Dear editors,

dear authors,

Thank you for giving me the opportunity to review the manuscript essd-2022-395 "HMAGLOFDB v1.0 - a comprehensive and version controlled database of glacier lake outburst floods in high mountain Asia" by Shrestha and co-authors. In this study, Shrestha et al. compiled an inventory of glacier lake outburst floods (GLOFs) from sources such as scientific literature, media, and eyewitness accounts since the mid-19th century. Of the 682 documented GLOFs, 49 cases have not been previously recorded in other databases and appear for the first time in this compilation. The authors define dozens of variables that characterize the location, size, impact, and consequences of these GLOFs and attempt to gather all available information to populate these attributes with numerical or descriptive information. An important observation is that reporting on GLOFs has not been systematic in recent decades, leaving large gaps in the database that could be filled by further research.

Shrestha et al. report on the largest GLOF database to date in High Mountain Asia. The authors validate each case using geomorphological evidence downstream of the lake (but not necessarily changes in lake size) on satellite imagery, as well as assessment by local experts. In my opinion, this assessment is among the most carefully prepared inventories in this region, and I commend the authors for this work. The authors also provide a list of previously reported GLOFs that they excluded from this database, and the reasons for excluding these cases (e.g., incorrect coordinates or little evidence of former lakes) sound largely reasonable. I also appreciate the interactive map that allows non-experts to have easy access to this data (but without a download option).

Despite these efforts to compile this database, its presentation and associated manuscript require thorough revision, not least to comply with Earth System Science Data guidelines. I will first comment on major issues associated with the database, followed by issues in the manuscript.

1.) The database is currently archived on at least four different platforms (Zenodo, Github, ArcGIS and ICIMOD), and some files are available on one platform while missing on others (e.g. HMAGLOFDB_removed.csv or the list of references is not on Zenodo). I strongly encourage the authors to use ONE non-proprietary, accessible platform, i.e. without the need for a login account, to deposit their data in order to pursue their goal of presenting a version controlled database. This repository should contain all files contributing to this database (e.g. metadata, the list of references). My preferred choice would be Zenodo, but I leave that decision to the authors.

2.) The references column should contain the full name of the reference (author, title, journal, year) to make the underlying source easy to find.

3.) This database contains dozens of new cases that have not been reported before. I believe it is important to properly document these new cases, e.g., with satellite imagery before and after the GLOF showing lake area change or downstream impacts. I would like to encourage the authors to provide supplementary material for these cases.

4.) The value of this database is that it can be used to calculate trends in GLOF occurrence, hazard, and risk. To this end, each case should be provided with at least an approximate time stamp. If the exact date is not known, the authors could at least indicate that the GLOF occurred before the first available satellite image (Landsat or Corona) was acquired or during a period embraced by two satellite images. This information should be added for the entries in lines 474 to 660 in the current database. For these cases, it is still difficult to assess whether the occurrence of these cases could be more accurately dated, e.g., by using satellite imagery that cover the outbursts or not. In addition, it would be good to split the Year_approx column into two columns, one with the latest possible date (or year) before the GLOF and one with the first possible date (or year) after the GLOF.

5.) I have checked some GLOFs (yet not all systematically) and some cases seem to have limited evidence on satellite imagery, or the coordinates seem to be in wrong locations, e.g. at Langco, Tulaco, Phyang, Langbu Tsho or Bugyai. Oubuguoco has two entries, and I wonder if this is a duplicate? The GLOF from Pogeco in 2002  appears to be a misidentified GLOF, according to Nie et al. (2018). Looking at the removed cases, I wonder why so many GLOFs from Lake Merzbacher were discarded? Ng and Liu (2009) and Kingslake and Ng (2013) provide a thorough compilation of these cases, including flood volumes and peak discharges.

Regarding the manuscript, I have the following concerns and recommendations for improvement.

6.) The reason why the variables were selected could deserve a much stronger motivation/ justification. Why is it important to document river basin and lake volume? Why is it important to distinguish between female and male fatalities from GLOFs? In this regard, the metadata table should also be part of the main manuscript and adequately explain the meaning and units of each variable in separate columns.

