the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 09 Apr 2025
Physics-based simulation of hydrological processes in a high-elevation glaciated environment focusing on groundwater
Abstract. Understanding the role of groundwater is crucial to improving the quantification of the hydrological response to climate change in high-elevation glaciated environments. However, few studies have been conducted due to the lack of in-situ hydroclimatic observations, the complex topography, and the difficulty of characterizing surface-subsurface water exchange processes in these terrains. In this study, we adopt a fully-distributed, physics-based hydrological model, WaSiM, with an integrated 2-dimensional groundwater module to quantify the observed streamflow variations and their interactions with groundwater in a high-elevation glaciated catchment (Martell Valley) in the central European Alps since the 2000s. Extensive field observations (meteorology, vegetation, glacier mass balance, soil properties, groundwater levels, river discharge) are collected to analyze hydrological processes and to constrain the model parameters. We observe that shallow alpine groundwater levels respond nearly as quickly as streamflow to snowmelt and heavy rainfall inputs, as their measured hydrographs show. Because hydrological models rarely simulate this quick groundwater response, this highlights the need for improved subsurface parametrization in hydrological modeling. Surprisingly, subsurface lateral flow plays a minor role in river discharge generation at the study site, providing new insights into the hydrological processes in such an environment. Lastly, our results underline the challenges of integrating point-scale groundwater observations into a distributed hydrological model, with important implications for future piezometer installation in the field. This study sheds new light on surface-subsurface hydrological processes in high-elevation glaciated environments. It highlights the importance of improving subsurface representation in hydrological modeling.
- Preprint
(20655 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
CC1: 'Comment on egusphere-2025-1500', Nima Zafarmomen, 11 Apr 2025
- AC1: 'Reply on CC1', Xinyang Fan, 13 Jun 2025
-
EC1: 'Comment on egusphere-2025-1500', Nunzio Romano, 30 Apr 2025
Dear Dr. Zafarmomen,
First of all, I thank you for your comments, and indeed it is always desirable for an article to receive comments not only from designated reviewers but also from individuals actively working on the subject of the article.Nevertheless, comments should maintain a plausible tone and, more importantly, when criticism is made, one should always try to provide the authors with actionable ways to address the problem raised.
Moreover, much more than in the past, one should be careful not to violate ethical norms. This happens, as in your case, when discussants or referees ask to cite their own articles. I do not think this is a good practice at all. I am also raising this question because I am aware of the existing literature on the topic of this submission, and your suggested papers are not the only ones.
Citation: https://doi.org/10.5194/egusphere-2025-1500-EC1 -
RC1: 'Comment on egusphere-2025-1500', Anonymous Referee #1, 23 May 2025
GENERAL COMMENTS
The authors applied a distributed hydrological model, WaSiM to a 77-km2 alpine catchment in the Italian Alps to examine the hydrological behaviour of the catchment with a particular focus on groundwater. The study objectives are stated as: (1) Can the alpine surface-subsurface processes be modeled with WaSiM? (2) How to set up a fully-distributed, physics-based model to reproduce streamflow? To answer these questions, the authors set up and run the model to simulate streamflow, thereby meeting the operational objectives. I commend the authors for the time and efforts that took to produce the results presented in the manuscript. Important scientific contribution of the study would have been to develop a model that actually captures the essential physics of the catchment, and examine the spatial distribution and temporal dynamics of groundwater in a physically realistic manner. Unfortunately, the model has fundamental deficiencies in representing the essential hydrogeological characteristics of the catchment. Therefore, the groundwater component of the model is neither physics-based nor fully-distributed, and some of the conclusions stated in the manuscript are not scientifically defensible. I will elaborate more on this in my specific comments below. I would encourage the authors to build a model that reflects at least the essential features (not the details) of groundwater in the alpine catchment using WaSiM or another modelling platform.
SPECIFIC COMMENTS
Line 171-180. Both domains. What does this mean? Soil and groundwater? How can they have the same depth?
Line 223. This assumption may be applicable to some catchments, but it is not applicable to alpine catchments, where main aquifers are either thick Quaternary deposit (talus, moraine, rock glaciers, etc.) or weathered bedrock. Without an ability to represent these alpine aquifers in a meaningful manner, the model is not ‘physics-based’ in this particular environment. I have been to the study site and have seen the hydrogeological characteristics of S4, S6, S7, and S8 sub-catchment. Based on my observation, Quaternary deposits are important aquifers in these sub-catchments.
Line 225. DTW are mostly within 1m. This may be true in the valley-floor sediments near the streams, but DTW are generally much deeper in Quaternary deposits and weather bedrock at higher elevation. Therefore, the model is lacking the essential characteristics of alpine aquifers feeding the valley floor. If the authors decide to build a new model, please represent the varying depth of different types of aquifers over the catchment based on satellite imagery or digital elevation model (e.g., Mueller et al., HESS, 26, 6029–6054).
Line 230. Spatially homogeneous. Over the past two decades, numerous field-based studies have shown that alpine aquifers consist of individual units of periglacial and paraglacial sediments distributed heterogeneously in the complex terrain, and underling weathered bedrock. Spatially homogeneous property (hydraulic conductivity, thickness, etc.) completely contradicts the findings of field-based studies. Therefore, the model may be spatially distributed for snow processes, but not for groundwater processes. If the authors decide to build a new model, please represent physically reasonable hydrogeological properties for different types of aquifers.
