the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 25 Feb 2025
Assessing economic impacts of future GLOFs in Nepal’s Everest region under different SSP scenarios using three-dimensional simulations
Abstract. This study investigates simulated glacial lake outburst floods (GLOFs) at five glacial lakes in the Everest region of Nepal using the three-dimensional model OpenFOAM. It presents the evolution of GLOF characteristics in the 21st century considering different moraine breach scenarios and two Shared Socioeconomic Pathways scenarios. The results demonstrate that in low-magnitude scenarios, the five lakes generate GLOFs that inundate between 0.35 and 2.23 km2 of agricultural land with an average water depth of 0.9 to 3.58 meters. These GLOFs reach distances of 59 to 84 km, affect 30 to 88 km of roads, and inundate 183 to 1,699 buildings with 1.2 to 4.9 m of water. In higher scenarios, GLOFs can extend over 100 km and also affect larger settlements in the foothills. Between 80 and 100 km of roads, between 735 and 1,989 houses and between 0.85 and 3.52 km2 of agricultural land could be inundated, with average water depths of up to 10 meters. The high precision of the 3D flood modeling, with detailed simulations of turbulence and viscosity, provides valuable insights into 21st-century GLOF evolution, supporting more accurate risk assessments and effective adaptation strategies.
- Preprint
(2471 KB) - Metadata XML
-
Supplement
(268 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2025-50', Adam Emmer, 21 Mar 2025
This study models potential future GLOFs from three existing and two future glacial lakes in Nepal. This study brings novelty while integrating OpenFOAM modelling with SSP scenarios. The study details the impacts of potential GLOF scenarios on people and infrastructure while it fails to justify some of the basic assumptions.
Strikingly, the breach scenarios (Table 1) are defined regardless moraine dam geometry, physical limits of breach development, internal structure and possibly overdeepened bedrock terrain. Why 30, 60 and 90 m? Why not 20, 40 or 60 m? Or 10, 20 and 30 m? It is important to highlight that anyhow sophisticated modelling outcomes are totally dependent on rather arbitrary definition of these breach scenarios. It is also important to highlight that these scenarios have different probabilities for individual studied lakes and that some are not even realistic.
Now what is called BR1 (lower boundary; 30 m breach depth) is already pretty harsh scenario and the term “lower boundary” is misleading in this context. How many examples of 30 m deep breaches do we have from lakes of similar size and topographic setting? I don't think about many. The BR2 (upper boundary; 60 m breach depth) is not only unlikely but also unrealistic for lakes with flat and wide dam geometry (such as Imja which dam height is 55 m, according to 10.5194/hess-29-733-2025 or 35 m according to 10.5194/hess-19-1401-2015).
For a comparison, the breach which developed during the 2023 South Lhonak GLOF – the largest GLOF from a moraine-dammed lake in High Mountain Asia in past decades – is 55 m deep (see 10.1126/science.ads2659). The use of as extreme scenario as BR3 (90 m breach depth) needs special justification on a case-by-case basis.
In conclusion, in the study of 5 lakes, breach scenarios should rather be tailored to specific dam properties of individual studied lakes and I encourage the authors to address this issue (I recommend major revisions). Thank you.
Citation: https://doi.org/10.5194/egusphere-2025-50-RC1 -
AC1: 'Reply on RC1', Wilhelm Furian, 27 Mar 2025
Dear Mr Emmer,
We sincerely thank you for recognizing the potential and novelty of our study. You are absolutely right that the morphological breach parameters have not been sufficiently explained, and that this lack of detail reduces the clarity of our approach. We will address your comments by incorporating the following clarifications into our manuscript, which we hope will improve the transparency of our methodology and clarify our assumptions and choices.
Regarding your second paragraph:
You are absolutely right in pointing out that our rationale for choosing these exact breach parameters was not sufficiently explained. We address this in the following paragraphs:
Due to the requirements of working with OpenFOAM, several concessions have to be made when creating breach scenarios. As we are working with this 3D numerical model, several options that can be included in 2D approaches are not available to us. We cannot define a desired hydrograph for each lake or couple a designated breach model to the simulation. Rather, the hydrograph in OpenFOAM results from the breach scenario and the lake volume and is not specified directly.