7.) I feel that the compilation of lake data distracts from the real topic of this manuscript, which is the compilation of GLOFs. One motivation for the authors seems to be that they want to add a unique ID to each GLOF according to a previously compiled lake inventory (is my assumption correct?). However, the number of lakes may change over time, and the inventories have different minimum mapping units, different criteria for mapping lakes, or choose different distances from glaciers to consider a lake a glacial origin or influence. The methods used to map and categorize glacial lakes are subject to very different challenges and uncertainties than the creation of GLOFs, and are therefore part of a very different story. It is therefore surprising that the results begin with a presentation of lake area change, given that the title and scope of this manuscript refers to GLOFs. There is also little information on how these lake inventories were mapped, particularly the multi-period inventories (2000, 2010, and 2020). Were these inventories prepared by the authors or as part of a previous study? That being said, I strongly recommend removing the lake inventories from this study and focusing on the GLOFs.

8.) Removing the lake inventory from the study would also help avoid comparing lakes with and without outbursts (e.g., in Figures 5 and 6). ESSD is a data-driven journal and discourages scientific interpretation of the data. Instead, the discussion might focus more on the completeness (or gaps) of the variables. Could these be closed in future updates of the database? If so, how? Is there a need to collect additional variables? If so, which ones?

Specific comments (line by line):

L1: It is difficult to say why this is version controlled. Have there been previous versions of that database?

L22: A number of cases seem to have occurred before 1833 according to the csv file?

L27: It is common practice in ESSD that the abstract ends with the link and a reference to the repository. Please add.

L29: ‘expanse’: amount?

L31: ‘on a large scale’: please be more specific

L32: ‘Many of these lakes’: better use another phrase. According to your database, about 1% of all lakes had outbursts; ‘results’: please check the verb here.

L34: ‘many decades in different parts’: please be more specific.

L35: ‘HKH’: please explain.

L39: ‘n=…’, just say 2916 lakes? Did these two studies use exactly the same methods, i.e. same mapping area, same buffer around glaciers?

L45: ‘potentially dangerous lakes’: I feel that this phrase adds little objectivity to the discussion of potential changes in GLOF hazard or risk, and suggest to remove it.

L50: ‘strain dams’: if they are overtopped?

L51: ‘seismic events’: you mean earthquakes?

L56: ‘in the shadow’: you mean downstream?

L77-92: As described above, I discourage from putting the compilation of lake inventory into this manuscript. If the authors would like to keep it, much more information is needed on how the lakes were mapped.

L103: ‘overreporting’: maybe use another phrase such as ‘misidentified’ / ‘confused’?

L107: What about changes in lake areas and exposed lake beds? Is this not a criterion for identifying GLOFs?

L112: This file is only on Github, as far as I can see. Please choose one repository that contains all data.

L126-135: The content in these sentences is largely a repetition from the introduction. Please merge this information with the introduction and delete here.

L134: ‘Multiple datasets’: please cite.

L135: dh/dt: please explain.

L136: ‘in .git on a rolling basis’: please explain.

L137: ‘RDS / DOI’: please explain

L138-140: The concept of this ID is very confusion: What does GF, E, and N stand for? How do represent the precision of a given coordinate, i.e. how would you write a coordinate of 79.200239? Doesn’t Z need to have two digits, if a lake bursts out more than 9 times, as reported in Table 3?

L144: Please add the entire reference to the database, not only author and year.

L146-147: add the name of this file here

Table 1: There is hardly any information on why these variables are part of the database. Please add more reason why you selected these variables. Please copy the information from the meta data file here. How did you extract Elevation [m a.s.l.]? How did you obtain Area? Did you map the lakes? ‘Displaced_disabilities’: missing ‘l’; column ‘Certainty’ is not part of the database, as far as I can see.

L155-166: As discussed above, the presentation of a lake data base is questionable, and I would encourage the authors to remove it.

L172: ‘ICIMOD’: copy the link here and avoid foot notes.