Line 237. Transformed separately to the stream. If I understood correctly, surface and subsurface runoff is routed to the stream with no inter-grid transfer. This approach may be useful for water balance simulation, but will likely misrepresent hillslope processes where lateral flow plays a major role both in surface and subsurface runoff. Therefore, the model simulation results cannot be used to substantiate conclusions regarding the nature of runoff processes (see my comment on Line 391-393).
Line 256. The depth of aquifer (= 1.3 m) is unrealistically small, except for the regions near stream channels, in alpine catchments. Please use variable depths representing the essential feature of alpine environments in a new model.
Line 266-267. A well-designed model is indeed useful for understanding the processes, when the model captures the essence of the system. Unfortunately, the current model completely misses the essential feature of alpine groundwater. Please develop a new model that captures the essence of the system and use it to gain new understanding of alpine hydrological processes.
Line 329-331. Did the authors have a spin-up period prior to 2007? In Figure 2c, the warm-up period starts in 2007, and the glacier mass balance is consistently negative after 2007. Did the increase in glacier coverage occur prior to 2007? Please explain.
Line 345-347. If the trend is statistically insignificant, then this statement cannot be substantiated by data. Please delete this and the following sentence.
Line 352. Groundwater levels. The figure shows the depth to water table, not the water level referenced to a common elevation datum. Please change the terminology to avoid confusion.
Line 355. 30m away from the river. These wells are close to the stream, and are likely strongly influenced by the stream water level. This is especially true if surface and subsurface runoff from hillslopes is directly routed to the stream (see my comment on Line 237).
Line 362. Highly heterogeneous. The water-table depths in ID 4478 and 4479 are quite similar, within 1m of each other. This is homogeneous in a relative sense. The more important heterogeneity in alpine catchments generally occurs between different aquifer units, for example between stream-side aquifer and talus on a hillslope. Please represent more important heterogeneity in a new model.
Line 385. Lateral flow. It is not clear how the lateral flow affects the water table in stream-side aquifers, if it is directly routed to the stream. Please explain.
Line 391-393. This is bold statement that challenges the commonly accepted paradigm of hillslope hydrology, which is based on decades of careful observation of hillslope processes in experimental catchments. To make such a bold statement, the model needs to reflect the essence of hillslope hydrology, which requires at least two-dimensional (vertical and horizontal) flow that captures the processes resulting in saturation overland flow and subsurface storm flow. Please consider using a more realistic model if the authors need to make the statement.
Line 414. Resolution of 25m. Please include a scale in Figures 5a and 5b.
Line 425. Figures 5c and 5d show the water-table depth. It is difficult to understand the interaction of neighboring wells from the water-table data alone. Please use the hydraulic head to demonstrate the dynamics of individual cells and their interaction.
Line 433. Underestimation of the discharge. Underestimation during the winter period is most likely due to the lack of representation of groundwater contributions from hillslope aquifers. They are not properly represented in the model, but numerous field studies have shown that these aquifers sustain winter flow in alpine streams. Please represent these aquifers in a new model.
Line 450. The model underestimates discharge. Please see my comment on Line 433.
Line 455. Groundwater baseflow in S7 and S8 in are very small (0.02-0.04 mm/d). This is unrealistic for alpine catchments in a humid region.
Line 471-473. Important contribution of baseflow. Baseflow should be almost 100% during winter months.
Line 521. Physics-based, fully-distributed. The model presented in this manuscript may be physics-based and fully-distributed for snow and other processes, but it is not for groundwater processes. Please build a new model that captures the essential physics of aquifers in alpine catchments.
Line 546-547. This is a bold statement. Please see my comment on Line 391-393.
Line 551. Please see my comment on Line 391-393.
Line 583-584. This statement is not convincing. Water table in stream-side aquifers are hydraulically connected, as demonstrated by the similarity between ID 4478 and 4479 in Figure 3a. This is different from snow accumulation, which is highly sensitive to local-scale variability resulting from the complex terrain. Please do not simply attribute the mismatch between observed and simulated data to scale issues.
Line 595. Very low discharge. This is not particularly low for winter flow in alpine catchments. Winter flow of 0.5-1 mm/d is commonly observed in alpine catchments in humid regions around the world.
Line 605. 0.2-0.6 mm/d. This only applies to S3. S2, S7, and S8 have 0.01-0.06 mm/d, which is completely unrealistic, most likely because distributed aquifers are not properly represented in the model.
Line 622. WaSiM is indeed a physics-based, distributed model; but its potential was not properly utilized in the current manuscript. Please utilize the full potential of WaSiM or another modelling platform to build a model that captures the essential physis of alpine aquifers.
Line 642. Crucial contribution of baseflow. The model-simulated baseflow is unrealistically small (see my comment on Line 605). Please use a new model to demonstrate the crucial contribution.
Citation: https://doi.org/10.5194/egusphere-2025-1500-RC1 -
AC2: 'Reply on RC1', Xinyang Fan, 13 Jun 2025
Dear reviewer, thanks for taking time to review our manuscript. Here we provide a generic response to discuss the key points raised in the comments (see the attached response). Should a revision is advised by the editor, we will provide a point-to-point response to address the comments in detail.
