the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 11 Dec 2025
New insights from fully-integrated surface-subsurface hydrological modeling in high-elevation glaciated environments
Abstract. Recent modeling efforts go towards distributed physics-based modeling for improved hydrological process understanding. The sparse hydro-climatological observations and complex topography in high-elevation mountainous environments have, however, hampered the application of such data-intensive and sophisticated techniques for detailed process exploration. In particular, characterizing surface-subsurface water exchange processes in these terrains remains understudied. Here we implement a fully-integrated and fully-distributed surface-subsurface hydrological model, WaSiM, to quantify the observed streamflow variations and their interactions with groundwater in the high-elevation glaciated environment (Martell Valley) in the central European Alps since the 2000s. Extensive field observations (meteorology, vegetation, glacier mass balance, soil properties, river discharge), especially in-situ groundwater levels, were collected to investigate hydrological processes and constrain model parameters in the surface and subsurface. We observe that in contrast to the delayed responses typically observed in aquifers, shallow alpine groundwater responds nearly as quickly as streamflow to peak snowmelt and extreme rainfall events, highlighting the need for improved subsurface parametrization in hydrological modeling. Surprisingly, we are able to obtain satisfactory model performance when subsurface lateral flow is constrained to zero, indicating its limited influence in river discharge generation at the site and providing new insights into hydrological processes in such an environment. Moreover, to overcome the challenges of integrating point-scale groundwater observations into a fully-distributed hydrological model in high-elevation catchments, we suggest pre-calculating the Topographic Wetness Index to guide piezometer placement. Despite the inherent uncertainties associated with the necessary assumptions of modeling, this study sheds new light on surface-subsurface hydrological processes in high-alpine landscapes.
- Preprint
(50994 KB) - Metadata XML
-
Supplement
(2228 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
CC1: 'Comment on egusphere-2025-6065', Nima Zafarmomen, 08 Jan 2026
- AC1: 'Reply on CC1', Xinyang Fan, 15 Jan 2026
-
RC1: 'Comment on egusphere-2025-6065', Anonymous Referee #1, 19 Mar 2026
Overall, this is robust research presented in a well written manuscript. I was impressed with the thorough analysis you performed when comparing the simulated results to the in situ field data. You also produced clear visuals to report the results. I have some comments below in response to a few places in the paper where I thought you could add additional context.
Introduction
A paragraph is needed introducing, explaining, and justifying the use of the Water Balance Simulation Model (WaSiM) as the hydrologic model for this study. Given that this research is the first to apply WaSiM o in high-elevation environment, this explanation/knowledge gap should also be addressed in the introduction.
Additionally, I think a paragraph is needed explaining how in situ data can be used in these fully-integrated models would be helpful – particularly by highlighting the rarity of in-situ groundwater data.
Study Area
Figure 1b – The basemap is present showing the Germany label cut off. I would recommend fixing your map.
Section 3.2 You mention using the sen slope and Mann-Kendall test – please also add your p-value significance threshold, confidence intervals, etc.
Section 4.4.3 It seems to me that there should be additional statistical significance tests such as RMSE, MAE and p-values.
I am recommending adding the RMSE and MAE analysis. Also, I recommend expanding section 4.4.3 to include the NSE, spearman rank correlation coefficient, and p-value thresholds that you considered “significant” in this analysis.
Line 368: please provide the river gauge and borehole ID 4479 water fluctuation in order to compare it to the 0.8 reported for the borehold ID 4478.
Line 565: I’m not sure that the hypothesis that “subsurface lateral flow may have limited contribution to streamflow generation at the high-elevation headwater landscapes.” Is fully explained in the introduction. Therefore, I would recommend ensuring that this narrative is present throughout the paper including an additional explanation in the introduction. I’m also not sure how this is an “outrageous hypothesis” – perhaps expand on why this hypothesis is unusual for hydrologic research?
Conclusion: I think you can add a paragraph on how the findings of this research can be applied to similar high alpine ecosystems throughout the world in order to give the research a “global application”
Citation: https://doi.org/10.5194/egusphere-2025-6065-RC1 - AC2: 'Reply on RC1', Xinyang Fan, 25 Mar 2026
-
RC2: 'Comment on egusphere-2025-6065', Anonymous Referee #2, 06 May 2026
Review of the Manuscript
This is a good study that demonstrates significant effort in conducting extensive fieldwork and modeling works. However, I believe the manuscript still needs improvements. I've outlined several comments to assist in this process, as there is potential for enhancement during revision (before publication).