L177: I couldn’t see the persons. Please add a circle.

L178: ‘Milad Dildar’: is this the photographer? Do you have permission to use the photos?

L183: ‘water line’: I couldn’t see it. Please add an arrow.

Figure 3: Please avoid red and green in figures at the same time, suggest using a viridis color scale: https://cran.r-project.org/web/packages/viridis/vignettes/intro-to-viridis.html; which study is the source of these 7 regions? Please add more space between the decades/ months. It is challenging to say when decade / month ends and the next begins. Would also be good to report the total sum per decade.

L190: I would welcome very much if the authors add a supplementary file that validates the occurrence of these newly reported cases.

L193: Is this statement referring to the removed cases?

L200: allowing ‘for’?

L207: ‘supersaturation’: unclear, please explain.

L209: Why not simply calling it a water pocket, similar to the study of Haeberli, 1983? ‘Glacier outburst’ sounds like if parts of the glacier get mobilized.

221: 22 ‘GLOFs’.

L224: ‘inventories’: which ones?

L226: ‘this data’: change to ‘these’ (data is plural)

L228: How likely is it that all the fatalities came from the GLOF itself? The GLOF was part of a large rainfall event with many debris flows (see also the image in Allen et al., 2016).

L254: ‘potentially’: why? Did these floods cross the borders or not?

L275: ‘hundreds’: suggest to add ‘to thousands of meters’?

L277: ‘SRTM’: please explain, and mention this part of the work also in the methods; ‘between 10 and 50m’: how do you know about these uncertainties?

L278: ‘errors’: of what?

L280: ‘satellite imagery close before the drainage date’: it’s still unclear whether (or not) you mapped the lake area before the GLOF in this study. Please be more specific. ‘further analysis’: which analysis?

L289: ‘wilful tampering with data for political reasons’: interesting thought, please elaborate.

L289: Please add more information on how you deem a source of information ‘generally trustworthy’.

L305: ‘susceptibility’: unclear how you estimated susceptibility. Please revise.

L305-306: Is there any hypothesis that says that lakes at higher elevations should be more prone to outburst?

L306: ‘less likely’: I guess this is not a probabilistic assessment, so please avoid; ‘a larger number of GLOFs happened [AT] low elevations’.

L320-321: interesting thought regarding the monitoring of debris flows. Is there any reference for that?

L326: could be backed up with more references.

L327: ‘increase of [reported] events’

L327-328: not exactly sure how the trends in the reported GLOFs and the research activity fit together. Please elaborate.

Paragraph 4.3: Suggest to avoid this discussion. ESSD is a data-driven journal, and too much science can be a reason for rejection.

Figure 6, second panel: shouldn’t lakes with outburst have negative lake area change?

L365: ‘risk’: you mean hazard (i.e. probability of failure)?

Figure 7: Honestly speaking, I cannot see the added value of this figure. ‘PZI/ RGI’: please explain.

Figure 8: Please elaborate how you obtained the Fahrböschung. What is the overall reason behind this analysis? I could not find any motivation in the introduction for this analysis. Why do you plot the data from Kääb et al. (2021) here? Please add axis labels to the inset, and describe in the figure caption.

L388: ‘mean reach angle’: all previous statements refer to the median?

L400: Do you consider the same study period of Carrivick and Tweed (2016)? Their study has been published 6 years ago, so the database is shorter.

L423: Did your appraisal account for road disruptions? What is a ‘ripple effect’?

L426: are thee minimum and maximum values behind the mean?

L447: It is not expected in ESSD that authors investigate these trends or provide deeper mechanistic insights. This phrase therefore can be deleted (also in L315 and L336).

L454-456: I could not follow this statement, please revise.

L478: The uncertainty and completeness of this variable (and many others) is not assessed or discussed in greater detail. Suggest to extend the associated paragraphs.

L496-499: Please add to the discussion how you would achieve that.

L500: The data availability statement usually comes before the conclusions.

L575: Please use the published version of that paper.

Dear editor and authors,

Please find attached my comments in detail.

Regards,