Large-scale isotopic fingerprinting of cryosphere and hydrological components in glacierized catchments

Vital, Melanie; Jara, Edison; Wade, Andrew John; Masse-Dufresne, Janie; Persoiu, Aurel; Temovski, Marjan; Saidaliyeva, Zarina; Vreča, Polona; Dàvila Roller, Luzmila; Shahgedanova, Maria; Pu, Tao; Fernandoy, Francisco; Jeonghoon, Lee; Nishonov, Bakhriddin; Ramirez, Edson; Vystavna, Yuliya

doi:10.5194/essd-2026-173

Preprints

https://doi.org/10.5194/essd-2026-173

Preprints

12 May 2026

| 12 May 2026

Status: a revised version of this preprint is currently under review for the journal ESSD.

Large-scale isotopic fingerprinting of cryosphere and hydrological components in glacierized catchments

Melanie Vital, Edison Jara, Andrew John Wade, Janie Masse-Dufresne, Aurel Persoiu, Marjan Temovski, Zarina Saidaliyeva, Polona Vreča, Luzmila Dàvila Roller, Maria Shahgedanova, Tao Pu, Francisco Fernandoy, Lee Jeonghoon, Bakhriddin Nishonov, Edson Ramirez, and Yuliya Vystavna

Abstract. Stable water isotopes (SWI; δ¹⁸O and δ²H) are widely used tracers for identifying water sources and reconstructing hydroclimatic processes in glacierized catchments. However, globally harmonized isotope datasets that integrate multiple cryospheric and hydrological endmembers remain limited. Here we present a global georeferenced database of stable water isotope measurements compiled across glacierized catchments worldwide. The dataset contains 12,348 individual δ¹⁸O and δ²H measurements collected from 63 publications, covering 19 countries across six continents. The temporal coverage of the isotope measurements spans from 1978 to 2023, based on data compiled from publications starting in 1981. The database integrates isotope measurements from precipitation, snowpack and snowmelt, glacier ice, firn, glacial and supraglacial meltwater, rock glaciers, ice-cored moraines, permafrost thaw waters, talus slopes, streams, lakes, and groundwater. Each record includes standardized metadata describing geographic location, elevation, sampling period, endmember classification, and analytical methods. Across the compiled dataset, δ¹⁸O values range from approximately −31 ‰ to −0.36 ‰, while δ²H values range from −247 ‰ to 0 ‰, reflecting strong variability in elevation, temperature, and moisture source conditions across glacierized environments. Cryosphere endmembers such as glacier ice and glacial meltwater typically exhibit depleted signatures (δ¹⁸O ≈ −14 ‰; δ²H ≈ −100 ‰) reflecting precipitation formed under colder climatic conditions and preserved within glacier storage, while groundwater and firn waters tend to show relatively enriched compositions due to mixing, recharge processes, and seasonal precipitation inputs. By harmonizing isotope observations across multiple cryosphere and hydrological components, this database provides a new global reference for identifying characteristic isotopic signatures of glacier-derived waters and their downstream mixing with other hydrological sources. The dataset supports comparative studies of glacier–hydrology interactions, endmember mixing analysis, and isotope-enabled hydrological modelling, contributing to improved understanding of water resources in glacier-fed catchments under changing climate conditions. The dataset is publicly available at https://doi.org/10.5281/zenodo.19062383 (Vital et al., 2026).

Received: 04 Mar 2026 – Discussion started: 12 May 2026

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Preprint (PDF, 1972 KB)

Supplement (610 KB)

Download & links

Status: final response (author comments only)

