Lee et al. present and discuss a new database of d18O and d2H in precipitation in Jeju Island (Korea). Apart from the usefulness of such data for scientific and practical purposes (e.g., water resources management) it is worth praising the authors for “digging up” decades all data, rather that just leave sit in drawers – it could set an example for groups, as well. To this extent, I suggest adding 1-2 lines, both in the abstract and the main text on the usefulness of such approach. Also, to better support this, I would suggest the authors to add a brief (within the limitations imposed by the journal’s instructions) discussion of how their data fits/complements similar studies in the region (see https://doi.org/10.1016/j.ejrh.2026.103141 and references therein).

Overall, given the scope of the journal, I would reduce the length of the discussions section, perhaps remove also the entire sections 4.3 and 4.4 and leave these for a dedicated paper. Some of the discussion could be reduced in length, as they are mostly theoretical considerations.

A couple of observations detailed below might nevertheless help the authors and future readers better shape and understand, respectively, the paper.

Specific comments

1. The introduction seems perhaps to long, there is really no need to underpin the uniqueness of volcanic islands when it comes to isotopes in precipitation. The lengthy introduction sets some expectations that this characteristic might somehow influence the stable isotopic composition of precipitation, which is not the case. I leave it to the authors to decide whether this comment warrants a change in the text.
2. “isotopic” jargon. In places, the text follows loose oral communications “rules” leading to, say, inconsistencies, see for example in lines 44-45 “complex atmospheric and hydrological conditions produce significant spatial variability in isotopic fractionation via Rayleigh distillation” or lines-76 “radial topography shapes not only precipitation distribution and orographic lifting processes but also atmospheric cooling trajectories that govern Rayleigh distillation and isotope fractionation”. Other examples include “isotopic depletion/enrichment” (change to “depletion/enrichment in 18O/2H”), “heavy isotopes are typically enriched” (they are not), “isotopically depleated” (the water is not depleted in isotopes) etc. Combing the entire ms for such loose formulation would greatly improve its accuracy in terms of scientific language.
3. The “methods” should better present the screening of data. It is not clear (line 152) how “Data screening criteria included excessive exposure time, leakage, and incomplete rainfall events.” Was used to remove outliers, “bad” data points etc.
4. The “results” section, includes numerous lines that fall under “discussions”, see for example the 1^st paragraph (lines 158-171). Either shorten the entire section to provide results only, or rename it “results and discussions”
5. Discussions:
   1. It is not clear in line 225 what “seasonal offsets in δ18O, δD, and d-excess” are. The entire section 4.1 si more like a theoretical background than real discussion. Further, the text discusses the usefulness of sinusoidal curves used, but no discussion is made on why the d18O and d2H values behave as they do. Also, the discussion of d-excess variability is to simplistic – it is asserted that high values indicate continental sources and low values, maritime. This is not always the case, especially e.g., during warm/dry summers, when sub-cloud evaporation can lead to very low d-excess, regardless of air mass origin. I suggest either support the source partitioning with moisture track analysis (e.g., Lagrangian models) or just leave it out. The journal’s paper format favors the later. The last paragraph of section 4.1 just comes out of nowhere, as no discussion of large-scale atmospheric circulation patterns was presented before.
   2. L258 – valid for every area with high altitude differences, not only on volcanic island. As per the earlier comment, continuous references to the volcanic nature of the island suggest that it could have some effect on the stable isotopic composition of precipitation. It does not.
   3. As for section 4.1, the 1^st paragraph in section 4.2 is theoretical and too long. The ms could do without it. The discussion for the different lapse rates needs some more depth. The proximity to the ocean does not explain lower condensation efficiency; the explanation must lie in a different place. Perhaps the strength of the circulation could do it? This would be a good place to insert a discussion on the role of atmospheric circulation patterns. The last paragraph of this section (4.2) is again too long and mostly theoretical. And again, it is unclear what it is being mean by “isotopic fractionation”. Is it a change in fractionation factor? Rate of change? Magnitude? It needs to be explained.
   4. Section 4.3. What is the spatial resolution of the grid cells used in the GCM? Was it only 3 cells for the entire island? The putative limitation in the representation of mesoscale convective systems would also affect summer precipitation, and I would somehow expect a stronger role of these during summer, not winter. Please be more specific about this explanation.

General comment

Lee et al. present monthly isotope precipitation data collected from 15 monitoring stations across the Jeju volcanic island in South Korea. The dataset spans the period from September 2000 to December 2003, allowing the analysis of seasonal variability. Considering the topography of the island, this isotope dataset is also useful to study the mechanisms behind the altitude effect.