Therefore, in our approach, the DEM for each simulation run, including the moraine breach, has to be manually created and transformed into an STL surface. As we ran almost 100 simulations, it would not be feasible to define individual moraine breaches for every lake in every scenario, taking into account, e.g., the changing moraine structure, the growing lake volume and depth etc.
We fully agree that the magnitude of the simulated floods mostly depends on the chosen breach scenarios and mention this in Table 4. However, we recognize the need to further explain the reasoning behind our choices and argue for their validity:
Since we chose to use a 3D model to improve the simulation accuracy, we had to limit the number of breach scenarios in order not to increase the computational time beyond the 10-12 weeks that the simulations already had to run (excluding the time needed to set-up the simulations).
The chosen breach scenarios follow 10.1016/j.jhydrol.2021.126208 (Sattar et al. 2021), who provide numerous GLOF scenarios for Lower Barun Lake. We excluded the lower and higher estimates of their study, as the higher ones seem unrealistic and the impact of the lower scenarios would be too small to justify the longer computation time. We therefore chose two main depths: 30 and 60 m, and a third extreme depth of 90 m.The 30m breach was chosen because it represents a significant, but not extreme, moraine breach. The 60m breach was chosen as the upper limit of GLOF events (as seen at South Lhonak Lake, 10.1126/science.ads2659). The 90m scenario was chosen as a potential GLOF of extreme magnitude, theoretically possible at all lakes except Imja Tsho. However, due to its unlikely nature, we did not evaluate this scenario further. Instead, we provide the results in the supplementary material.
As a degree of generalization is required for large-scale OpenFOAM simulations, we use these breach parameters for all five lakes. Apart from the fact that it would not be feasible to individually create different moraine breaches for our 3D approach, many of the parameters needed to delineate individual breaches are not easily quantifiable for the future scenarios: Internal moraine ice can melt, and moraines can be damaged by earthquakes or lowered by internal piping, etc.
However, this is not made clear in our manuscript and we will add a subchapter detailing the reasoning behind the choice of these breach parameters and justifying their use for each lake.We do not aim to investigate the probabilities of specific events, as it would be beyond the scope of our study to approximate the necessary parameters for the whole of the 21st century. However, this is not made sufficiently clear in our manuscript and we thank you for pointing this out. We will include a clearer description of the objectives of our study and a more substantial justification for our choice of breach parameters.
Regarding your third paragraph:
We agree that the phrase "lower boundary" is misleading for a 30m breach, as it may well be the lower boundary in this study, but not in the wider scientific context. We will rephrase the relevant paragraphs.
You are absolutely right, a 60m breach would be very unlikely at Imja Tsho, as it would represent more than a complete moraine incision. We used an approach similar to the recently published 10.5194/hess-29-733-2025, where “the maximum breach depth is considered to reach the marine dam’s maximum height and extend from the dam crest down to the point where the hummocky terrain ends” (p. 737). However, we have failed to explain this in detail and will therefore adapt our manuscript to exclude the 60m scenario for this lake.
According to 10.5194/hess-29-733-2025, the results of the 30m breach are within realistic boundaries at Imja Tsho. Our study found ~11,800 m3 s−1 peak discharge for a 30m breach, while Chen et al. (2025) estimated a peak discharge of between 8,000 and 30,000 m3 s−1 for a full breach, with a mean of ~15,000 m3 s−1.
For Tsho Rolpa, both 30m and 60m breaches are possible and our results (regarding discharge and inundation depth) are in good agreement with previous studies:
10.13101/ijece.5.123 found 90,000 m3 s−1 peak discharge in the highest scenario vs. 81,000 m3 s−1 peak discharge in our highest scenario
10.5194/hess-29-733-2025 give 13,000 to 15,000 m3 s−1 as the peak discharge vs. the ~12,300 m3 s−1 peak discharge in our study
10.5194/piahs-387-59-2024 used 20m and 40m breach depth, and for both inundation depth and discharge our results agree well with theirs.We will adapt the relevant parts of the manuscript and include a subchapter comparing the performance of our model, thus justifying the use of the 30m and 60m breach for four lakes and the use of 30m at Imja Tsho.
Regarding your fourth paragraph:
We have already mentioned the unrealistic nature of the 90m breach in the manuscript, but we will rephrase the sentences to emphasise that we are excluding it from our analysis for this reason.
We thank you again for your valuable comments, which help to improve the quality of our manuscript. We hope that our clarifications and adjustments address your concerns and look forward to your response.