-
AC2: 'Reply on RC1', Xinyang Fan, 13 Jun 2025
-
RC2: 'Comment on egusphere-2025-1500', Anonymous Referee #2, 04 Jun 2025
General Comments
The manuscript by Fan et al. presents a detailed and technically sophisticated application of a fully distributed, physics-based hydrological model, the Water Balance Simulation Model (WaSiM) to investigate surface–subsurface hydrological interactions in the glacierized catchment of Martell Valley, South Tyrol, European Alps. This study addresses a significant gap in our understanding of groundwater dynamics in alpine cryospheric environments. I appreciate the authors' effort in undertaking this important and challenging task.The strength of this manuscript is the use of extensive observational data, often lacking in high-mountain environments. The comprehensive implementation of WaSiM to simulate both surface and subsurface hydrological processes is impressive. However, the manuscript would benefit from a clear articulation of its novel scientific contributions in relation to existing studies. I suggest including a dedicated comparison with similar modelling efforts to better highlight what is gained or potentially lost by the chosen approach.
Additionally, the limitations of applying such a model in complex mountain terrain, particularly with respect to spatial resolution, assumptions regarding subsurface properties, and data coarseness, should be discussed in detail with a separate section addressing model uncertainties, assumptions (e.g. uniform aquifer thickness). Further elaboration on key methodological decisions, such as the use of dual melt approaches, justification for subsurface parameterization, and the model’s ability to represent delayed responses in groundwater will also help strengthen the manuscript.
Specific Comments
The authors position their study as one of the first to develop and implement a physics-based modelling framework to simulate surface–subsurface interactions in a glacierized environment. I agree with this claim, as there are indeed only a limited number of comparable studies. However, I recommend that the authors include a dedicated section comparing their approach with similar studies to clarify what is gained (or lost) by using this approach.The authors provide information about the catchment surface conditions, e.g. 40% bare rock, 34% grassland, and so on. I recommend that the authors include similar information about the valley floor settings (sediment-filled region), as most of the movement occurs in this region. Additionally, providing information on the land cover (bare rock, grassland, forest) in the map (Figure 1) can help readers better understand the area and relevance for hydrology.
In Section 4.1 and Table 3: the description of the snow/rain partitioning scheme using the parameters TRS (i.e. temperature at which half of the precipitation falls as snow [0 °C]) and Ttrans (i.e. half of the temperature range from snow to rain [+2 °C]) could benefit from further clarification or simplification. Some readers who are less familiar with hydrological modelling may find the current phrasing difficult to understand. I suggest explicitly stating that this defines a linear transition of snow fraction from 100% at –2 °C, 50% at 0 °C, and 0% at +2 °C. Additionally, the authors may explain the use of two different approaches (EB and T-index) for snow and ice melt, even though both processes are forms of cryospheric melt that are influenced by similar energy exchanges?
4.2. Unsaturated zone and groundwater model: In a mountain environment, especially when glacierized, the use of uniform aquifer properties ignores the highly heterogeneous nature of the subsurface system. Such assumptions may hold true in plains, where natural depositions are relatively uniform, but not in mountainous environments. The authors later in the results (L361–362, and on several other occasions) imply the heterogeneity of the subsurface properties. Also, why are vertical flows modelled but not horizontal flows in the unsaturated zone? Please explain. Are the authors sure about the absence of permafrost in the upper parts of the study area?
4.3. Streamflow generation: The model computes surface runoff, subsurface lateral flow, and baseflow separately and routes them using a reservoir cascade. This lumped treatment may oversimplify the interactions and timing differences among flow components.
Since groundwater/subsurface water dynamics are central to the study, I suggest the authors justify the use of 1.3 m as subsurface thickness across the entire catchment. The region is dominated by bare rock, grasslands, forest, and glaciers. It is obvious that subsurface conditions and thickness are not uniform. I believe the thickness of 1.3 m is low for such regions, especially in valley floors, with former glacial depositions or landslides. The authors mention numerical reasons and limitations, I understand that. In that case, the authors need to do a sensitivity test to see the influence of variable thickness on subsurface flow and storage, and on the overall outcomes. Additionally, the use of 1.3 m contradicts Figure 3(a), as the hydraulic head of borewell ID 4478 is around the same depth.
The model ran at 25×25 m resolution. Isn’t the soil profile from the global database (Harmonized World Soil Database), which has a resolution of 1 km, too coarse, particularly in a topographically complex region? The use of such data in a 77 km² area may not represent local settings (like moraine or similar glacial deposits). Please clarify.
The authors have adopted manual calibration following Fatichi et al. (2015) to avoid computational load and time. This is justifiable, but manual calibration may be inefficient and subjective when dealing with many parameters across modules. Please clarify how the authors overcame this and what standardized approaches were used.
The authors point out an 8% underestimation of the annual glacier MB. An additional explanation about this is required. Since one glacier is used for calibration, the model may not have captured the heterogeneity of glacier responses in the catchment, as glacier MB is also affected by aspect, elevation, and other non-climatic parameters.
In Figure 2(c), please also include information about the plot within the circle. Though it is mentioned in the text, I believe some explanation is also required in the caption.
