the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 19 Mar 2026
A non-explicit representation of macropores in the SVS land surface model improves streamflow simulations under frozen soil conditions
Abstract. Soil freezing is a major cold region process that influences hydrological response of northern catchments, in particular during winter rainfall and snowmelt events. Ice within the soil matrix reduces the pore space available for water to infiltrate, while the presence of soil macropores in structured soils maintains rapid water percolation even in frozen conditions. Representing the complex effect of soil freezing on water infiltration in land surface models is a challenging task. This is particularly true for operational models, where physical process integration must balance performance improvements against computational efficiency and complexity. In this study, we propose a conceptual approach to represent the effects of macropores on frozen soil infiltration into the Soil, Vegetation, and Snow (SVS) model used within the operational prediction systems of Environment and Climate Change Canada (ECCC). We assessed the effects of this new configuration (Fr-MP) on streamflow simulations at more than 580 hydrometric stations located in the Great-Lakes and Saint-Lawrence domain over a five-year period. The conceptual representation of macropores improves the Kling-Gupta Efficiency (KGE) at 88 % of the assessed stations, resulting in an increase in the median KGE of 0.28 compared to the configuration without macropores and soil freezing. Detailed analysis of a decomposed hydrograph shows that the Fr-MP configuration increases SVS soil drainage (slow response) and reduces surface runoff and lateral flow (quick response). To ensure that the proposed change is also acceptable in the context of operational numerical weather prediction, an evaluation of its impact on soil freezing depth as well as screen-level temperature and dew point temperature predictions is performed against in-situ observations. These results support the potential operational implementation of the Fr-MP configuration at ECCC for numerical weather and streamflow prediction.
- Preprint
(4194 KB) - Metadata XML
-
Supplement
(1233 KB) - BibTeX
- EndNote
Status: open (until 30 Apr 2026)
- RC1: 'Comment on egusphere-2026-928', Anonymous Referee #1, 29 Mar 2026 reply
-
RC2: 'Comment on egusphere-2026-928', Anonymous Referee #2, 17 Apr 2026
reply
General comment:
This study presents a computationally efficient approach to improve infiltration under frozen soil conditions in a land surface model. The manuscript clearly demonstrates that the current soil freezing formulation leads to unrealistic runoff responses and that increasing infiltration can substantially improve streamflow simulations at large spatial scales. The topic is highly relevant, particularly in the context of operational hydrological forecasting, and the study provides a valuable and pragmatic contribution in this regard.However, the proposed “macropore representation” is, in its current form, a conceptual parameterization that mimics the effect of preferential flow rather than explicitly representing macropore processes. The implementation effectively removes the ice-induced hydraulic impedance once a soil moisture threshold is exceeded and additionally modifies the runoff generation scheme. As such, the approach represents a combined structural modification of flow partitioning rather than a direct representation of macropore flow.
In its current form, the manuscript does not sufficiently disentangle these effects. A clearer separation of the impacts of the increased vertical hydraulic conductivity and the modified runoff generation would be necessary to properly attribute the improvements in model performance.
Furthermore, the physical interpretation of the results could be strengthened. In particular, the manuscript would benefit from a more explicit discussion of:
- the conceptual nature and limitations of the parameterization,
- the implications of representing subgrid-scale preferential flow using a bulk threshold formulation,
- the spatial variability in model performance, especially in relation to land use, soil texture, and the exclusion of stations from the evaluation.
Finally, the transferability of the calibrated parameter α_MP within and beyond the study domain is not addressed and should be discussed.
Overall, I consider the manuscript suitable for publication after revision. The proposed approach is particularly interesting from an operational perspective, as it provides a simple and efficient way to improve infiltration under frozen conditions. I am especially interested to see how this concept will evolve in future work toward more physically based representations of preferential flow processes.
Minor comments and technical notes:
L. 44: The sentence is difficult to follow. Consider restructuring it for clarity (e.g. “This makes the representation … complex and challenging.”).