Major Issues
- Structure and language: The organization and clarity of the paper should be enhanced to improve comprehension.
- Subsurface flow concerns: I remain unconvinced by the claims regarding limited subsurface flow contributing to river flow.Specific Comments:
- Lines 9-11: The phrase “highlighting the need for improved subsurface parameterization” is confusing. Did you address this in your manuscript? If so, how? If it's a future recommendation, how do you plan to achieve it?
- Lines 11-13: The statement about subsurface lateral flow being constrained to zero and its limited influence on river discharge is questionable. Could this subsurface flow be as rapid as overland flow? What is your model's temporal resolution? If overland flow dominates, can you cite evidence for this, such as land erosion?
- Lines 14-16: Please clarify how point scale groundwater observations were incorporated in your modeling. Or did you just use them for model validation or calibration?
- Research Questions: The connection between research questions and the main text is unclear. Specifically, what do you mean by “the commonly adopted more simplified” in research question (iii)? How does your study differ from existing researches (literature)? Please clarify them. Clarification on how point-scale groundwater observations address your earlier identified research gaps is essential.
◦ Line 75: The term "alone" needs clarification.
- Line 76: Could you explain “external coupling framework”?
- Line 81: The phrase “with all its limits” should be reconsidered for clarity. “A missing piece in the literature”: This phrasing may imply you’re the first; please revise, as while you may be the first to implement WaSim in high-elevation glaciated environments, you may not be the first to model in this setting?
- Table 4: There appear to be no spatial variations for the parameters presented. Could you clarify this assumption? If no spatial variation exists, your claim of an “improved” subsurface schematization is, perhaps, questionable?
- Chapter 4: What is the temporal resolution of your model? I couldn’t locate this information in the manuscript.
- For soil analysis, I'm just wondering why not using SoliGrids (250 m resolution) for accuracy.
- Section 6.1: I think many paragraphs of this section should be moved to the Introduction. This part showed what has been done in previous researches and how your study cover the research gaps.
- Lines 350-354: The flow of these sentences could be improved; the conclusion about declining water availability seems unsupported by the previous sentences. While glacier melt (10 mm/year) may offset declining rainfall (-4.5 mm/year), could there be times when glacier melt ceases?
- Figure 8: The groundwater heads appear to represent groundwater depth (or depth to the water table).
- Lines 474-476: Clarification is needed regarding recharge. are you implying that some of it is released outside your study area? How significant is this recharge?
- Lines 645-646: The reasoning for being unable to model permafrost due to a lack of soil temperature data is unclear. If such data are not existing, can we just simulate these temperature?
- Lines 657-659: What evidence supports the claim of limited subsurface flow? If fast flow is primarily overland, then soil erosion may also provide insights.Citation: https://doi.org/10.5194/egusphere-2025-6065-RC2 -
RC3: 'Comment on egusphere-2025-6065', Anonymous Referee #3, 07 May 2026
Review for manuscript “New insights from fully-integrated surface-subsurface hydrological
modeling in high-elevation glaciated environments” by Fan et al., 2026
Overall, I found the manuscript well written and clearly structured. The modelling decisions are generally well supported, and the approach seems appropriate given the complexity of the system and the amount of available data. The model appears to be reasonably well constrained, especially with respect to glacier and snow processes.
I particularly appreciated the comparison between simulations with and without lateral flow. This provides a clear and interesting result: lateral flow appears to need to be inhibited in order to reproduce observed groundwater levels. However, I found the implications of this result somewhat difficult to interpret. Does this suggest that, in this system, most water transfer occurs vertically rather than laterally between cells? Or could this be a structural artefact of the modelling framework? I think this result is important and would benefit from a more in-depth discussion of its physical interpretation, limitations, and broader implications.
Figure 1 Could the meteorological station symbols be made larger or shown in a brighter colour? It took me some time to identify the 12 stations. The piezometer labels are also somewhat difficult to read against the background; adding a light label background or halo might improve readability.
Line 116 The phrase “are requested” sounds slightly awkward. Would “are obtained” or “were obtained” be more appropriate? Also, because the stations are not labelled in Figure 1, it is difficult for the reader to know where specific stations such as Zufall or Careser are located. It may be helpful either to label them in Figure 1 or to move some of this information to the data availability statement.
Table 1 Why is wind speed abbreviated as “WG”? “WS” may be more intuitive, unless there is a specific reason for using “WG”.