Best regards -
Wilhelm FurianCitation: https://doi.org/10.5194/egusphere-2025-50-AC1
-
AC1: 'Reply on RC1', Wilhelm Furian, 27 Mar 2025
-
CC1: 'Comment on egusphere-2025-50', Nitesh Khadka, 28 Mar 2025
The paper provides new insights into GLOF modeling and the resulting socio-economic consequences in the Nepal Himalaya. I have a few suggestions:
- How does the used model depict accurate GLOF rheology (how sediments are incorporated in the flow)?
- Can we expect inundation of 20 m only for large lakes such Ngojumba Tsho?
- Trekking trails are shown as roads, such as in Phakding (Figure 8)? Will it not have impact when carrying out the socio-economic (monetary) assessment?
Citation: https://doi.org/10.5194/egusphere-2025-50-CC1 -
AC2: 'Reply on CC1', Wilhelm Furian, 10 Apr 2025
Dear Mr. Khadka,
Thank you for recognizing the novelty of our study and for taking the time to suggest improvements and clarifications! We address them in the following paragraphs:
1.
While we do not incorporate a specific sediment transport model, we account for the increased viscosity and simulate the GLOFs as hyperconcentrated flows, following 10.1029/WR021i010p01511 and 10.1007/s10346-023-02157-wUsing a dedicated two-phase solver for sediment transport (like sedFoam, 10.5194/gmd-10-4367-2017) would allow for the simulation of the sediment and the water. However, this would only be possible if the vertical and horizontal extent of the flood would be known in advance. While, theoretically, it would be a possibility to first run interFoam, create a mesh from the inundation depth and then run sedFoam to simulate the sediment transport in that water body, it would by far exceed the scope of our study.2.
You are right in that larger lakes have the potential to produce GLOFs capable of causing massive inundation depths. However, we do not use a physical model to simulate the moraine incision, but rather employ a parametric scenario-based approach and provide the breach parameters to OpenFOAM to simulate the according hydrograph.Therefore, in our study, the lake volume represents the potential for a GLOF's magnitude, while the breach size is responsible for realizing this potential.Since our breaches are of the same magnitude for all lakes, the inundation depths of all lakes are of the same order of magnitude. Please also see our response to Adam Emmer's valuable comment above, where we supply previously lacking information on our reasoning behind the breach parameter selection.3.You are absolutely right, the classification of every trail and trek as "roads" is misleading and will be changed in the relevant figures and paragraphs.Regarding the impact on the socio-economic assessment: The lack of detailed classification in the OSM-roads data prevents an in-depth assessment of the financial impact of the destruction of infrastructure. Therefore, we unfortunately had to refrain from providing monetary values associated with the inundated routes. Instead, we point out the aggravating nature of road/trek destruction during a GLOF event as it hinders post-disaster assistance.Citation: https://doi.org/10.5194/egusphere-2025-50-AC2
-
CC2: 'Comment on egusphere-2025-50', Binod Diwadi, 23 Apr 2025
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2025/egusphere-2025-50/egusphere-2025-50-CC2-supplement.pdf
-
AC3: 'Reply on CC2', Wilhelm Furian, 26 Apr 2025
Dear Mr Diwadi,
Thank you very much for taking the time to review our manuscript, for recognizing its potential, and for providing valuable and constructive feedback. In the following, we address your comments point by point.
Main points:
- We appreciate your suggestions regarding the description of the Everest Region. We will incorporate the studies you mentioned to improve the representation of previous scientific efforts.
- A) We chose not to include additional equations from OpenFOAM in the manuscript, as they are both highly numerous and not directly specific to our study. However, to support interested readers, we will add a reference to the OpenFOAM user manual, where the relevant equations and their interrelations are comprehensively described.
B) The dam break flow modeling approach is detailed in Chapter 4.2 and in Table 1 of the manuscript. There, we describe the relation between different breach scenarios, their corresponding opening times, and the use of the baffle method to simulate progressive breach formation.
C) We fully agree with your observation that simulating GLOFs as hyperconcentrated flows leads to improved realism compared to traditional clear-water models. We also recognize the limitations inherent to our approach and have discussed some of these in Chapter 6.1. Following the suggestions of previous reviewers as well as your own, we will expand this discussion further to better highlight the limitations and uncertainties.