RC1:
'Comment on essd-2026-173', Anonymous Referee #1, 21 Sep 2026
Reviewer Report
Title: Large-scale isotopic fingerprinting of cryosphere and hydrological components in glacierized catchments
This manuscript presents a global database of stable water isotope observations from glacierized catchments, integrating δ¹⁸O and δ²H measurements from precipitation, snowpack, snowmelt, glacier ice, firn, glacier meltwater, supraglacial meltwater, rock glaciers, ice-cored moraines, permafrost thaw waters, talus slopes, streams, lakes, and groundwater. The authors report 12,348 individual isotope records compiled from 63 publications and distributed across six continents. The database contains extensive metadata concerning geographic location, sampling period, elevation, endmember classification, analytical methods and isotope measurements. The manuscript addresses an important need in isotope hydrology because globally harmonized datasets that combine cryospheric and hydrological endmembers remain limited. The public availability of the database and associated code is also a major strength. The discussion of analytical, environmental, temporal, spatial and metadata-related uncertainties is appropriate and demonstrates awareness of the challenges associated with compiling heterogeneous isotope observations. However, several aspects require substantial clarification and revision before the manuscript can be considered suitable for publication. In particular, the manuscript currently moves from data compilation to global hydroclimatic interpretation and statistical inference without sufficiently demonstrating that the underlying observations are comparable or statistically independent. The statistical methodology is also insufficiently described for the analyses presented in Section 4.6. Furthermore, several numerical and methodological inconsistencies occur throughout the manuscript.
Major Comments
The database construction and literature-search methodology needs substantially more detail. The manuscript states that the database was constructed from 63 publications using searches primarily in Web of Science and Scopus and provides several search terms. However, the procedure is not sufficiently reproducible. The authors should provide: exact search strings; dates on which the searches were conducted; databases searched; inclusion and exclusion criteria; screening procedure; number of records/publications identified initially; number excluded at each stage and reasons for exclusion; procedure for identifying duplicate datasets; criteria for deciding whether a dataset from a publication was sufficiently documented for inclusion; procedure for handling supplementary datasets; procedure for incorporating unpublished data generated by the authors. A PRISMA-like flow diagram, even if adapted for a database-construction study rather than a systematic review, would substantially improve transparency. This is particularly important because the paper describes the database as a global compilation, while the resulting coverage is highly uneven.

There are important inconsistencies concerning the temporal and geographic coverage. Several statements concerning the temporal and spatial extent of the database appear inconsistent. For example, the Abstract states that the database contains data from 19 countries, whereas Section 2.2 states that it covers 20 countries. The Abstract also states that isotope measurements span 1978-2023, whereas Section 2.2 states that the temporal coverage extends from 1981 to the most recent publications up to 2025. These discrepancies need to be resolved throughout the manuscript, including: number of countries; number of catchments; earliest isotope observation; latest isotope observation; publication-year range; sampling-year range. The authors should clearly distinguish publication year from sampling year, as these are not equivalent. Besides lot of work on glaciers/cryosphere using stable water isotopes in the Indus River Basin has been carried out which is completely missing when it comes to Himalaya or Hindu-kush (e.g. Jeelani et al., Lone et al., 2021; 2022; etc.).

The manuscript needs a much clearer distinction between observations, samples and records. The manuscript repeatedly uses terms such as individual entries, measurements, samples, records, and catchments. It is not sufficiently clear whether one database row represents: one physical water sample; one isotope measurement; one sampling event; one observation extracted from a publication; or one sample aggregated from a publication table. The statement that the database contains 12,348 “individual entries” and excludes averages needs clarification. This distinction is particularly important because the statistical analysis appears to treat the individual isotope values as observations. I recommend adding a precise definition of record, sample, sampling event, study, catchment, and publication.

Statistical independence and pseudo replication are major concerns. This is, in my view, one of the most important issues in the manuscript. The database combines thousands of measurements from a relatively small number of publications and catchments. Consequently, the individual observations cannot automatically be considered statistically independent. For example, hundreds of measurements from the same catchment, publication, monitoring station or laboratory may contribute disproportionately to a particular continental or endmember distribution. Nevertheless, Section 4.6 reports pairwise statistical comparisons and Dunn's tests, concluding that most endmembers are statistically distinguishable. The authors need to explain: What constitutes the statistical unit? Are individual isotope measurements treated as independent? Were observations weighted equally? Were measurements grouped by catchment? Were multiple observations from the same publication treated as repeated observations? Were differences among studies/catchments/laboratories considered? Was hierarchical structure accounted for? Were sample sizes dominated by a few highly sampled catchments? A conventional Dunn test applied directly to thousands of pooled measurements may produce very small p-values simply because of the very large sample size, even when the practical separation among endmembers is modest. The authors should therefore reconsider whether the reported significance testing is statistically appropriate. A hierarchical or study/catchment-aware approach may be more appropriate, or the authors should explicitly justify why individual observations can be treated as independent.