Line 142 I am not sure the reader needs all the details about who installed each station. It may be sufficient to include the years of operation and where the data can be accessed. The installation details could potentially be moved to the data availability statement. Similarly, if the model ultimately uses daily data, is the measurement interval necessary here?
Line 147 November is not technically winter, so “at the onset of the low-flow season” may be more accurate. I would also be cautious about stating that a single water sample collected at the end of fall helps characterize winter interactions. This may be true in a limited sense, but the current wording may overstate what can be inferred from one sampling period.
Line 159 Could you clarify how reasonable the soil profile and soil type assumptions are? Were these compared with any field measurements? The 1 km × 1 km resolution seems coarse compared with some of the other spatial inputs, so it would be helpful to briefly justify this choice or discuss its implications.
Line 165 It sounds like the model is calibrated to glacier area change and evaluated against mass balance. This is useful information, but I wonder whether it belongs in the spatial data section or would fit better in the modelling approach section.
Line 192 “Flexibly activated” is sufficient; “activated/deactivated” is probably not necessary, since deactivation is implied.
Line 197 This paragraph seems to mix the general model description with specific parameter choices. For example, why was the range of -2 to +2 selected? Were these values calibrated or assumed? This information may fit better in Section 4.4, where model parameterization and calibration decisions are discussed.
Line 199 If the bucket approach is not used, it may not need to be mentioned here.
Line 201 Could the algorithms be named more specifically, if they have formal names? The phrase “state of the art” feels somewhat overstated unless the methods are identified and their status is briefly justified.
Line 332 The sentence beginning “as global warming continues…” could likely be removed, as it introduces a broader climate-change framing that is not needed for the calibration discussion. If you want to contextualize the simulated value of -126 mm yr⁻¹, it may be more useful to compare it with existing glacier mass-balance estimates, for example from Dussaillant et al. (2025).
Dussaillant, I., Hugonnet, R., Huss, M., Berthier, E., Bannwart, J., Paul, F., and Zemp, M.: Annual mass change of the world's glaciers from 1976 to 2024 by temporal downscaling of satellite data with in situ observations, Earth System Science Data, 17, 1977–1991, 2025. https://doi.org/10.5194/essd-17-1977-2025
Line 342 Could you clarify how snowmelt and glacier melt are distinguished in the model? For example, is snowmelt occurring on glacierized cells counted as snowmelt, or is it included in glacier melt? This distinction is important for interpreting the reported melt contributions.
Line 352 The sentence beginning “Overall…” may not be necessary here. The point seems more appropriate for the introduction or broader framing rather than this section.
Figure 2 “Warm-up period” is understandable, but “spin-up period” is more commonly used in the modelling literature. Consider changing the terminology for consistency.
Figure 3 Could you clarify the reference level for river water level? The river appears to be above 0, so it is not immediately clear what 0 represents. Is this water level above the riverbed, with 0 indicating dry conditions? Or is it referenced to ground surface or another datum? A short clarification in the caption or text would help.
Figure 5 Could the line colours in panels c and d reflect the TWI classes shown in panels a and b? It may be clearer to group all simulations by TWI rather than by selected groundwater ID. At present, it is difficult to identify the IDs in panels a and b, and the groundwater borehole cell colour is similar to the TWI colour scale, which makes the figure harder to interpret. Also, where are the time series for wells 4473 and 4472?
Line 420 I found this conclusion somewhat vague. What does this result imply about the use of individual grid cells for comparison with groundwater observations? It could suggest that well-location selection needs to be done very carefully, that multiple neighbouring cells should be considered, or that some mismatch between real-world point observations and model grid cells is unavoidable. Since this issue is discussed more fully in Section 6.3, you could either clarify the implication here or leave the interpretation for the discussion section.
Figure 6 / Line 432 I think the description of low-flow performance at Pilam needs to be more careful. In Figure 6, there appears to be winter flow at S7, while the model simulates zero flow. Describing this as a “slight underestimate” seems misleading if the model is simulating no flow during periods when observations indicate some winter discharge. It also looks as though some winter periods may have measurement issues, while others, such as 2019–2020, show measurable winter flow. The 2021 winter drop also looks suspicious and may need explanation. Overall, I suggest revising the text to more explicitly acknowledge that the model does not fully capture winter low-flow dynamics at S7.
Line 457 The difficulty of the modelling task has already been mentioned several times. This sentence could probably be shortened or removed to avoid repetition.
