Referee report on

“Reconstruction of the 1989 Laguna del Cerro Largo GLOF: an Interdisciplinary Understanding of GLOF Impacts”

General comments

This manuscript presents a comprehensive, interdisciplinary reconstruction of the 1989 Glacial Lake Outburst Flood (GLOF) in Valle Soler, Chile. By combining 2D hydrodynamic modelling with local witness accounts and field validation, the authors revisit an event for which instrumental data are scarce and historical estimates remain uncertain.

The study is significant because it challenges historical estimates of this event using modern hydrodynamic modelling and provides new insights into GLOF processes. The discrepancy identified between the ~2,000 m³ s⁻¹ estimate reported in the original study and the authors’ modelled peak discharge highlights a likely underestimation of GLOF hazards, with implications for regional hazard assessment.

The manuscript is generally well structured and clearly written. The compilation of modelling results, sedimentary evidence, and historical information represents a promising dataset. However, the integration between the different lines of evidence remains somewhat descriptive, and several methodological aspects would benefit from clarification regarding the robustness of the reconstruction.

Overall, the study has clear potential but requires minor revisions to strengthen methodological clarity and interdisciplinary synthesis.

Specific major comments

1. Framework and independence of evidence

The manuscript adopts a transdisciplinary approach combining modelling, field evidence, and witness testimonies. While this integration is a clear strength of the study, the analytical framework linking these datasets could be described more explicitly.

It would be helpful to clarify how the different datasets are used within the reconstruction (e.g. model inputs, validation constraints, contextual evidence). For instance, witness testimonies appear to provide both input parameters for the hydraulic simulations (e.g. timing or duration of the flood) and supporting evidence for the plausibility of the modelling results. The manuscript would benefit from clarifying which elements were used to parameterise the models and which were retained as independent validation constraints, in order to avoid potential circular reasoning.

A short conceptual workflow figure summarising how the different datasets contribute to the reconstruction could help clarify this methodological framework.

2. Pre-event topography and modelling

The hydraulic simulations are based on the Copernicus 30 m DEM acquired several decades after the 1989 event. The manuscript acknowledges that post-flood erosion and deposition may have modified channel geometry and flow pathways, but the implications for the hydraulic reconstruction are not discussed in detail.

Given that fluvio-glacial plains are particularly prone to channel migration, sediment redistribution, and morphological reorganisation, the DEM used may differ substantially from the pre-event terrain that controlled flood routing. The authors may wish to consider whether an approximate reconstruction of the pre-event topography could be attempted; for example, by adjusting DEM contours using georeferenced historical aerial imagery where available (e.g. trimetrogon photography is available for some periods in this region).

Alternatively, a short discussion of how DEM uncertainty may influence model results would strengthen the interpretation of the simulations.

3. "Vanishing Evidence" and Proxy Limitations

The manuscript reports that dendrogeomorphic sampling was hindered by fungal decay affecting Nothofagus species, which limited the usefulness of tree-ring evidence for reconstructing the 1989 event. Such preservation issues are well known in humid Patagonian environments. To help the authors enhance their discussion and better integrate these results, I offer the following comments:

• Sampling locations: Were samples collected primarily within the active flood track, along floodplain margins, or within depositional zones? The spatial distribution of affected trees is important for distinguishing between GLOF impacts and smaller torrential floods. A map overlaying sampling sites with the modelled flood extent could help clarify whether the “vanishing evidence” reflects biological preservation issues or geomorphic patterns.
• Sampling strategy: I would be interested to learn more about the criteria used to select trees for sampling (e.g. targeting visibly disturbed individuals versus systematic sampling across the floodplain) and how the final sample size was determined.
• Please clarify whether the sampling strategy was independent of the model results. Confirming whether the botanical evidence served as an independent test of the model’s reach (rather than being guided by it) would strengthen the validation strategy.
• Even if dendrochronological dating proved difficult, physical markers such as scar height and orientation may still provide useful information on flood magnitude and debris transport. If such observations were recorded, it would be interesting to briefly discuss how these compare with the modelled water surface elevation during the 1989 event.