The statistical methodology is inadequately described. Section 4.6 reports: “The post-hoc Dunn’s test results indicate that most cryosphere and hydrological endmembers are statistically distinguishable…” However, the Methods section does not adequately describe the statistical workflow used to obtain these results. The manuscript should explicitly state: which omnibus test was performed before Dunn's test; whether Kruskal–Wallis tests were used; assumptions/checks; treatment of missing values; multiple-comparison correction; correction method used; significance threshold; software/package used; exact statistical model; whether analyses were conducted globally and separately by continent; sample size used in each comparison; whether outliers were retained. Without these details, the analysis cannot be independently reproduced.

Statistical significance is sometimes interpreted as hydrological separability. The manuscript states that statistically significant differences demonstrate “robust separation among major water sources.” This interpretation needs to be substantially moderated. A statistically significant difference in isotope distributions does not necessarily mean that the two endmembers are sufficiently separated to be useful in an isotope mixing model. For isotope hydrograph separation or EMMA, the relevant question is not simply: Are the two distributions statistically different? but rather: Are they sufficiently distinct relative to their within-source variability and analytical uncertainty to provide useful discriminatory power? The authors themselves later acknowledge that endmembers should be defined carefully and that uncertainty needs to be considered in mixing analyses. This is an important point, but it should be integrated more explicitly into the statistical interpretation. I recommend reporting effect sizes and distributional overlap in addition to p-values.

The treatment of spatial sampling bias requires stronger quantitative assessment. The authors correctly acknowledge that isotope observations are unevenly distributed geographically, with substantial clustering in Europe, Asia and North America and very limited representation in Africa and Oceania. This is an important limitation, but the implications are not fully incorporated into the subsequent global analyses. For example, the manuscript compares continental distributions despite extremely different numbers of observations: Europe: 5,899; Asia: 3,132; North America: 3,022; South America: 249; Africa: 32; Oceania: 12. With only 12 observations in Oceania and 32 in Africa, continental-scale conclusions are necessarily very limited. The authors should distinguish clearly between: “distribution of available observations” and “representative distribution of isotope compositions across the continent.” The latter cannot be inferred from the current sampling density.

Table 2 requires clarification concerning what constitutes a “catchment”. Table 2 compares the number of catchments in the database with global glacier area and ice volume. However, the manuscript needs to define precisely how a catchment was identified. For example: Does a publication containing samples from three sites within the same drainage basin count as one catchment? Are sub-catchments counted separately? Are multiple studies in the same catchment counted once or multiple times? Are catchment boundaries derived from the source publications or standardized using a GIS dataset? This is particularly important if the table is intended to quantify geographical representativeness.

The claim that the database provides “characteristic isotopic signatures” should be moderated. The manuscript repeatedly refers to characteristic isotope values of different cryosphere endmembers. For example, the Abstract describes glacier ice and glacial meltwater as typically having δ¹⁸O around −14‰ and δ²H around −100‰. However, the manuscript itself demonstrates substantial overlap among several endmembers. This raises an important conceptual issue: there may not be a globally universal isotopic fingerprint for a particular endmember such as glacier meltwater or snowmelt. The observed isotope composition depends on: elevation; latitude; precipitation season; moisture source; continentality; accumulation history; snowpack evolution; sublimation; refreezing; melt timing; glacier age; geographic setting. Therefore, I suggest replacing language implying universal “fingerprints” with terminology such as observed isotopic distributions, regional ranges, or empirical isotope signatures within the compiled dataset, where appropriate.

The interpretation of d-excess is too categorical. The manuscript interprets d-excess largely in terms of moisture-source relative humidity, sea-surface temperature and kinetic fractionation. This is broadly reasonable, but the interpretation of d-excess in snow, glacier ice, groundwater and meltwater is more complicated because secondary processes can substantially modify d-excess. The authors themselves recognize the roles of: sublimation; snow metamorphism; evaporation; refreezing; atmospheric exchange; groundwater mixing. Therefore, statements such as elevated d-excess indicating particular moisture-source conditions should be framed more cautiously, especially for secondary hydrological endmembers. For example, the statement that elevated Asian d-excess “reflects a strong influence of moisture derived from relatively dry, cool oceanic sources” requires stronger evidence than the isotope distribution alone provides.