Figure 8 I found the distinction between percolation and groundwater recharge confusing. Percolation is often understood as water draining toward groundwater, while groundwater recharge is commonly interpreted as the amount of water entering the groundwater system. In the figure, groundwater recharge appears much lower than percolation, which makes me wonder whether “groundwater recharge” is actually being used to mean net groundwater storage change or net groundwater volume change. If so, a different term may be clearer. Otherwise, the current terminology may be misleading.
I also wondered whether the model assumes a constant head beneath the glacier. If so, was this described earlier in the manuscript? Is that assumption based on model structure, observations, or another constraint?
Figure 9 When discussing this figure in the text, could you also include the terms “gaining stream” and “losing stream”? These terms are widely used to describe groundwater–surface-water exchange and would help make the interpretation clearer.
Line 515 The direct quotation feels somewhat unnecessary. The same point could likely be paraphrased and cited instead.
Line 518 I would like to see a more detailed discussion of the implications of lateral flow. Why does this result matter? What physical conditions could explain the need to inhibit lateral flow? Alternatively, could this reflect limitations of the model structure or spatial discretization?
Line 525 You could also cite Aubry-Wake et al. (2024), who used shallow groundwater head observations in a fully integrated hydrological modelling study of glacier–groundwater interactions. That study was HRU-based rather than fully distributed, but it is still relevant to the use of groundwater observations for constraining glacierized basin models:
Aubry-Wake, C., McNamara, G., Somers, L. D., McKenzie, J. M., Pomeroy, J. W., & Hellström, R. (2024). Sensitivity of surface water and groundwater contributions to streamflow in a tropical glacierized basin under climate change scenarios. Environmental Research Letters, 19, 114036. https://doi.org/10.1088/1748-9326/ad7c68
Line 567 “Outrageous” feels too strong and informal. Possible alternatives include “unlikely,” “physically improbable,” “counterintuitive,” or “inconsistent with established understanding of flow behaviour.”
Line 779 I really liked this section. It was clear and helpful.
Citation: https://doi.org/10.5194/egusphere-2025-6065-RC3
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 1,826 | 2,682 | 218 | 4,726 | 299 | 164 | 180 |
- HTML: 1,826
- PDF: 2,682
- XML: 218
- Total: 4,726
- Supplement: 299
- BibTeX: 164
- EndNote: 180
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This study implements a fully-integrated and fully-distributed surface-subsurface hydrological model using the WaSiM framework to examine water exchange processes in the high-elevation, glaciated Martell Valley in the Italian Alps. The authors utilize extensive field data, including rare in-situ high-elevation groundwater level observations—to constrain model parameters across snow, glacier, and subsurface modules
This paper is worthy of publication as it addresses a significant gap in mountain hydrology: the quantification of surface-subsurface interactions in cryosphere-dominated environments. The integration of groundwater level observations into a distributed physics-based model for a high-alpine catchment is a notable technical achievement.
1- While the model achieved satisfactory performance with subsurface lateral flow constrained to zero, this remains a counter-intuitive finding for steep alpine terrains. The authors acknowledge that their piezometers were placed in relatively flat floodplains. The finding might be a localized artifact of piezometer placement rather than a catchment-wide process. More discussion is needed on how this parameterization might affect the model’s ability to simulate hillslope-to-stream connectivity in steeper parts of the catchment.
2- The study notes a specific numerical requirement in WaSiM where the lower boundary of the soil layers must be identical to the lower boundary of the aquifer (calibrated here to 1.30 m). This 1.30 m constraint is quite shallow for an entire catchment. While it may represent the shallow porous aquifers observed, it automatically excludes any simulation of deeper groundwater flow paths. The authors should explicitly clarify if this thickness (1.30 m) is applied uniformly across the entire 77 km2 catchment, including the valley floor where sedimentary deposits are likely much deeper.
3- In Section 6.3, the recommendation to match the TWI spatial resolution to the model resolution (25m×25m) is clear. It might be helpful to include a brief sentence advising how practitioners should handle areas where the 25m grid might smooth over critical local topographic features (like small moraines) mentioned earlier in the text.
4- It would be beneficial to explicitly discuss if this "zero interflow" finding is a physical characteristic of the Martell Valley (e.g., due to high vertical conductivity in porous aquifers) or if it highlights a structural limitation in how current physics-based models distribute lateral flow in steep, shallow-soil alpine terrains. In Section 5.3, consider clarifying if "zero interflow" refers to the model parameter dr being set to 0, and briefly reiterate the physical implication for the reader.
5- As you are working on surface-subsurface hydrological model I do strongly recommend to broad your literature review and cite "Assimilation of sentinel‐based leaf area index for modeling surface‐ground water interactions in irrigation districts"