 4. Process interpretation versus modelling assumptions

The discussion suggests that the initial flood may have behaved as a dense, debris-flow-like slurry given the mobilisation of very large boulders. However, both modelling approaches are based on shallow-water, Newtonian flow assumptions. It could be interesting to briefly discuss how this interpretation of potentially debris-rich flow conditions relates to the modelling framework, and whether such processes could influence the reconstructed peak discharge or flow depths.

5. Contextualizing the revised peak discharge estimate

A key conclusion of the study is that peak discharge during the 1989 event may have been substantially higher than estimates initially reported in Hauser (1993). This is an important result, and the discussion could be strengthened by placing these revised estimates in a broader regional context.

For example, it may be useful to briefly compare the reconstructed discharge with other documented GLOF events in Patagonia or to clarify how such revised estimates might influence current interpretations of regional hazard.

Minor/Technical suggestions

• Consider adding a detailed map showing the locations of dendrogeomorphic samples and sediment cores relative to the modelled flood extent.
• Data Summary: A summary table correlating each dataset (witness, sediment, tree-rings) with the specific flood characteristic it constrains (magnitude, timing, extent) would greatly improve readability.
• Italicise species names.

This manuscript investigates the 1989 GLOF that occurred on the eastern side of the NPI, in Valle Soler, Chilean Patagonia. It does so by combining multiple diverse datasets and a model approach. The datasets include sediment cores (Russian peat corer), dendrological evidence, two radiocarbon dates, UAV-based DEM, and witness accounts which are used as input parameters for the two different hydrological models. The authors present new convincing findings regarding the flood dynamics and the triggering mechanism of the GLOF compared to what was previously reported about this particular GLOF by Hauser (1993).

In general, this work is a substantial contribution to the scientific literature on GLOFs in Patagonia and their physical and human impact. It was a pleasure to read this manuscript (though I do have a comment on the writing style - see below), and I believe that with moderate revision, this article will be ready for publication soon.

I hope these comments help improve the manuscript.

With kind regards,

Loic Piret

General comments:

(1)  Datasets not integrated in the discussion: The authors have created and compiled an extensive and diverse dataset in this data-sparse region. The discussion section focuses on the findings based on the model output and the witness interviews. However, not all results are integrated in the discussion:

Firstly, the UAV-derived DEM is announced in the abstract and introduction section and its creation is detailed in the methodology section. The UAV-derived output however does not seem to further appear in the manuscript. Please include these results (more explicitly) in the manuscript and how they helped understand the 1989 GLOF dynamics.

Secondly, the sediment cores are presented in the results but receive very little attention in the discussion. I'm left with the question of what the sediment analysis actually showed. The two clastic layers found in the cores are interpreted to represent the two GLOF's in the valley which occurred in the valley. But what can we tell about the sediment grain-size gradations, the core-to-core correlations and variability in thickness of the deposits between the sites? Is it not possible to get information about the flood energy, duration, and spatial extent of the flood using these sediment cores? This data can then be linked back to the hydrologic model outputs, which can then help validate (or challenge) the model outputs.

Thirdly, the abandonment of the dendrogeomorphologic dataset is disclosed honestly and it is important we mention such "failed" methods. However, at L325-340, some other (non-abandoned) findings are mentioned about where trunks were found but it is unclear what actually can be deduced from these findings. Finally, the Carbon-14 dating needs some revision and integration in the discussion as well (see also specific comments).

In short: The manuscript would benefit from a new more explicitly data-driven discussion section.

(2) Writing style: The authors occasionally adopt an Americanized journalistic (and at times, anecdotal) tone rather than a rigorous, formal analytical tone which is considered more suited for a peer-reviewed scientific article. Though I enjoyed reading it, I'm unsure whether this writing style is 100% fitting for this format. Ultimately, I leave the decision on which writing styles are appropriate for this journal to the discretion of the editorial team. Some specific examples of the journalistic/anecdotal approach are noted by line number here: L78-80, L168-169, L255-258 (this is a particularly dramatic anecdotal passage), L413-414 (anecdotal).