L358: Such low discharge can also be contributed by frozen streams in winter at lower elevations, apart from the reasons mentioned.
L369–370: Is it due to the use of 1.3 m as the subsurface thickness and homogeneous subsurface properties in the model? I believe the groundwater level should show a lag time in response to surface conditions. It may also be due to the model time step (i.e. daily).
L385–394: I believe that the delayed response, especially in the early melting period, is valid. Additionally, the statement challenging the commonly adopted hydrological modelling approach about the role of soil needs more evidence.
Section 5.6 can be made shorter; I see several repetitions of points already mentioned earlier.
Citation: https://doi.org/10.5194/egusphere-2025-1500-RC2 - AC3: 'Reply on RC2', Xinyang Fan, 13 Jun 2025
Status: closed
-
CC1: 'Comment on egusphere-2025-1500', Nima Zafarmomen, 11 Apr 2025
- The reliance on manual calibration for a complex, fully-distributed model with numerous parameters is a significant limitation. While the authors justify this choice due to computational constraints and data availability, the manuscript would benefit from a more thorough discussion of the implications of manual calibration on parameter uncertainty and model robustness. For instance, how sensitive are the key findings (e.g., minor role of subsurface lateral flow) to parameter choices? A sensitivity analysis, even if limited, could strengthen the credibility of the results.
- The multi-signal calibration approach is a strength, but the sequential calibration process (surface to subsurface) may introduce biases. The authors should discuss potential dependencies between modules (e.g., snow module influencing groundwater recharge) and how these were addressed to ensure consistency across the calibration steps.- The finding that subsurface lateral flow plays a minor role in streamflow generation is intriguing but requires further scrutiny. The assumption of homogeneous subsurface properties (e.g., hydraulic conductivity, storage coefficient) across the catchment may oversimplify the complex geology of the Martell Valley, which is noted to be heterogeneous (Section 2). This assumption could bias the model toward underestimating lateral flow. The authors should explore whether spatially variable subsurface parameters, informed by available geological data, could alter this conclusion.
- The model’s inability to reproduce the strong winter recession at borehole ID 4478 (Section 5.3) suggests limitations in capturing preferential flow paths or other subsurface processes. The manuscript would benefit from a deeper discussion of alternative mechanisms (e.g., macropore flow, fractured bedrock) that could explain this discrepancy, potentially supported by literature or additional field observations.- The challenges of integrating point-scale groundwater observations into a distributed model are well-articulated, but the proposed solutions (e.g., using TWI to guide piezometer placement, comparing neighboring cells) need more rigorous evaluation. For instance, how representative are the TWI-based recommendations for other high-alpine catchments with different topographic or geologic characteristics? A sensitivity analysis of TWI resolution or comparison with other topographic indices could enhance the generalizability of these recommendations.
- The manuscript highlights the mismatch between observed and modeled river networks due to DEM uncertainties (Section 5.6.3). This issue could significantly affect groundwater-surface water interactions, yet it is only briefly addressed. A quantitative assessment of DEM uncertainty (e.g., comparing simulations with different DEM resolutions) would strengthen the discussion and provide more concrete guidance for future studies.- The authors note that the Martell Valley is relatively dry compared to other Alpine catchments (Section 5.6.2), which may limit the applicability of findings to wetter environments. Similarly, the lithology (crystalline bedrocks, shallow soils) may not be representative of other high-alpine settings. The discussion should more explicitly address the conditions under which the key findings (e.g., minor role of lateral flow, rapid groundwater response) are likely to hold, potentially by comparing with studies in contrasting catchments.
- The manuscript claims that the rapid groundwater response is rarely simulated by hydrological models (Section 6), but this statement requires more substantiation. A brief review of other physics-based models (e.g., HydroGeoSphere, ParFlow) and their ability to capture such dynamics would contextualize the novelty of WaSiM’s performance and clarify the need for improved subsurface parameterization.- The underestimation of winter baseflow (Sections 5.4, 6) is attributed to shallow river channels and homogeneous subsurface parameterization, but observational uncertainties in low-flow measurements (e.g., sensor limitations in freezing conditions) are also significant (Section 5.6.4). The manuscript should more clearly disentangle model limitations from observational uncertainties, possibly by discussing the reliability of winter discharge data or exploring alternative data sources (e.g., tracer studies) to validate baseflow contributions.
- The claim that baseflow contributes significantly to winter streamflow (up to 40% in some subcatchments, Section 5.4) is compelling but relies on model simulations rather than direct observations. Additional evidence, such as isotopic or chemical tracers, could corroborate this finding and enhance confidence in the model’s representation of baseflow dynamics.
- The introduction is comprehensive but lengthy, with some repetition (e.g., challenges of alpine hydrology are mentioned multiple times). Streamlining the introduction to focus on key gaps and the study’s objectives would improve readability.
- Section 5.6 is titled “Challenges and opportunities for modeling high-alpine glaciated environment,” but it primarily discusses challenges. Explicitly addressing opportunities (e.g., leveraging remote sensing, integrating machine learning for parameter estimation) would balance the narrative and highlight future research directions.