The interpretation of Figure 2 and the regression requires revision. The manuscript reports a regression slope of approximately 8.23 and intercept of approximately 15‰, and interprets this as indicating particular fractionation and moisture-source processes. However, several issues need clarification: Was ordinary least squares regression used? Were all 12,348 measurements included? Were observations weighted? Were different endmembers pooled? Were repeated measurements from the same catchment treated independently? Was uncertainty in both δ¹⁸O and δ²H considered? Were outliers included? What are the confidence intervals and R²? Is the reported slope statistically different from 8? Pooling different water types from different environments and then fitting a single global regression can produce a line that does not necessarily represent a physically meaningful meteoric water line. I recommend reporting the regression methodology and uncertainty explicitly and avoiding a strong process interpretation unless supported by additional analysis.

Figure 3 requires greater transparency regarding sample size. The figure caption states that sample sizes are shown for each category. However, because the sample sizes differ substantially among endmembers, the figure should be accompanied by either: an explicit table of n values; median and interquartile range; preferably confidence intervals; and, where relevant, number of independent catchments/studies contributing to each category. A category with thousands of observations from a few studies should not be visually or statistically interpreted in the same way as a category with tens of observations from several independent catchments.

The manuscript should distinguish measurement uncertainty from environmental variability. The uncertainty section is conceptually useful, but the manuscript sometimes combines several fundamentally different uncertainty sources. Analytical uncertainty, sampling variability, spatial heterogeneity and between-study methodological differences are not interchangeable. For example: analytical precision is measurement uncertainty; storm-to-storm isotope variability is environmental variability; different sampling seasons represent sampling design; differences among laboratories represent inter laboratory uncertainty; differences among catchments represent spatial variability. The manuscript should provide a clearer uncertainty framework, perhaps separating: Measurement uncertainty-sampling uncertainty-environmental variability-representativeness uncertainty- database harmonization uncertainty. This would make the database substantially more useful to downstream users.

Quality-control procedures need to be documented more rigorously. The authors state that outliers were flagged rather than removed and provide thresholds for δ¹⁸O, δ²H and d-excess. However, the rationale for these thresholds is not sufficiently explained. For example: Why was δ¹⁸O outside -40 to +20‰ selected? Why was δ²H outside -320 to +120‰ selected? Why is d-excess outside 0-25‰ flagged? Are these physical thresholds, statistical thresholds or screening thresholds? Are flagged observations included in Figures 2-6? Are they included in the statistical analyses? How many records were flagged? How many were subsequently verified or retained? This should be reported quantitatively.

Harmonization to VSMOW/VSMOW2 needs more explanation. The manuscript states that isotope values were reported relative to VSMOW/VSMOW2 and that data were reported to VSMOW-SLAP using two-point normalization. However, the database spans several decades and multiple laboratories. The authors should explain how measurements reported using older standards or calibration schemes were harmonized. In particular: Were original isotope values converted? Were values simply retained as reported? Were laboratory-specific corrections available? Were data measured before adoption of current reference materials treated differently? This is important for a global dataset where differences of less than 1‰ may be scientifically relevant.

The use of “EMMA” terminology should be reconsidered. The manuscript states that endmember labels were standardized to a controlled vocabulary used in endmember mixing analysis (EMMA). However, the database contains categories that are not necessarily conventional endmembers in a strict mixing-model sense. For example: groundwater; stream water; lakes; talus water; permafrost thaw; supraglacial water. Some of these can themselves represent mixtures rather than chemically or isotopically discrete sources. I suggest distinguishing between: source/endmember categories and observed hydrological compartments. This distinction would make the dataset much more useful for mixing-model applications.

Some interpretations are stronger than the data support. Several conclusions appear to infer hydrological processes from isotope distributions alone. For example, the manuscript states that stream waters show intermediate values indicating mixing between cryosphere and meteoric sources and that these values imply dominant glacier/snowmelt contributions in some regions. An intermediate isotope composition does not by itself quantify the contribution of each source. Without discharge-weighted source sampling, temporal sampling, or mixing-model analysis, isotope values alone should not be interpreted as demonstrating a dominant contribution. The manuscript should carefully distinguish: “consistent with mixing” from “demonstrates/indicates a dominant contribution.”

