Review comments

Thank you, HESS, for inviting me to review this paper. I like the paper as it covers the global scale analysis of various reliable precipitation datasets using a global hydrological model. Selecting the best-performing global dataset for precipitation is meaningful for areas where observation data is scarce.  

Comments

1. There are no keywords provided in this manuscript.
2. What are the values of the x and y-axis of the inset histograms? Mention them in the caption at least.
3. Long sentences in lines 60-66, 87-92, 97-101, 123-127, 196-201, 220-222
4. Line 203, write table 1 as Table 1
5. Line 208, remove or take this “Cohen et al. (2022) report R²=0.99 in 30-year average prediction against USGS gauge data and a global river dataset.” sentence in the discussion section
6. In Figures 3 and 10, you have provided the best-performing precipitation dataset based on annual CC and KGE. However, I can't see the values associated with CC and KGE in the figure; it shows the global distribution of the dataset.
7. In Figures 2 and 5, the authors said KGE values lower than -1 are highlighted in yellow. But as I can see the x-axis it seems different.
8. For Figures 6 and 9, please make a superscript of the units for discharge and area inside the figures and other parts of the manuscript.

General comments:

In the submitted manuscript, the authors assess the efficacy of six global/near-global precipitation datasets in streamflow modeling across 1825 gauging stations, employing the WBMsed hydrological model. Building upon prior research, this investigation scrutinises precipitation datasets that have either been updated or not been included in previous studies. Notably, the assessment extends beyond the daily time scale, including monthly and annual perspectives, and evaluates the model’s proficiency in simulating both high and low flows for each of the selected precipitation sets.

From a linguistic perspective, the manuscript is well-written and exhibits a clear structure. It falls comfortably within the scope of the journal and will capture the interest of members within the hydrological modeling community. However, it is imperative to acknowledge certain significant shortcomings which need to be addressed before the manuscript can be considered for publication.

1. The methods section is comparatively short with a main focus on the description of the precipitation sets evaluated and it does not provide sufficient detail for reproduction of the study.

   1. Eleven of the authors claim their main contribution to this study was extensive data quality control, however, I cannot find a single sentence mentioning or describing the quality control procedure applied in the methods section.

   2. It is unclear which datasets were used to drive the model as all that is given are two references to previous papers and a rather confusing description of some update applied.

   3. The reader is left guessing as how the mismatch between resolution of the precipitation data and the resolution of the hydrological model was addressed.

   4. No rationale is given for how and why the 1825 gauging stations were selected for evaluation. Furthermore, the authors do not provide a list containing the GRDC number and/or name of the selected gauging stations, which makes reproduction of this list impossible (maybe this is planned to be provided in electronic form at a later stage?).

2. According to the abstract, the authors want to provide guidance on the selection of precipitation sets for hydrological modelling. They also make an enormous effort and use five different metrics for performance evaluation. Sadly, the manuscript does not provide much beyond the pure quantitative evaluation. Giving a rationale for the selection of those specific metrics and discussing why one/some of them provide a benefit for the evaluation under certain conditions would tremendously improve the manuscript. It would also be of value if the manuscript could discuss the performance within specific climatic zones (e.g tropics vs temperate) as suggested in the abstract.

3. The discussion section would benefit from further editing. It contains paragraphs/sentences which would be better placed in other sections (see specific comments below) and it does not discuss any limitations of the study.

4. Presenting maps at a global scale is no easy task and a lot of effort has been put into a substantial number of figures. However, it is very difficult to actually see what is going on. The coloured points often overlap each other and it is not easy to distinguish the colours or to discern the differences between the precipitation sets. Providing maps limited to regions which exhibit major differences between the sets might be a better option. There also seems a consistent problem within the caption where ‘yellow’ lines or highlights are referred to but the presented colour seems to be more ‘orange’ or ‘red’. Replacing the coloured line with a dashed black line and highlighting the bars in grey would avoid any potential colour confusion. Furthermore, the colour legends seems to miss a title describing what value is actually presented.

Specific comments:

Line 30: The term ‘more than 1800’ is rather vague and should be changed to the exact number of stations.

Lines 34-37: The sentence seems overly complex. Removing the vague quantification and replacing it by the information in the brackets should address the issue.

Line 88: Again, the expression ‘multiple’ is rather vague and should be changed to the exact number of sets evaluated. ‘

Line 88-91: Listing all those example sets bloats the paragraph. I recommend to remove the list of example sets in the brackets.

Line 92-93: Same as in line 88-91

Lines 86-105: The authors mention Voisin et al. (2008) but do not give any information on the scope of this cited study or how it relates to the presented work.

Line 121: I suggest changing the header to ‘Precipitation sets’

Line 122: What is the rationale behind choosing datasets of length >30 years? Why those specific six sets?

Line 130-136: It is not entirely clear which version of the set has been used. Presumably ERA5-Land? Why was it selected and was it used/evaluated in previous studies?