- The use of abbreviations (e.g., PEQ, TWI, DTW) is frequent, and a glossary or table defining these terms would aid readers unfamiliar with the terminology.- Figure 8 is visually rich but overwhelming due to the number of panels. Consider splitting it into two figures (e.g., one for percolation/recharge, another for groundwater level/exfiltration/infiltration) or using a subset of months to improve clarity.
- Table 3 and Table 4 list calibrated parameters but lack units for some parameters (e.g., “Scaling for max.deposition” in Table 3). Ensuring consistency in units and providing brief explanations for less intuitive parameters would enhance accessibility.
- The caption for Figure 5 could clarify that panels (c-d) show simulations for multiple grid cells, as this is not immediately obvious from the figure alone.I highly recommend to discuss the SWAT-MODFLOW papers which integrates SW-GW and cite below paper:
Assimilation of sentinel‐based leaf area index for modeling surface‐ground water interactions in irrigation districtsCitation: https://doi.org/10.5194/egusphere-2025-1500-CC1 - AC1: 'Reply on CC1', Xinyang Fan, 13 Jun 2025
-
EC1: 'Comment on egusphere-2025-1500', Nunzio Romano, 30 Apr 2025
Dear Dr. Zafarmomen,
First of all, I thank you for your comments, and indeed it is always desirable for an article to receive comments not only from designated reviewers but also from individuals actively working on the subject of the article.Nevertheless, comments should maintain a plausible tone and, more importantly, when criticism is made, one should always try to provide the authors with actionable ways to address the problem raised.
Moreover, much more than in the past, one should be careful not to violate ethical norms. This happens, as in your case, when discussants or referees ask to cite their own articles. I do not think this is a good practice at all. I am also raising this question because I am aware of the existing literature on the topic of this submission, and your suggested papers are not the only ones.
Citation: https://doi.org/10.5194/egusphere-2025-1500-EC1 -
RC1: 'Comment on egusphere-2025-1500', Anonymous Referee #1, 23 May 2025
GENERAL COMMENTS
The authors applied a distributed hydrological model, WaSiM to a 77-km2 alpine catchment in the Italian Alps to examine the hydrological behaviour of the catchment with a particular focus on groundwater. The study objectives are stated as: (1) Can the alpine surface-subsurface processes be modeled with WaSiM? (2) How to set up a fully-distributed, physics-based model to reproduce streamflow? To answer these questions, the authors set up and run the model to simulate streamflow, thereby meeting the operational objectives. I commend the authors for the time and efforts that took to produce the results presented in the manuscript. Important scientific contribution of the study would have been to develop a model that actually captures the essential physics of the catchment, and examine the spatial distribution and temporal dynamics of groundwater in a physically realistic manner. Unfortunately, the model has fundamental deficiencies in representing the essential hydrogeological characteristics of the catchment. Therefore, the groundwater component of the model is neither physics-based nor fully-distributed, and some of the conclusions stated in the manuscript are not scientifically defensible. I will elaborate more on this in my specific comments below. I would encourage the authors to build a model that reflects at least the essential features (not the details) of groundwater in the alpine catchment using WaSiM or another modelling platform.
SPECIFIC COMMENTS
Line 171-180. Both domains. What does this mean? Soil and groundwater? How can they have the same depth?
Line 223. This assumption may be applicable to some catchments, but it is not applicable to alpine catchments, where main aquifers are either thick Quaternary deposit (talus, moraine, rock glaciers, etc.) or weathered bedrock. Without an ability to represent these alpine aquifers in a meaningful manner, the model is not ‘physics-based’ in this particular environment. I have been to the study site and have seen the hydrogeological characteristics of S4, S6, S7, and S8 sub-catchment. Based on my observation, Quaternary deposits are important aquifers in these sub-catchments.
Line 225. DTW are mostly within 1m. This may be true in the valley-floor sediments near the streams, but DTW are generally much deeper in Quaternary deposits and weather bedrock at higher elevation. Therefore, the model is lacking the essential characteristics of alpine aquifers feeding the valley floor. If the authors decide to build a new model, please represent the varying depth of different types of aquifers over the catchment based on satellite imagery or digital elevation model (e.g., Mueller et al., HESS, 26, 6029–6054).
Line 230. Spatially homogeneous. Over the past two decades, numerous field-based studies have shown that alpine aquifers consist of individual units of periglacial and paraglacial sediments distributed heterogeneously in the complex terrain, and underling weathered bedrock. Spatially homogeneous property (hydraulic conductivity, thickness, etc.) completely contradicts the findings of field-based studies. Therefore, the model may be spatially distributed for snow processes, but not for groundwater processes. If the authors decide to build a new model, please represent physically reasonable hydrogeological properties for different types of aquifers.
Line 237. Transformed separately to the stream. If I understood correctly, surface and subsurface runoff is routed to the stream with no inter-grid transfer. This approach may be useful for water balance simulation, but will likely misrepresent hillslope processes where lateral flow plays a major role both in surface and subsurface runoff. Therefore, the model simulation results cannot be used to substantiate conclusions regarding the nature of runoff processes (see my comment on Line 391-393).
Line 256. The depth of aquifer (= 1.3 m) is unrealistically small, except for the regions near stream channels, in alpine catchments. Please use variable depths representing the essential feature of alpine environments in a new model.