The continental analysis should be substantially revised. The continental analysis is potentially interesting, but several interpretations are based on very different sample sizes and climatic/geographic mixtures. For example, Europe contains 5,899 observations while Africa contains only 32. Consequently, statements about continental differences should explicitly include: n observations; n studies; n catchments; geographical coverage; elevation range; climatic zones represented. The authors should avoid implying that the observed distributions are representative of entire continents.

The 60° latitude division for North America needs stronger justification. The manuscript separates North America into >60° and ≤60° latitudes to distinguish Arctic/sub-Arctic from temperate/alpine settings. This is a useful exploratory division, but the choice of 60° is somewhat arbitrary and may combine highly different mountain and climatic systems within each group. The authors should either: justify the threshold more rigorously; conduct sensitivity analyses using alternative boundaries; or describe this explicitly as an exploratory stratification.

Figure 4 and the continental comparisons should include independent sample counts. The figures show boxplots of continental isotope distributions, but the manuscript would be much stronger if the figure or supplementary information reported: n observations / n studies / n catchments for every continent × endmember combination. This would allow readers to immediately distinguish well-constrained distributions from poorly sampled categories.

The manuscript should provide a comprehensive data dictionary. The manuscript states that the database contains 44 fields. A complete data dictionary should be provided in the Supplementary Information or Zenodo repository. This is particularly important for an ESSD data publication.

The conclusions should be more focused on the database rather than broad hydroclimatic conclusions. The strongest contribution of the manuscript is the database and its harmonization, rather than the broad global interpretation of isotope patterns. The conclusions would therefore benefit from emphasizing: database scope; spatial and temporal coverage; endmember categories; metadata; quality-control procedure; uncertainty; limitations; applications; future updates. Some broader process interpretations should be shortened or moved to a more clearly qualified discussion.

Data and code availability should be made more explicit. The manuscript provides Zenodo DOIs for both data and code. This is a major strength. However, the data availability statement currently contains a grammatical error: “Data described in this manuscript can are publicly available…” The statement should be corrected and should specify: exact dataset version; exact code version; whether the Zenodo record is versioned; whether all supplementary tables are included; whether raw source data can be traced to the original publications. The manuscript currently identifies the dataset and software repositories, which is good practice.
Citation: https://doi.org/10.5194/essd-2026-173-RC1
RC2:
'Comment on essd-2026-173', Anonymous Referee #2, 24 Sep 2026
General comments
This manuscript presents a compilation of more than 12,000 δ¹⁸O and δ²H measurements from glacierized catchments around the world. Rock glaciers, ice-cored moraines, talus slopes and permafrost thaw are rarely sampled, and having them gathered in one place is very useful. I also think it was the right decision to keep individual measurements rather than site averages. A dataset like this has clear value for isotope hydrology and glaciology, and it is a good fit for ESSD. I enjoyed reading the paper.
I went through the manuscript and also looked at the Zenodo file. My first three comments concern the dataset itself and how it is documented (quality control, compilation and metadata, terminology), since users will rely on this paper to understand what they download. The next two concern the geographical coverage and the interpretation of pooled global distributions, which I think goes further than the data allow. The last one deals with the d-excess discussion and with places where the text does not match the figures.
None of this is out of reach, but together it amounts to a fair amount of work, so I recommend major revision.
Specific comments
1) Quality control and data file
Section 2.2 states that outliers were flagged rather than removed (L179–189). This is good practice, but I could not find these flags in the file: there is no flag column, and some values far outside the stated bounds are not marked (see below). The "Parameters definition" sheet of the file is a very useful start. It would help to add the flag field and the units where they are missing, and to align it with the text, which mentions 44 fields (L166) while the file and the sheet contain 39. The "std", "error" and "uncertainty" fields also overlap in meaning and are unevenly filled ("std" is almost empty), so a clearer definition, or merging them, would help users. At L320–323, the authors also suggest that the spread of values within each category gives insight into both environmental variability and potential analytical uncertainty. I would nuance the second part, since that spread is dominated by differences between sites and climates, which makes analytical uncertainty hard to see.
A few things that are not mentioned in the manuscript came up when I checked the file. Some of them may have a simple explanation.
The d-excess distribution stops just above 0 ‰, the lower flag bound, with no negative value anywhere, while high values are kept without any mark. This looks as if negative values were removed rather than flagged. It matters for the interpretation too. At L360–364 and L382–385, the lack of strong isotopic enrichment in lakes is explained by cold conditions, in contrast with arid settings, whereas Sect. 4.4 (L526–530) attributes their lower d-excess to evaporation and variable residence times. It would help to reconcile these two readings. Aridity is not limited to warm regions, and several sites in the database may be cold and arid. There, evaporated waters with low or negative d-excess would be expected, and where evaporation signals are absent, a short residence time may be the better explanation. I would suggest restoring the removed values, if any, and adding a nuance on this point.

