the publicly shared database?

 Figure 7: Why does Res-CN under or overshoot storage for many reservoirs (e.g., panels 7,8 etc.)?

Figure 8: I can’t really tell whether this is a good/bad match between the three? I suggest adding some statistical measures such as RMSE and also a seasonal climatology panel on the right (could be just for the period with observed data).

Figure 9 caption: are these just “topographic” characteristics or in general “catchment” characteristics?

Sections 3.4.2 – 3.4.4: The results and graphics here are nice; however, I wonder what the utility of these data/outcomes are. I suggest that the authors shed some light in the intro section and subsequently in the results section regarding why these specific attributes are chosen, and why/how these are useful, for example, for modeling hydrology considering reservoirs.

Section 3.4.5: Again, why are these specific human activities selected for analysis and how are those useful?

Related to the above comments on the utility of various characteristics, I would suggest adding one figure on the ratio of reservoir storage and/or surface area to catchment size.

Overall/General: the number of reservoirs selected for various purposes is different and validation is provided for a limited subset. Please try to have consistency and expand the validation effort.

Minor/Editorial comments:

• L48, “…especially driven by climate warming and …”: not clear “what” is driven by climate and population; revisions needed.
• L80: should be “altimetry-based reservoir datasets” and “Chinese reservoirs”
• L101: please spell out GEE
• L108: delete “for”
• Figure 1 caption and elsewhere: I suggest “water SURFACE area” instead of “water area”; this applies to Section 2.2 as well.
• L133: please check grammar.
• Figure 3 caption: “dimensionless XX? is indicated ….”

References

Dang, H., Pokhrel, Y., Shin, S., Stelly, J., Ahlquist, D., Du Bui, D., 2022. Hydrologic balance and inundation dynamics of Southeast Asia's largest inland lake altered by hydropower dams in the Mekong River basin. Science of The Total Environment, 831: 154833.

Dang, T.D., Vu, D.T., Chowdhury, A.K., Galelli, S., 2020. A software package for the representation and optimization of water reservoir operations in the VIC hydrologic model. Environmental Modelling & Software, 126: 104673.

Galelli, S., Dang, T.D., Ng, J.Y., Chowdhury, A., Arias, M.E., 2022. Opportunities to curb hydrological alterations via dam re-operation in the Mekong. Nature Sustainability: 1-12.

Shin, S., Pokhrel, Y., Yamazaki, D., Huang, X., Torbick, N., Qi, J., Pattanakiat, S., Ngo‐Duc, T., Nguyen, T.D., 2020. High Resolution Modeling of River‐Floodplain‐Reservoir Inundation Dynamics in the Mekong River Basin. Water Resources Research, 56(5): e2019WR026449.

General comments

Dams and reservoirs play an important role in water resource management and regulation. The authors provided new and comprehensive reservoir datasets over China (the Reservoir dataset in China, Res-CN), which featured reservoir-catchment characteristics for 3254 reservoirs. I have the following concerns for authors in ongoing revision and improvements.

Authors may need to provide more details on why they only focus on reservoirs in China, and why they chose to use GeoDAR while a more recent study published more comprehensive reservoir dataset for China.

Authors have provided comprehensive climatic characteristics (L450~L451) and human activity characteristics of reservoirs but did not explicitly state why these characteristics should be provided. Therefore, I suggest that the authors to offer a more compelling motivation to start their Introduction, and also discuss why these much informaiton is needed for understanding reservoir changes. Otherwise it may look like too much information to digest for certain users.

Although authors argued the need for intermediate catchments, however, I still failed to understand how data for these intermediate ones can be used to understand changes in reservoirs as I thought it may be missing essential water balance components? Can authors add more discussions and also clarify?

Other Comments

L243: there's no need to mention computational time, unless you can provide details on the platform because it is highly platform dependent.

L365~L370: I suggest that the author should place these introductions after L358 (GRSAD and RealSAT) to make the content more cohesive.

L381~L382: Why the time period of reservoir storage variation is from 1984 to 2020, not 1984~2021?

L395: Please explain the two peak values of in-situ in 2021 (Figure.7 and Figure.8).

L425: Please confirm that the time period is 1984-2020.

L437: When the author first introduced the MERIT-river database, please add citations.

L463~L464: Why the time period is different between here (1990~2018) and L456 (1980~2020)?

L524: Why the color of Fig.11i is red?