First, it's a good thing that this data, collected more than 20 years ago, is being published rather than left to gather dust in a drawer. The paper fits the scope of the journal, even if some improvements in the presentation of the data and its analyses are necessary before acceptance for publication in ESSD. Especially, there are many repetitions about how useful the data could be, missing information about the meteorological data and IsoGSM outputs, and sections that should be re-written in the discussion section. Also, there are several statements about local climate conditions without citing any reference or showing any maps.

Specific comments

• The same statements about the how useful the dataset could be are repeated all along the paper. For example, lines 62-24 and 66-68 are very similar, again at lines 84-85, 198-201, 218-221, 250-253, and so on. This is way too repetitive and, contrary to the authors’ intention, it makes loose impact about the value of this data.
• At lines 66-68, I do not agree so much with the authors. First, I am not very convinced on how this data could be useful for paleoclimate reconstruction. The sampling period is rather recent, and the time length is rather short (39 months), meaning it is not possible to investigate important climate changes related to interannual variability for example. For the modeling side, of course this is useful, but an isotope-enabled GCM (so global scale), even at high resolution such as 50 km like IsoGSM, cannot resolve the effects due to the topography of the island (maybe there is even no land mask for this island in the simulation setup). So, I would tone down a bit this statement. One aspect, which could be interesting, would be to use a regional isotope-enabled model such as IsoRSM with moisture tracking ability.
• Related to the point above, “long-term” sounds a bit odd to me considering the dataset is “only” 39-month long.I suggest to replace long-term by a more neutral and descriptive term such as pluriannual for example.
• Data in the excel file: Add a README tab with the description of the site, coordinates and altitude information of each monitoring site. This very important information is missing in the excel file. Moreover, I cannot understand the tab “Sampling Points”, please write in English and provide the right information.
• Data availability: There is no link and/or information about the standard climatic data. They need to be provided as well. IsoGSM data need to be cited as well. Considering the acknowledgement “The authors thank Kei Yoshimura and Hayoung Bong for sharing their model data.”, I suppose the data are from IsoGSM4-HR at T248 spatial resolution (around 50 km), nudged to ERA5, right? Please add the citation and DOI of the Zenodo dataset in the data Availability section (https://zenodo.org/records/15073235, https://zenodo.org/records/14681370, and https://zenodo.org/records/14676091 I suppose).
• About IsoGSM, There is no description of the model setup in the paper. Please add it: model version (IsoGSM4), spatial resolution, nudging to ERA5, citations, etc…). You talk about 3 grid cells (line 309), please provide the coordinates.
• Lines 86-91: please provide information about the period, reference, and show some plots.
• Line 147: define also the d-excess.
• Section 3.1: It would be nice to plot also the individual records, and not only the average.
• Lines 179-185: again, it would be nice to show map of typical climate conditions in summer and winter (temperature, precipitation, winds or water vapor transport, humidity, etc…). The authors can use local meteorological station data or ERA5.
• Section 3.2: The authors refer to the figure 4, but this section is more related to the Figure 6 about the altitude effect.
• Lines 211-214: again, show it. You can use ERA5 data to check the moisture circulation.
• Section 4: “Discussion” sounds odd for a data paper. I would call it more “Technical validation”, in order to show for which purpose this data can be used.
• Line 224: Remove the “long-term” term, which is misleading, all along the manuscript.
• Line 227: We do not use isotope data to calibrate isotope-enabled GCMs. The data is used to evaluate models and check the missing processes.
• Line 232-233: remove this sentence.
• The section 4.2 is kind of repetition of the results section. Some parts could be shortened a lot.
• Section 4.3: Add the description of the model here or, better, add a sub-section IsoGSM in the section 2.
• Line 315-316: Show it!
• Line 321: The model is not coarse (~50 km) considering this a global model. Such very high resolution for a GCM is quite rare (usually, this is 200 and 100 km). Instead, I would say to go further, we need a regional model or an icosahedral model (NICAM-wiso) with the ability to resolve convection.
• Lines 327-328: show it.

• Lines 334-335: No this is not the problem. The problem is that a GCM cannot resolve such convective processes and parameterizations will never fix completely the issue. You need a higher-resolution model where the convection is modeled (and not parameterized), such as the icosahedral model NICAM.
• The section 4.4 is not convincing at all and should be removed.