A few entries look wrong, for example an unrealistic δ¹⁸O value (record 7081) or publication years (and data-owner labels) increasing by one per row from 2015 to 2071 for Gordon et al. (2015). These are most likely cells shifted or auto-incremented during copy-paste in a spreadsheet, which is easy to fix, and a simple check of value ranges and coordinates should catch any remaining cases.

About a dozen pairs of records share the same site, date and isotope values, mostly in Michelon et al. Some may be laboratory replicates, which would be worth indicating in the file; otherwise they should probably be removed.

Could the authors explain how the groundwater samples from Mount Fuji fit the selection criteria? I may have missed the link with glacier or snow processes, and a sentence would help readers understand why they are included.

2) Data compilation and metadata
Section 2.2 describes the compilation in a few lines: 63 publications, listed in Table S1, plus unpublished data, with individual values only, no averages (L156–166). The databases and keywords used for the search are given (L161–164). Since the paper will be the main reference for the dataset, I would like to see a little more on how it was put together: the date of the search, the number of studies found and screened, the criteria used to include or exclude them, how values were extracted (tables, supplements, digitized figures, contact with authors), how unpublished data were measured and checked, and how overlaps between papers or with GNIP/GNIR were handled.
Also, only 63 publications seemed to me a little surprising given how much has probably been published, and a short note on why some studies were not included would be helpful. Section 3.4 states that the coordinates and elevation of sampling locations were retained (L294–297), and this is the case for most records. In some catchments, however, different water types share a single coordinate, sometimes rounded to the degree, which suggests a catchment-level location. Would it be possible to give the sampling location for these records as well, or at least a field stating whether coordinates refer to the sample or the catchment? This would make later reuse and reinterpretation much easier, for instance relating samples to elevation, distance from the glacier or local topography. It would also help to state where elevations come from and what counts as a catchment.
Regarding L303, it would be useful to indicate for each record whether the original study reported a calibration against VSMOW and SLAP (or VSMOW2/SLAP2), or whether this was assumed, as it is hard to verify for older data. The "number of samples" field mentioned at L169 does not appear in the file, and it would in any case be at odds with "single data" (L158). Finally, the number of countries differs between the abstract (19) and L160 (20), and the temporal coverage at L160–161 seems to refer to publication years rather than to the sampling period given in the abstract.
For precipitation, the text acknowledges that samples may be single events, daily or monthly composites (L257–259), and later states that they are mostly monthly (L625). The sampling interval and the representativeness of each sample are crucial here: an event sample and a seasonal composite do not carry the same information, and they cannot be meaningfully pooled or averaged without weighting by precipitation amount. The file has sample period fields, but they are filled for only a small part of the precipitation records. I would suggest completing them where possible, and adding the phase and amount of precipitation.
3) Use of the term "endmember"
Throughout the paper, the 14 categories are called endmembers. I would suggest calling them "water types" and keeping "endmember" for the way the data may be used in mixing models. Section 2.1 itself defines endmembers as pure or unmixed sources, and Table 1 describes streams and groundwater as integrating several contributions. Streams are precisely the mixtures that EMMA aims to decompose, and lakes and groundwater are themselves mixtures, even though groundwater is often used as an endmember for streamflow: they integrate rain, snowmelt, ice melt or subsurface water in proportions that vary in space and time, so their composition is an outcome of mixing rather than an input to it. Even snowpack or glacier ice act as endmembers only with respect to a given mixture in a given catchment. A reader less familiar with mixing models could otherwise take lakes or groundwater for unmixed sources. Simple rules for assigning samples to categories (e.g. glacial meltwater vs. stream) would also help users. Note also that the file contains a fifteenth category, "snow" (two records), which is defined in the file but not in Table 1.
4) Spatial coverage (Sects. 2.3 and 4.3)