L. 87, 204, 250, 273, Figure 1 caption: The manuscript frequently refers to a “macropore representation” or “macropore configuration”. Given that the approach does not explicitly represent macropore flow but rather mimics its effect on infiltration, I suggest using a more precise terminology (e.g. “conceptual macropore parameterization” or “macropore flow correction”) consistently throughout the manuscript.
L. 150: The statement that a unique residual unfrozen water content is assigned independently of soil temperature is somewhat unclear, as Eq. (1) explicitly includes temperature. I may be misinterpreting the formulation. However, it would be helpful to clarify whether this temperature dependence is averaged out to obtain a constant value for each soil type.L. 214: The proposed parameterization represents a strong simplification of preferential flow processes. In particular it does not account for dynamic effects such as refreezing or the dependence of macropore flow on soil type and saturation. Some experimental studies even suggest that preferential flow can decrease at high saturation levels or vary significantly with soil structure. A short discussion (maybe later in the manuscript) of these limitations would strengthen the manuscript.
L. 311: The exclusion of stations based on NSE/KGE thresholds is understandable. However, it would be helpful to briefly explain the choice of the KGE threshold (-0.41) and its interpretation, as this value is not immediately intuitive for the readers.L. 359: There appears to be a missing “Fr-” before the word “experiment”.
Figure3/Figure 5: The spatial distribution of excluded stations (red crosses) shows a pronounced clustering in specific regions rather than a random pattern. While the manuscript discusses potential causes for poor model performance in these areas, it remains unclear whether the station filtering systematically removes specific hydro-climatic or physiographic conditions from the evaluation. Please clarify how this affects the representativeness of the results.
L. 372: Please consider using more precise terminology such as “macropore parameterization” instead of implying an explicit representation.
L. 406: Please clarify whether this statement applies to all stations or only the majority of them.
L. 408: The underlying reasons for the observed behavior are not entirely clear. Additional explanation would improve the interpretation.
L. 414: Could the observed spatial variability in model performance be linked to differences in land use or soil properties? While some aspects are discussed, it might be helpful to further elaborate on the controlling factors (e.g. soil texture, land cover, or hydrological regime).
L. 418: The results indicate substantial performance degradations associated with the soil freezing configuration in some regions. While general limitations of the freezing scheme are discussed later, it would be helpful to more explicitly link these limitations to the observed regional patterns in model performance.
L. 421: The influence of station exclusion on the interpretation of PBIAS should be clarified, especially since different configurations are evaluated on different subsets of stations. This may affect the comparability of the reported bias between experiments. In addition, the use of the term “demonstrates” appears too strong in this context. Given the differences in station subsets and the indirect nature of the comparison, “suggests” may be more appropriate.
L. 439: While some explanations are provided, the underlying drivers of the observed spatial patterns are not fully clear. Could these patterns be more explicitly linked to dominant hydrological processes (e.g. snowmelt dynamics, flow partitioning, or model structural limitations)?
Figure 8: Adding a finer temporal reference (e.g. months or seasons) to the x-axis would improve readability and help distinguish between winter, snowmelt, and summer events.
Furthermore, differences between the model configurations appear to persist during summer periods, when soil freezing should not play a significant role. This may be related to the modified runoff generation scheme introduced together with the macropore parameterization. Please clarify whether such differences under unfrozen conditions are expected within the current model formulation.Figure 9: Please clarify whether the improved agreement is related to specific freezing conditions during the selected period.
L. 610: The influence of soil type on the presented results could be discussed in more detail, as it may significantly affect the applicability and transferability of the parameterization. In particular, experimental studies suggest that the role of preferential flow under frozen conditions can vary with soil properties and saturation state. A brief discussion of this variability would strengthen the manuscript.
L. 615: The macropore configuration is implemented together with a modification of the runoff generation scheme by disabling the subgrid-scale interflow routine at the surface. This represents an additional structural change to the model. It would be helpful to more clearly separate and discuss the respective impacts of this modification and the macropore parameterization on the results.
