Referee report for: GRDC-Caravan: Extending Caravan with Data from the Global Runoff Data CentreGeneral Assessment

This manuscript introduces the GRDC-Caravan dataset, which extends the Caravan hydrology dataset by incorporating discharge data from the Global Runoff Data Centre (GRDC). By expanding spatial and temporal coverage to over 5,000 stations in 25 countries, it’s clear that this work makes an important contribution to large-sample hydrology (LSH). The dataset is particularly valuable for underrepresented regions, providing a solid foundation for global hydrological research while aligning with open science principles. Personally I feel impresed by both the valuable job of GRDC as institution and the intitiative of the team to further contribute with this Caravan extension. I am sure thart this will provide a powerfull new dataset for the hydrology community. While the manuscript is undeniably significant and deserves publication, there are a few minor areas that need improvement to meet the standards of Earth System Science Data (ESSD). Addressing these issues will help improve clarity, structure, and reproducibility. With these adjustments I believe that the paper will be ready for the journal.

Suggestions for improvement

1. Manuscript Structure:

The manuscript lacks a clear distinction between methods and results, making it difficult to follow.

• Please consider spliting the "Dataset Description" section into two separate sections: "Methodology" and "Dataset Description." 

1. Figures and Captions

I have the impression that while the figures are visually informative, they lack context and detailed descriptions.

• Improve captions to explain the figures' relevance to the dataset's purpose
• Provide more discussion of the figures in the main text to guide readers.

1. Comparisons with Existing Datasets

I feel that the paper presents a limited discussion of how GRDC-Caravan compares with other LSH datasets.

• Add a detailed overview of LSH releases. Where did it start? MOPEX? CAMELS? Which are the main datasets currently available? Which countries are covered? How did Caravan arrive to existence? What about other datasets and similar initiatives? GSIM? How is the situation of LSH datasets by the end of 2024?
• Include a comparative analysis highlighting GRDC-Caravan’s unique contributions (e.g., coverage in underrepresented regions, temporal/spatial resolution, data quality). Remember that you should sell the idea of why GRDC-Caravan matters.

1. Data Records and Repository Description

I feel that there is still insufficient details about the organization of the Zenodo repository and dataset files.

• Describe the repository structure, including folders and file organization. Of course, a further description of the data included, and units is already provided by Caravan, but for users, it is always beneficial to know what to expect after the download.
• L245: Specify the version number of the dataset being referred to on Zenodo. Pay attention to always place the specific version being described in the paper, not the general DOI for all versions.

1. Reproducibility and Code Availability

The authors state that the GRDCFlowTools R package is only available upon request, limiting reproducibility.

• If possible, publish the package on a public repository (e.g., GitHub or CRAN).
• Include usage examples to facilitate adoption and reproducibility.

1. Duplicates

While the authors acknowledge, the duplicated catchments are not explicitly labeled in the dataset.

• If possible, consider  adding a column that identifies duplicated catchments in the dataset.

1. Dataset Analysis and Validation

This paper already covers a clear description of the dataset being published. But to be fair to other similar dataset papers published in ESSD, it indeed lacks a bit of a overview and preliminary analysis of such dataset, which could also act as a technical validation of what is being published. While I see that the GRDC makes available data which has been checked by the NMHS, I think that future users could benefit from a general overview, which could also act as a validation. Personally, I do not see the need to perform any modelling analysis with the data as previous artiodactyls papers have done.

• Include plots showing spatial distributions of key streamflow signatures and climatic indices (e.g., Q mean, BFI, aridity).
• Try to include a creative plot showing an overview of this new discharge data for users. Why not some small boxplots of the Q mean covered among the GEnS climatic zones? Something inspired in Figure 4 by Marvin Höge et al. 2024 in CAMELS_CH? Take this as a suggestion (inspiration), not a requirement.
• Validate catchment boundaries with visual aids or references to prior successful applications of MERIT Hydro and delineator.py.

1. Outlook Section

The outlook could be more concise and impactful.

• Consider Emphasizing more the televance of GRDC-Caravan for global hydrology. Describe the need for new Caravan extensions, and the Importance of open data policies from NMHS.
• Streamline the section to improve rhythm and flow.

Specific points:

General:

Some sentences are long and complex, making them hard to follow. Please while reviewing the manuscript make sure to go through such points. For example:

L 171: Consider rephrasing "After the calculation of all potential pour points was completed, a rating score R was calculated...” to: "Once all potential pour points were calculated, a rating score (R) was assigned based on the following formula."

Introduction

L33–34: Please consider to rephrase "is the primary objective of the principal reason" to something in the lines of  "The primary objective of the Global Runoff Data Centre (GRDC) is to (…)."

Dataset description

L115: Acknowledge that Figure 3 adapts elements from the original Caravan publication. Add a note in the caption, e.g., "This figure is adapted from the original Caravan publication (Kratzert et al., 2023)."

Review of the Manuscript by Färber et al.

This is my first review of the manuscript by Färber et al. The manuscript addresses a relevant topic and aligns well with the journal's scope. Utilizing a river discharge dataset with an extended record length is valuable for numerous research applications. However, I found the paper to be poorly organized at times and not always clear in its presentation. Furthermore, it lacks a thorough discussion of the novel contributions of this dataset, remaining overly generic in certain aspects.

Abstract

The abstract is unclear and should explicitly state the manuscript's objective, which only becomes apparent after a careful reading of the paper.

Introduction

The introduction includes essential material but could benefit from improvement, particularly in its concluding section, which is overly cryptic. The introduction should conclude with a clear and well-articulated statement of the manuscript's objective.

Dataset Description and Methods

The manuscript conflates the dataset description with the methods. These should be clearly separated for better clarity. For instance:

• Figure 1 is not cited in the text.
• River discharge data should separately describe the current GRDC dataset and the Caravan dataset.
• Catchment attributes and DEM should be included as part of the dataset description.

Outlook Section

The outlook section lacks a crucial comparison between the datasets prior to and after the extension. It should emphasize what has been improved, the capabilities now available that were previously unattainable, and the added value of this new dataset.

Additional minor comments are reported in the annotated pdf.