Specific comments:

L80: The term "epistemological pluralism" (and, later in the manuscript at L421: "Epistemological Blind Spots") is quite heavy. I would suggest to either clarify what "epistemological pluralism" means or to refrain from using it. While this concept is valuable, the specialized terminology may form a barrier for non-native English-speaking scientists. I would encourage the use of slightly more accessible language.

L233-237: Are these findings on the grain size based on your sediment core results or were they previously published in either Burton et al 2020 or Sincavage et al 2023 (this ref is not in reference list)? In case of the former: add them to the result section of the sediment cores. In case of the latter: it would be better to place this in the discussion section and compare with your own grain-size results from the sediment cores.

Section 3.3. (L284): The single C14 date shows that this top clastic layer is from the 1989 GLOF. However, can we be sure that this clastic layer is all deposited at once? Maybe there is sedimentological evidence pointing to a single-event deposit  (downcore grain-size patterns for instance). It would make sense that this is a single deposit based on the context and witness accounts but independent evidence is preferred.

L289-294: Does the grain size only vary with distance downstream or is it also dependent on the specific geomorphic setting (e.g. low-lying, raised, near constriction, etc.)?

L347-349: "a counterintuitive coarsening of the grain size, before a gradual fining": It is unclear what is meant here. Is this a downstream trend in the valley? Is this a downcore grain-size trend? The Sincavage reference is not in the reference list.

At L354, it is mentioned that organic material was extracted from the clastic layer that is assumed to represent deposits of the 1989 GLOF. One date confirms this but the other result dates the sediment back to 1899 CE. This is confusing. Please show where the samples were taken on the sediment logs (figure 6) to clarify this.

L373-374: Please provide some appropriate examples of this triggering mechanism in similar situations to strengthen this idea. It is currently unclear where this idea came from. Furthermore, are there an other triggering mechanisms that should be considered?

The results regarding policy impacts appear overstated. While I acknowledge the limited discussion concerning policy-making in L72-73 and L416–420, the claims in these sections and at lines 460–461 require some more substantiation. The authors could provide references and insights into the existing relevant policy and legal frameworks to demonstrate how the 'original report' actually shaped the political landscape.

Tangential to this comment: If no such policy framework exists, or if it is outdated, a policy-brief should be written to inform the local policymakers!

Fig 4  caption - L267: Could this be the other way around? The GeoClaw model shows the wave progression almost at a confluence at t=0.25, whereas the HEC-RAS 2D model shows a more limited wave progression at the same time. The wave in the GeoClaw model already reaches the rightmost edge of the study-area rectangle at t=1.5 hours, whereas in the HEC-RAS model, it only reaches the rightmost edge of the study-area rectangle at t=2.5 hours. Therefore, the HEC-RAS model shows a lag in wavefront propagation relative to the GeoClaw model.



Fig 6: Adding a dedicated depth scale instead of just the tape measure would improve the readability of this figure.

Fig 9: Add the F14C value of the sample to the graph to show how you deduct an age of 1989 from this result. Also mention that technically, it could be from the 1950's but given the context, it is highly likely from 1989.

Technical corrections:

L49: I suggest to use the gender-neutral alternative "uncrewed" instead of unmanned. https://cordis.europa.eu/article/id/442165-not-unmanned-but-uncrewed

L80: Nightingale 2016 missing from reference list.

L237: Sincavage et al., 2023 missing from the reference list. Please verify that all references in the text are in the reference list and vice versa.

L256: Please explain what a coigue (i.e. Nothofagus dombeyi (?) ) is.

L289: Please explain Spanish words such as "campensino".

L355-356: "The 14C samples taken from the layer expected to have originated returned "modern age", ..." Currently, this sentence is grammatically incomplete. Is there a part missing from this sentence - particularly after "originated"? Also, should it be "fractional carbon of 1.1921; or "at" (as it is now).

L371: Add reference to the original report here.