Overall opinions

The manuscript aims to showcase the WoKaS-Iso database. It consists of very useful isotope measurement dataset for various applications in karst hydro(geo)logy and make the data easily accessible to the whole scientific community. The isotope measurements come with precipitation MSWEP, temperature ERA5 and evapotranspiration GLEAM, giving a ready to use dataset with independent estimations of P, T and ET.

The manuscript is well written and pleasant to read. The figures are well designed and clear. The data are well described and the author give clear recommendation on the use of the data by affecting various flags depending on the data quality. This is really useful for the reader and future user of the dataset.

I would recommend this manuscript for publication in ESSD after addressing some minor concerns given as specific comments below. This initiative is of clear interest for the scientific community and will benefit from additional contributions in the future.

Specific comments

Line 301: It could be nice to have one figure with some example of time series, maybe showing some different behavior across north America, Europe and Asia (since these are regions with the more data).

By looking at some time series, I have the impression that Input-FR-S-0016@Lison and Input-FR-S-0017@Loue are the same data. Maybe need to cross checked

This paper introduces the world's first systematically integrated database (WoKaS-Iso) containing time-series data on oxygen-18 and deuterium-1 isotopes from karst springs and cave dripwater, providing detailed descriptions of data sources, standardization procedures, quality control measures, global product evaluations, and data structure. The establishment of this database holds significant value for karst hydrogeology, paleoclimate reconstruction in caves, and hydrological model calibration, particularly for large-scale comparative studies and data-scarce regions. The comprehensive workload and robust data integration align with the data paper positioning of the ESSD journal. However, the paper still exhibits notable shortcomings in several aspects that require reconsideration following substantial revisions.

1. The text repeatedly mentions that the database is built upon WoKaS and SISAL_mon_v1, but fails to systematically elaborate on its incremental contributions compared to these two foundational databases (e.g., the number of newly added sites, time series length, data types, etc.). It is recommended to include a "Comparison with Previous Work" section in the introduction or methods section to clearly outline the incremental features.
2. The evaluation of the three precipitation isotope modelsIsoGSM, Isoscape, and Sine Curve (Section 2.2.2) was based solely on data from 55 stations, encompassing a mix of cave, spring, and weather station sites, which may introduce bias in the results. The criteria for spatiotemporal matching should be specified more rigorously (e.g., whether matching is performed monthly or whether precipitation amounts are weighted); the assessment findings for δ²H should also be presented in the main text rather than merely mentioned in the appendix.
3. In the quality scoring system (Table 6), "Data Collection" receives 3 points, "Digital Data" receives 2 points, and "Global Products" receives 1 point. This scoring criterion does not account for variations in the accuracy of digital data or differences in the reliability of global products across specific regions. It is recommended to introduce more detailed quality metrics (such as data missing rate, temporal continuity, and deviations from adjacent sites) or at least conduct a sensitivity analysis of this scoring system.
4. Figure 3 simultaneously displays data combination types (colors) and sampling resolution (symbol sizes), making them visually difficult to distinguish. It is recommended to split the data or adopt a more intuitive legend design.
5. Section 2.1.3 contains significant overlap with the content of Figures 4 and 5; it is recommended to consolidate and streamline the content.
6. Although detailed CSV structures and MATLAB scripts are provided, typical use cases (e.g., how to extract time series from a specific site or perform simple analyses using global product data) are lacking. It is recommended to include a concise example or flowchart in Section 3.4, "Usage Notes."
7. L80: "The database incorporates data sourced from 236 springs and 74 caves, totaling 997 time series..." It does not specify how many of these time series contain both δ¹⁸O and δ²H, or how many contain only one of them.
8. The statement L91 ("serving 10%–25% of the global population (Ford and Williams, 2007; Stevanovic, 2019)" cites literature from nearly two decades ago, making its data somewhat outdated; more recent global assessments could be incorporated.
9. L145 “they largely omit δ¹⁸O and δ²H data from karst springs and cave drip waters”GNIP/GNIR. The design itself was not intended for karst water; the criticism seems somewhat excessive.
10. Table 1 contains only figure captions; the main text does not explain the steps illustrated in Figure 1.
11. In Figure 3, colors represent data combinations (flow rate + δ¹⁸O + δ²H, etc.), while symbol sizes indicate resolution; however, the legend does not explain the meaning of these sizes.
12. “only GNIP stations within 25 km were retained”The reason for choosing 25 km was not explained, nor were altitude differences considered.
13. The stated IsoGSM spatial resolution of "200 km" in Table 2 is inaccurate. IsoGSM typically has a T42 spectral resolution (approximately 2.8 × 2.8, or about 300 km); please verify this.
14. Section 3.3 mentions that extraction scripts for three products are provided, but IsoGSM and Isoscape are not included. Explain the reason or provide a simple NetCDF reading example.