Regarding your concern: The computational demands of two-phase GLOF simulations are already considerable. While three-phase solvers exist within OpenFOAM, they are primarily designed for small-scale industrial processes, such as multiphase flows in oil extraction or nuclear reactors. Using a three-phase solver to model GLOFs at a large spatial scale would not have been feasible with the computational resources available to us. Tracking three interfaces (air, water, sediment/boulders) would significantly increase computational complexity, particularly in terms of interface tracking within the VOF method.
D) For your question regarding inundation depths, we refer you to our response to Mr. Khadka.
E) Regarding surface roughness, we employed a constant Manning value across the domain. Given the scarcity of high-resolution surface roughness data—especially for projecting into future conditions—this approach follows common practice in GLOF modeling studies. While Manning’s n may vary between mountainous and more lowland river reaches, accurately delineating these transitions would introduce considerable uncertainties. Additionally, implementing spatially variable roughness would require splitting the computational mesh into multiple patches or implementing customized boundary conditions, both of which would significantly increase computational time. We will include a discussion on the limitations associated with assuming a constant Manning value in the revised manuscript. - A) You are correct regarding the misleading legend entry for trails in Phakding. This concern has already been addressed in our response to the previous reviewer.
B) We appreciate your comments regarding the potential impact of GLOFs on the tourism sector. We will add a paragraph discussing this important aspect in the revised manuscript. - We agree that the formation of glacial lakes depends on a wide range of factors that cannot be predicted with certainty. We will include a brief discussion to acknowledge this complexity in the revised manuscript, while a comprehensive discussion of all relevant processes is beyond the scope of our study (as these topics are addressed more fully in previous specialized research, e.g., https://doi.org/10.18452/23250 and https://doi.org/10.3389/feart.2022.821798).
Minor points:
We appreciate your detailed suggestions and will implement your advice regarding additional references, improvements in grammar and style, as well as clarifications where necessary. We will also include explicit references to the reflectance and OSM datasets and revise Figure 3 to improve its clarity. Below, we address your specific comments in detail:
- The dark blue areas on the glaciers indicate existing supraglacial lakes. The label “Mt Everest” was intended to show the general location of the mountain, rather than the exact position of the summit. However, we agree that its placement could be improved and will adjust it slightly.
- For the topographical potential, we refer to our previous publications (https://doi.org/10.18452/23250 and https://doi.org/10.3389/feart.2022.821798), where the surrounding slopes of potential glacial lakes and the lakes’ evolution are discussed in detail. In this context, Ngojumba Tsho shows a slightly higher predisposition for mass movements than Bhote Tsho. However, as our study does not model specific GLOF triggers, we cannot provide estimates on potential avalanche volumes. We agree that this would be an interesting subject for future research.
- The depth-damage relationship is described at the end of Chapter 4.1, where we explain the different damage classes and provide references for further reading. To maintain the focus of the manuscript, we prefer not to reproduce the full damage-depth curves.
- The empirical coefficients in Equation 2 originate from O’Brien et al. (1988) and were derived through regression analysis. They define the relationship between yield stress, viscosity, and the volumetric sediment concentration. For a detailed discussion, we refer readers to the original publication.
- Chapter 5.1 is titled “Overview” because it summarizes the results of the GLOF simulations. Given the large number of model runs, we believe an introductory chapter providing a concise overview before discussing individual aspects in detail is necessary. We therefore consider the title appropriate.
- The term “cross-section” refers to a vertical slice through the computational mesh along the GLOF path, perpendicular to the primary flow direction. At these cross-sections, we assess key parameters such as valley discharge, inundation depth, etc.
- Thank you for the information regarding the status of the EWS at Imja Tsho. We will revise our manuscript to note that the EWS has been discontinued.
- Regarding your comment on “modifying the extent (edges)” of the final three figures: we would like to clarify that the figures present the unmodified simulation results after postprocessing and integration into ArcGIS. No manual adjustments to the flood extent were made.
Citation: https://doi.org/10.5194/egusphere-2025-50-AC3
-
AC3: 'Reply on CC2', Wilhelm Furian, 26 Apr 2025
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 274 | 71 | 13 | 358 | 41 | 8 | 22 |
- HTML: 274
- PDF: 71
- XML: 13
- Total: 358
- Supplement: 41
- BibTeX: 8
- EndNote: 22
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1