Line 137-147: Which version has been used and at what resolution? Why was CHIRPS selected for evaluation?

Line 148-161: The description of the set is very detailed and I recommend to re-consider listing all the different data sets which were combined to create MSWEP. Instead you could provide information on which temporal resolution was selected and why this set was chosen for evaluation.

Line 162-168: Why was TERRAClimate selected for evaluation? Has it been used in other studies?

Line 169-174: Why was CPCU selected, what is the temporal and spatial resolution of this set? Has it been used/evaluated in previous studies?

Lines 121-184 Section 2.1: The section provides a description of the sets (some more extensive than others) but does not provide any information on how the sets were prepared for use in the hydrological model. As the model was run at daily time steps and 0.1° resolution I assume that most of the sets were interpolated to the model’s resolution?. Furthermore, there is no information provided on how the monthly data in TERRA was converted to daily data.

Line 194: What does high-resolution mean in this context? Is it finer, equal or coarser than the 0.1° resolution at which the model was run?

Lines 196-201: It is unclear which sets have been used to drive the model. Please, provide a table listing all the data sets and where to obtain them. Has the model been calibrated? Which parameter sets were used?

Lines 202-204: It is mentioned that the option to disaggregate TERRA from monthly to daily time steps was not used. If that is the case, I wonder how the model could be run at daily time steps. Was another method used for disaggregation?

Lines 205: The model was run at 0.1° resolution but several of the evaluated precipitation sets have a different resolution. How was this mismatch addressed? Did you apply any interpolation?

Lines 208-209: The last sentence in this section (Cohen et al. (2022) report … ) should be removed. It could be used in the discussion section.

Lines 211-218: How were the stations selected? Were stations limited to those with a record of > 10 years ? Were gaps allowed in the record? Will a list of the stations be provided? As 11 authors claim to have performed extensive quality control, at least one full paragraph outlining the steps involved should be provided.

Lines 237-239: I recommend to remove these two sentences.

Lines 259-261: Does the figure show monthly or annual CC values in the histograms? I assume the ‘yellow’ line refers to the ‘orange’ or ‘light red’ line in the inset?

Lines 265-267: The description reads ‘values lower than -1 are highlighted in yellow’ but I cannot see any yellow highlights, just an ‘orange’ bar and an ‘orange’ line. Furthermore, the highlighted bars seem to refer to values between -0.8 and -1.0 rather than to those < -1.

Line 268 (Figure 3): I understand the reason behind presenting this map but it is really hard to see what is going on. Maybe limiting the presentation to areas in which there are clear differences between the datasets would be better?

Lines 280-282: ‘Yellow’ looks actually ‘orange’, see above.

Lines 292-296: The same issue as in Lines 265-267.

Line 312 (Figure 6): This figure is really difficult to read. I would suggest to remove the annotation stating the catchment area as this information is already provided in Table 2 and because it distracts from the time series. Can the readability be increased by showing one single time series per row and increasing the height of the figure?

Line 327 (Figure 7): Yellow’ looks ‘orange’, see above.

Line 330 (Figure 8): Yellow’ looks ‘orange’, see above.

Line 349 (Figure 9): I have a hard time understanding what is going on in this figure. Reading the description I was expecting much less cluttered plots than presented. Why are there so many points per catchment? On what time scale did you actually calculate the extremes? I suggest, you add a section in the methods detailing the calculation of your Q10/Q90 to clarify potential misunderstandings. Furthermore, I suggest to remove the annotations presenting the lat, lon and area information as this is already given in Table 2 and is unnecessary here.

Line 355-372: I suggest to remove this paragraph and put it into the introduction instead to provide more background information.

Line 373-374: This first sentence (In light of the above … for global hydrological modelling) better suits the introduction as it outlines the aim of the study.

Lines 374-376: This is a crucial bit of information which I was looking for in the methodology (It is important to note … replicating observed river discharge). It should be placed into the section describing the model setup.

Lines 376-387: This sentence (Within this context … various precipitation sets.) would be better placed into the introduction, outlining the objectives of the study.

Lines 379-380: This sentence (Based on the evaluation … better than other datasets) is an excellent opening sentence for the discussion section.

Line 400-401: Why is it necessary to present a new figure here? The variability in performance has already been shown in Figure 3. This last sentence should be removed.

Line 402 (Figure 10): I do not think that this figure provides anything that has not yet been covered by the other figures. Better place it into the supplemental material.

Technical corrections:

Line 33: Remove comma after ‘Whilst’

Line 99: This is the first time ERA5 is mentioned in the main body and the acronym hence needs explaining. Should be changed to ‘such as ECMWF Reanalysis v5 (ERA5)’

Line 101: Is it PERCCDR (as in the abstract and methods) or PERSIANN-CCS-CDR? Please, double check consistency of acronyms.

Line 123-127: Some of the acronyms have already been defined in Lines 99-101 and do not need to be defined again.