Citation: https://doi.org/10.5194/egusphere-2026-928-RC2
Model code and software
Code of the Soil Vegetation and Snow (SVS) land surface scheme integrated in the ECCC Surface Prediction System with the official physics package that includes the macropore configuration Benjamin Bouchard, Vincent Vionnet, Étienne Gaborit, and Vincent Fortin https://doi.org/10.5281/zenodo.18664365
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 85 | 42 | 9 | 136 | 21 | 8 | 10 |
- HTML: 85
- PDF: 42
- XML: 9
- Total: 136
- Supplement: 21
- BibTeX: 8
- EndNote: 10
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The manuscript addresses a significant and timely topic in cold-region hydrology. The methodology is sound, and the results provide a clear path toward improving operational streamflow forecasting. However, the manuscript requires a more critical discussion regarding the physical basis of the macropore parameterization and a deeper analysis of the spatial variability in model performance. Addressing the major comments regarding the physical justification and the sensitivity of the alpha_MP parameter will significantly enhance the impact and scientific rigor of the work. I look forward to seeing a revised version of this study.
Major Comments：
1. The authors propose a non-explicit, threshold-based representation of macropores. While this is computationally efficient for operational forecasting, the manuscript lacks a robust discussion on the physical limitations of this simplification. Specifically, how does this approach account for the spatial heterogeneity of macropore networks at the grid-cell scale? A more thorough comparison with existing dual-domain or dual-permeability models (e.g., those cited in the introduction) is needed to justify why this simplified approach is sufficient for large-scale hydrological modeling.
2. The calibration of alpha_MP is a central component of the study. However, the manuscript does not sufficiently address the potential for equifinality or the transferability of this parameter across different hydro-climatic regions. Are the optimal values of alpha_MP consistent across the diverse landscapes of the Great Lakes and Saint-Lawrence domain? A more detailed analysis of the spatial variability of the optimal alpha_MP and its relationship with soil/land-cover characteristics would significantly strengthen the paper.
3. The results indicate that the Fr-MP configuration performs differently in agricultural areas compared to forested catchments. The authors attribute this to the interaction between tile drainage and frozen soil. I suggest a more in-depth analysis of this interaction. Does the current model structure adequately decouple the effects of macropores from those of anthropogenic drainage systems? This is crucial for the model's reliability in human-altered landscapes.
4. While the authors evaluate the impact on TT and TD, the improvements are described as "slight" or "neutral." Given that the primary goal is to improve streamflow, the authors should clarify whether these meteorological improvements are statistically significant and if they have any meaningful impact on the broader numerical weather prediction (NWP) performance at ECCC.
Minor Comments：
1. Page 2, Line 40: The term "discrepancy" is used well here; ensure this is consistently linked to the scale-dependency of the processes throughout the discussion.
2. Page 7, Eq. 9: Please clarify if alpha_MP is a static parameter or if it could be dynamic based on seasonal vegetation growth.
3. Page 8, Figure 1: The conceptual diagram is excellent. Please ensure the legend clearly distinguishes between the "matrix flow" and "macropore flow" pathways in panel (c).
4. Page 10, Section 2.4.1: The exclusion criteria for stations (NSE < 0 or KGE < -0.41) are reasonable, but please provide a brief justification for why these specific thresholds were chosen over others.
5. Page 13, Line 355: The mention of the "perched aquifer" in Michigan is interesting. Could the authors provide a bit more detail on how this specific geological feature might be better represented in future iterations of the model?
6. Page 16, Line 411: The observation that only 4 stations show a Delta KGE < -0.5 is a strong point. Please emphasize this in the abstract as a measure of the model's robustness.
7. Page 23, Line 495: The authors mention that wetlands and meanders are not fully represented in Watroute. It would be beneficial to include a brief sentence on how the transition to the "Raven" routing scheme might specifically address these limitations.
8. Clarity of Figures: In Figure 4 and Figure 5, the color scales are effective, but please ensure that the symbols for different drainage areas are easily distinguishable in the printed version.
9. References: Ensure that all cited preprints (e.g., Bauer et al., 2025) are updated to their final publication status if available by the time of revision.