Line 266-267. A well-designed model is indeed useful for understanding the processes, when the model captures the essence of the system. Unfortunately, the current model completely misses the essential feature of alpine groundwater. Please develop a new model that captures the essence of the system and use it to gain new understanding of alpine hydrological processes.
Line 329-331. Did the authors have a spin-up period prior to 2007? In Figure 2c, the warm-up period starts in 2007, and the glacier mass balance is consistently negative after 2007. Did the increase in glacier coverage occur prior to 2007? Please explain.
Line 345-347. If the trend is statistically insignificant, then this statement cannot be substantiated by data. Please delete this and the following sentence.
Line 352. Groundwater levels. The figure shows the depth to water table, not the water level referenced to a common elevation datum. Please change the terminology to avoid confusion.
Line 355. 30m away from the river. These wells are close to the stream, and are likely strongly influenced by the stream water level. This is especially true if surface and subsurface runoff from hillslopes is directly routed to the stream (see my comment on Line 237).
Line 362. Highly heterogeneous. The water-table depths in ID 4478 and 4479 are quite similar, within 1m of each other. This is homogeneous in a relative sense. The more important heterogeneity in alpine catchments generally occurs between different aquifer units, for example between stream-side aquifer and talus on a hillslope. Please represent more important heterogeneity in a new model.
Line 385. Lateral flow. It is not clear how the lateral flow affects the water table in stream-side aquifers, if it is directly routed to the stream. Please explain.
Line 391-393. This is bold statement that challenges the commonly accepted paradigm of hillslope hydrology, which is based on decades of careful observation of hillslope processes in experimental catchments. To make such a bold statement, the model needs to reflect the essence of hillslope hydrology, which requires at least two-dimensional (vertical and horizontal) flow that captures the processes resulting in saturation overland flow and subsurface storm flow. Please consider using a more realistic model if the authors need to make the statement.
Line 414. Resolution of 25m. Please include a scale in Figures 5a and 5b.
Line 425. Figures 5c and 5d show the water-table depth. It is difficult to understand the interaction of neighboring wells from the water-table data alone. Please use the hydraulic head to demonstrate the dynamics of individual cells and their interaction.
Line 433. Underestimation of the discharge. Underestimation during the winter period is most likely due to the lack of representation of groundwater contributions from hillslope aquifers. They are not properly represented in the model, but numerous field studies have shown that these aquifers sustain winter flow in alpine streams. Please represent these aquifers in a new model.
Line 450. The model underestimates discharge. Please see my comment on Line 433.
Line 455. Groundwater baseflow in S7 and S8 in are very small (0.02-0.04 mm/d). This is unrealistic for alpine catchments in a humid region.
Line 471-473. Important contribution of baseflow. Baseflow should be almost 100% during winter months.
Line 521. Physics-based, fully-distributed. The model presented in this manuscript may be physics-based and fully-distributed for snow and other processes, but it is not for groundwater processes. Please build a new model that captures the essential physics of aquifers in alpine catchments.
Line 546-547. This is a bold statement. Please see my comment on Line 391-393.
Line 551. Please see my comment on Line 391-393.
Line 583-584. This statement is not convincing. Water table in stream-side aquifers are hydraulically connected, as demonstrated by the similarity between ID 4478 and 4479 in Figure 3a. This is different from snow accumulation, which is highly sensitive to local-scale variability resulting from the complex terrain. Please do not simply attribute the mismatch between observed and simulated data to scale issues.
Line 595. Very low discharge. This is not particularly low for winter flow in alpine catchments. Winter flow of 0.5-1 mm/d is commonly observed in alpine catchments in humid regions around the world.
Line 605. 0.2-0.6 mm/d. This only applies to S3. S2, S7, and S8 have 0.01-0.06 mm/d, which is completely unrealistic, most likely because distributed aquifers are not properly represented in the model.
Line 622. WaSiM is indeed a physics-based, distributed model; but its potential was not properly utilized in the current manuscript. Please utilize the full potential of WaSiM or another modelling platform to build a model that captures the essential physis of alpine aquifers.
Line 642. Crucial contribution of baseflow. The model-simulated baseflow is unrealistically small (see my comment on Line 605). Please use a new model to demonstrate the crucial contribution.
Citation: https://doi.org/10.5194/egusphere-2025-1500-RC1 -
AC2: 'Reply on RC1', Xinyang Fan, 13 Jun 2025
Dear reviewer, thanks for taking time to review our manuscript. Here we provide a generic response to discuss the key points raised in the comments (see the attached response). Should a revision is advised by the editor, we will provide a point-to-point response to address the comments in detail.
-
AC2: 'Reply on RC1', Xinyang Fan, 13 Jun 2025
-
RC2: 'Comment on egusphere-2025-1500', Anonymous Referee #2, 04 Jun 2025
General Comments
The manuscript by Fan et al. presents a detailed and technically sophisticated application of a fully distributed, physics-based hydrological model, the Water Balance Simulation Model (WaSiM) to investigate surface–subsurface hydrological interactions in the glacierized catchment of Martell Valley, South Tyrol, European Alps. This study addresses a significant gap in our understanding of groundwater dynamics in alpine cryospheric environments. I appreciate the authors' effort in undertaking this important and challenging task.The strength of this manuscript is the use of extensive observational data, often lacking in high-mountain environments. The comprehensive implementation of WaSiM to simulate both surface and subsurface hydrological processes is impressive. However, the manuscript would benefit from a clear articulation of its novel scientific contributions in relation to existing studies. I suggest including a dedicated comparison with similar modelling efforts to better highlight what is gained or potentially lost by the chosen approach.