Section 2.3 concludes that North America, with 18.8 % of the catchments, is "broadly consistent with its large global share of glacier area and volume" (L202–203). I am not sure this holds, since Table 2 gives North America about a third of the global glacier area and volume. Conversely, Europe provides nearly half of the records (L198), mostly from the Alps (L192), from only 13 catchments (Table 2). Counting catchments hides this imbalance: a single intensively sampled catchment weighs much more in the medians and tests than one with a handful of samples. The authors already acknowledge this kind of bias at L286–294, and note the lack of data from northern Europe at L484–485. I would suggest carrying this into Sect. 2.3, redoing the comparison with the number of records per region, and naming the glacierized regions that have no data. Users would then know where the dataset is representative, and where results should not be extrapolated.
In Sect. 4.3, results are grouped by continent, while the discussion itself invokes different mechanisms: monsoon precipitation in Asia (L475), the influence of the tropical Pacific in the Andes (L448), westerly transport in Europe (L486). Each continent, however, covers several climate regimes. For Asia, for example, the monsoon and westerly domains described by Yao et al. (2013), already cited at L471, would be a natural basis for the grouping. It would also be useful to give the number of sites and sources behind each panel of Figs. 4 and 6. Several panels rest on a single site (Africa) or mainly on one study (North America >60°), and some boxplots are drawn from one value; these would be better shown as points and not interpreted.
5) Global comparisons between water types (Sects. 4.2 and 4.6)
Sections 4.2 and 4.6 explain differences between water types by processes (e.g. L379–385, L592–601), and recommend using in EMMA only the endmembers that show "a clear, significant separation" (L628–629). However, rare water types often come from one or a few sites (permafrost thaw from a single one), and precipitation was sampled over different intervals than the other components (L625–627). Pooled differences therefore mostly reflect where and when samples were taken, rather than processes. I would suggest presenting these global comparisons as descriptive, and basing any process interpretation on comparisons within the same catchment, for instance the difference between each water type and local precipitation or stream water, which removes most of the site effect.
The Dunn's tests (L589, Figs. S1 and S2) should be described in the Methods, including any preceding global test and the correction for multiple comparisons. As samples are grouped by study and site and are very numerous, almost any difference becomes significant; effect sizes or overlap measures that account for site would be more informative. Similarly, the Fig. 2 regression mixes regional gradients, so its intercept should not be read as a moisture-source signal (L348–350). Local meteoric water lines per catchment would be a useful addition to the dataset.
6) d-excess (Sect. 4.4) and agreement with the figures
Sect. 4.4 reads d-excess mostly as a moisture-source signal, while in the cryosphere post-depositional processes (sublimation, melt, refreezing) can play a major role, which the text acknowledges only briefly (L521–523). The text also states that continental recycling lowers d-excess (L556–562), whereas recycling tends to raise it, or at least not to lower it, and it is sub-cloud evaporation that lowers it. I think this section would benefit from a thorough revision, keeping in mind the cold-arid point raised above for lakes, for which lake type is not documented and the sample is small.
In several places the text and figures do not agree, which the authors will easily check. In Sect. 4.4, the ranking of snowpack, meltwater and streams (L517–526) and the description of North America >60° as elevated (L563) do not match Figs. 5 and 6. In Sect. 4.3, the medians given for Asia (L469–474), Europe (L485) and Africa (L458–459) differ from Fig. 4; for Asia, snowpack and streams appear inverted. L480 and L491 cite Fig. 5 instead of Fig. 4. Oceania is described as having a single endmember (L494), whereas the file contains three water types, and it is absent from Figs. 4 and 6. Finally, the abstract highlights groundwater and firn as relatively enriched, whereas Fig. 3 shows snowpack melt as the most enriched type, and firn rests on samples from a single study. Other inconsistencies of this kind may exist elsewhere in the manuscript, and I would encourage the authors to cross-check the text against the figures and tables throughout.
Technical corrections
Please check the typography and make sure that all references cited in the text are listed in the reference section; about fifteen to twenty seem to be missing, about ten others differ in author name or year, and several listed references are never cited.
Citation: https://doi.org/10.5194/essd-2026-173-RC2
AC1: 'Comment on essd-2026-173', Melanie Vital, 12 Oct 2026

Response to Reviewers 1 and 2
We sincerely thank both reviewers for their careful evaluation of our manuscript and their detailed and constructive comments. Their suggestions have helped us improve the clarity, methodological transparency, and scientific presentation of the database.
We have revised the manuscript accordingly, with particular attention to the database construction and harmonization procedures, statistical analysis, spatial coverage, quality control, uncertainty, and terminology. We have also clarified the scope and limitations of the dataset and moderated several interpretations to avoid overstating the representativeness or hydrological significance of the observed isotope distributions.
We would like to emphasize that this manuscript is primarily a database paper rather than a hypothesis-driven study. Its main novelty and scientific contribution lie in the compilation and harmonization of stable water isotope data from glacierized environments into a single, openly accessible database. By providing observed ranges of δ¹⁸O and δ²H across different hydrological components, the database offers a practical reference for researchers and field hydrologists. These ranges can help guide field sampling strategies, provide an initial understanding of expected isotopic variability, place new measurements in a broader context, and support the preliminary assessment of potential endmembers for applications such as End-Member Mixing Analysis (EMMA) and isotope hydrograph separation.
We recognize that isotope compositions are influenced by local hydroclimatic conditions, elevation, seasonality, and hydrological processes. The reported ranges are therefore intended as empirical references rather than universally transferable isotopic signatures. The statistical analyses are presented primarily to illustrate the variability and potential applications of the compiled data, rather than to establish universal distinctions among water sources.
In response to the reviewers' comments, we have also corrected identified errors in the database and published an updated version on Zenodo: https://doi.org/10.5281/zenodo.23184536.
Detailed point-by-point responses to both reviewers are provided in the attached supplementary document, together with explanations of the corresponding manuscript revisions.
We thank both reviewers again for their valuable contributions to improving the manuscript.

Citation: https://doi.org/10.5194/essd-2026-173-AC1

Supplement

https://doi.org/10.5194/essd-2026-173-supplement

Model code and software

Code for: Large-scale isotopic fingerprinting of cryosphere and hydrological components in glacierized catchments E. Jara et al. https://doi.org/10.5281/zenodo.18982251

Viewed

Total article views: 792 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
537	154	101	792	73	120	103

HTML: 537
PDF: 154
XML: 101
Total: 792
Supplement: 73
BibTeX: 120
EndNote: 103

Views and downloads (calculated since 12 May 2026)

Month	HTML	PDF	XML	Total
May 2026	230	67	8	305
Jun 2026	42	5	4	51
Jul 2026	51	10	7	68
Aug 2026	46	11	6	63
Sep 2026	135	41	61	237
Oct 2026	33	20	15	68

Cumulative views and downloads (calculated since 12 May 2026)

Month	HTML	PDF	XML	Total
May 2026	230	67	8	305
Jun 2026	42	5	4	51
Jul 2026	51	10	7	68
Aug 2026	46	11	6	63
Sep 2026	135	41	61	237
Oct 2026	33	20	15	68

Viewed (geographical distribution)

Total article views: 739 (including HTML, PDF, and XML) Thereof 739 with geography defined and 0 with unknown origin.

Country	#	Views	%

Cited

Latest update: 12 Oct 2026

Download

Preprint (1972 KB)
Metadata XML

Short summary

We created the first global collection of water measurements from glacier‑influenced rivers, lakes, snow, and ice to understand how melting glaciers shape water supplies. By combining data from many regions, we show clear patterns linked to climate and elevation. This work helps scientists and water managers better predict future water availability as glaciers continue to shrink.


Total:	0
HTML:	0
PDF:	0
XML:	0

Large-scale isotopic fingerprinting of cryosphere and hydrological components in glacierized catchments

Supplement

Model code and software

Viewed

Viewed (geographical distribution)

Cited

1 citations as recorded by crossref.