Additionally, the limitations of applying such a model in complex mountain terrain, particularly with respect to spatial resolution, assumptions regarding subsurface properties, and data coarseness, should be discussed in detail with a separate section addressing model uncertainties, assumptions (e.g. uniform aquifer thickness). Further elaboration on key methodological decisions, such as the use of dual melt approaches, justification for subsurface parameterization, and the model’s ability to represent delayed responses in groundwater will also help strengthen the manuscript.
Specific Comments
The authors position their study as one of the first to develop and implement a physics-based modelling framework to simulate surface–subsurface interactions in a glacierized environment. I agree with this claim, as there are indeed only a limited number of comparable studies. However, I recommend that the authors include a dedicated section comparing their approach with similar studies to clarify what is gained (or lost) by using this approach.The authors provide information about the catchment surface conditions, e.g. 40% bare rock, 34% grassland, and so on. I recommend that the authors include similar information about the valley floor settings (sediment-filled region), as most of the movement occurs in this region. Additionally, providing information on the land cover (bare rock, grassland, forest) in the map (Figure 1) can help readers better understand the area and relevance for hydrology.
In Section 4.1 and Table 3: the description of the snow/rain partitioning scheme using the parameters TRS (i.e. temperature at which half of the precipitation falls as snow [0 °C]) and Ttrans (i.e. half of the temperature range from snow to rain [+2 °C]) could benefit from further clarification or simplification. Some readers who are less familiar with hydrological modelling may find the current phrasing difficult to understand. I suggest explicitly stating that this defines a linear transition of snow fraction from 100% at –2 °C, 50% at 0 °C, and 0% at +2 °C. Additionally, the authors may explain the use of two different approaches (EB and T-index) for snow and ice melt, even though both processes are forms of cryospheric melt that are influenced by similar energy exchanges?
4.2. Unsaturated zone and groundwater model: In a mountain environment, especially when glacierized, the use of uniform aquifer properties ignores the highly heterogeneous nature of the subsurface system. Such assumptions may hold true in plains, where natural depositions are relatively uniform, but not in mountainous environments. The authors later in the results (L361–362, and on several other occasions) imply the heterogeneity of the subsurface properties. Also, why are vertical flows modelled but not horizontal flows in the unsaturated zone? Please explain. Are the authors sure about the absence of permafrost in the upper parts of the study area?
4.3. Streamflow generation: The model computes surface runoff, subsurface lateral flow, and baseflow separately and routes them using a reservoir cascade. This lumped treatment may oversimplify the interactions and timing differences among flow components.
Since groundwater/subsurface water dynamics are central to the study, I suggest the authors justify the use of 1.3 m as subsurface thickness across the entire catchment. The region is dominated by bare rock, grasslands, forest, and glaciers. It is obvious that subsurface conditions and thickness are not uniform. I believe the thickness of 1.3 m is low for such regions, especially in valley floors, with former glacial depositions or landslides. The authors mention numerical reasons and limitations, I understand that. In that case, the authors need to do a sensitivity test to see the influence of variable thickness on subsurface flow and storage, and on the overall outcomes. Additionally, the use of 1.3 m contradicts Figure 3(a), as the hydraulic head of borewell ID 4478 is around the same depth.
The model ran at 25×25 m resolution. Isn’t the soil profile from the global database (Harmonized World Soil Database), which has a resolution of 1 km, too coarse, particularly in a topographically complex region? The use of such data in a 77 km² area may not represent local settings (like moraine or similar glacial deposits). Please clarify.
The authors have adopted manual calibration following Fatichi et al. (2015) to avoid computational load and time. This is justifiable, but manual calibration may be inefficient and subjective when dealing with many parameters across modules. Please clarify how the authors overcame this and what standardized approaches were used.
The authors point out an 8% underestimation of the annual glacier MB. An additional explanation about this is required. Since one glacier is used for calibration, the model may not have captured the heterogeneity of glacier responses in the catchment, as glacier MB is also affected by aspect, elevation, and other non-climatic parameters.
In Figure 2(c), please also include information about the plot within the circle. Though it is mentioned in the text, I believe some explanation is also required in the caption.
L358: Such low discharge can also be contributed by frozen streams in winter at lower elevations, apart from the reasons mentioned.
L369–370: Is it due to the use of 1.3 m as the subsurface thickness and homogeneous subsurface properties in the model? I believe the groundwater level should show a lag time in response to surface conditions. It may also be due to the model time step (i.e. daily).
L385–394: I believe that the delayed response, especially in the early melting period, is valid. Additionally, the statement challenging the commonly adopted hydrological modelling approach about the role of soil needs more evidence.
Section 5.6 can be made shorter; I see several repetitions of points already mentioned earlier.
Citation: https://doi.org/10.5194/egusphere-2025-1500-RC2 - AC3: 'Reply on RC2', Xinyang Fan, 13 Jun 2025
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 573 | 147 | 24 | 744 | 19 | 36 |
- HTML: 573
- PDF: 147
- XML: 24
- Total: 744
- BibTeX: 19
- EndNote: 36
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
- The reliance on manual calibration for a complex, fully-distributed model with numerous parameters is a significant limitation. While the authors justify this choice due to computational constraints and data availability, the manuscript would benefit from a more thorough discussion of the implications of manual calibration on parameter uncertainty and model robustness. For instance, how sensitive are the key findings (e.g., minor role of subsurface lateral flow) to parameter choices? A sensitivity analysis, even if limited, could strengthen the credibility of the results.
- The multi-signal calibration approach is a strength, but the sequential calibration process (surface to subsurface) may introduce biases. The authors should discuss potential dependencies between modules (e.g., snow module influencing groundwater recharge) and how these were addressed to ensure consistency across the calibration steps.
- The finding that subsurface lateral flow plays a minor role in streamflow generation is intriguing but requires further scrutiny. The assumption of homogeneous subsurface properties (e.g., hydraulic conductivity, storage coefficient) across the catchment may oversimplify the complex geology of the Martell Valley, which is noted to be heterogeneous (Section 2). This assumption could bias the model toward underestimating lateral flow. The authors should explore whether spatially variable subsurface parameters, informed by available geological data, could alter this conclusion.
- The model’s inability to reproduce the strong winter recession at borehole ID 4478 (Section 5.3) suggests limitations in capturing preferential flow paths or other subsurface processes. The manuscript would benefit from a deeper discussion of alternative mechanisms (e.g., macropore flow, fractured bedrock) that could explain this discrepancy, potentially supported by literature or additional field observations.
- The challenges of integrating point-scale groundwater observations into a distributed model are well-articulated, but the proposed solutions (e.g., using TWI to guide piezometer placement, comparing neighboring cells) need more rigorous evaluation. For instance, how representative are the TWI-based recommendations for other high-alpine catchments with different topographic or geologic characteristics? A sensitivity analysis of TWI resolution or comparison with other topographic indices could enhance the generalizability of these recommendations.
- The manuscript highlights the mismatch between observed and modeled river networks due to DEM uncertainties (Section 5.6.3). This issue could significantly affect groundwater-surface water interactions, yet it is only briefly addressed. A quantitative assessment of DEM uncertainty (e.g., comparing simulations with different DEM resolutions) would strengthen the discussion and provide more concrete guidance for future studies.
- The authors note that the Martell Valley is relatively dry compared to other Alpine catchments (Section 5.6.2), which may limit the applicability of findings to wetter environments. Similarly, the lithology (crystalline bedrocks, shallow soils) may not be representative of other high-alpine settings. The discussion should more explicitly address the conditions under which the key findings (e.g., minor role of lateral flow, rapid groundwater response) are likely to hold, potentially by comparing with studies in contrasting catchments.
- The manuscript claims that the rapid groundwater response is rarely simulated by hydrological models (Section 6), but this statement requires more substantiation. A brief review of other physics-based models (e.g., HydroGeoSphere, ParFlow) and their ability to capture such dynamics would contextualize the novelty of WaSiM’s performance and clarify the need for improved subsurface parameterization.
- The underestimation of winter baseflow (Sections 5.4, 6) is attributed to shallow river channels and homogeneous subsurface parameterization, but observational uncertainties in low-flow measurements (e.g., sensor limitations in freezing conditions) are also significant (Section 5.6.4). The manuscript should more clearly disentangle model limitations from observational uncertainties, possibly by discussing the reliability of winter discharge data or exploring alternative data sources (e.g., tracer studies) to validate baseflow contributions.
- The claim that baseflow contributes significantly to winter streamflow (up to 40% in some subcatchments, Section 5.4) is compelling but relies on model simulations rather than direct observations. Additional evidence, such as isotopic or chemical tracers, could corroborate this finding and enhance confidence in the model’s representation of baseflow dynamics.
- The introduction is comprehensive but lengthy, with some repetition (e.g., challenges of alpine hydrology are mentioned multiple times). Streamlining the introduction to focus on key gaps and the study’s objectives would improve readability.
- Section 5.6 is titled “Challenges and opportunities for modeling high-alpine glaciated environment,” but it primarily discusses challenges. Explicitly addressing opportunities (e.g., leveraging remote sensing, integrating machine learning for parameter estimation) would balance the narrative and highlight future research directions.
- The use of abbreviations (e.g., PEQ, TWI, DTW) is frequent, and a glossary or table defining these terms would aid readers unfamiliar with the terminology.
- Figure 8 is visually rich but overwhelming due to the number of panels. Consider splitting it into two figures (e.g., one for percolation/recharge, another for groundwater level/exfiltration/infiltration) or using a subset of months to improve clarity.
- Table 3 and Table 4 list calibrated parameters but lack units for some parameters (e.g., “Scaling for max.deposition” in Table 3). Ensuring consistency in units and providing brief explanations for less intuitive parameters would enhance accessibility.
- The caption for Figure 5 could clarify that panels (c-d) show simulations for multiple grid cells, as this is not immediately obvious from the figure alone.
I highly recommend to discuss the SWAT-MODFLOW papers which integrates SW-GW and cite below paper:
Assimilation of sentinel‐based leaf area index for modeling surface‐ground water interactions in irrigation districts