Devoie et al. (2025) Modelling near-surface ice content and midwinter melt events in mineral soils

Identification

• Journal: Environmental Modelling & Software
• Year: 2025
• Date: 2025-12-05
• Authors: Élise Devoie, Renato Pardo Lara, Aaron Berg, William L. Quinton, James R. Craig
• DOI: 10.1016/j.envsoft.2025.106816

Research Groups

• Department of Civil Engineering, Queen’s University, Kingston, ON, Canada
• Department of Geography, Environment and Geomatics, University of Guelph, Guelph, ON, Canada
• Cold Regions Research Centre, Wilfrid Laurier University, Waterloo, ON, Canada
• Department of Civil and Environmental Engineering, University of Waterloo, Waterloo, ON, Canada

Short Summary

This study presents a numerically efficient, semi-analytical coupled thermal and mass transport model to represent near-surface soil ice content and midwinter melt events in mineral soils. The model capably reproduces field observations of frozen, thawed, or transitioning soils and offers significant computational advantages over existing continuum models, making it suitable for regional hydrologic applications.

Objective

• To extend a previously developed semi-analytical interface model for organic soils with permafrost to mineral soils without permafrost.
• To evaluate the extended model against a continuum model benchmark.
• To apply the model to intermittently frozen soil data collected at the Kenaston Field site in Saskatchewan, Canada, focusing on short-duration freezing in the near-surface soil.

Study Configuration

• Spatial Scale: Field data collected from 22 stations within the 40 km² Kenaston Network (Brightwater Creek basin, Saskatchewan, Canada), with most instrumentation in a flat 10 km² sub-region. Soil moisture and temperature sensors installed at 5 cm, 20 cm, and 50 cm depths. Model domain extended to a vertical depth of 15 m. The model includes a near-surface "buffer" layer of 85 mm thickness.
• Temporal Scale: Field data collected at 30-minute intervals over 4–6 years (2014–2020). Model simulations run for 5 years. Interface model simulations typically used 1-day or 1-hour timesteps, while the continuum model used timesteps on the order of minutes (e.g., 1 hour, 3 minutes) depending on spatial discretization.

Methodology and Data

• Models used:
  • Interface model: A semi-analytical coupled thermal and mass transport model, extended from Devoie and Craig (2020), tracking pore ice formation and mean soil temperature in terms of enthalpy, and the depth of freezing/thawing.
  • Continuum model: A coupled solution of the unsaturated Richards’ equation and an energy balance (including conduction, advection, and phase change), solved via a finite volume method with operator splitting (from Devoie et al., 2019), used for benchmarking.
  • Degree-day method: A simplified model (from Fox, 1992) used for comparison.
• Data sources:
  • Field observations: Soil moisture (HydraProbes electromagnetic sensors), soil temperature, and precipitation (tipping bucket rain gauges) from the Kenaston Network, Saskatchewan, Canada.
  • Soil properties: Representative soil from Kenaston site 1 (28% sand, 53% silt, 19% clay) used for base computational tests. Site-specific freezing point depression estimated using a logistic growth model fit to soil freezing curves (Pardo Lara et al., 2020).
  • Boundary conditions: Surface temperature estimated from 5 cm soil temperature data. Mass flux at the surface applied seasonally (average evapotranspiration rate of -2.07 mm/d in spring/summer, average recharge rate of 2.42 mm/d in fall). Fixed soil temperature of 5 °C at 15 m depth.

Main Results

• The interface model capably reproduces field observations of frozen, thawed, or transitioning soils, showing excellent agreement with the continuum model (RMSE of 0.04 for total ice content).
• The interface model is significantly more computationally efficient, running approximately three orders of magnitude faster (4.5 seconds vs. 2 hours 22 minutes for a specific event simulation) than the continuum model.
• The interface model accurately captures the timing of freeze/thaw events, with overall agreement between modelled and measured near-surface freeze/thaw states exceeding 90% (e.g., 92% for site 3, 94% for sites 15 and 20).
• The model is most sensitive to the freezing point and buffer layer thickness, and less sensitive to residual unfrozen water content and domain length.
• The interface model is particularly effective at representing early fall freezing events, which have significant hydrological importance.
• Methods are proposed for integrating the interface model into hydrologic models to improve predictions of hydraulic conductivity, infiltration, and storage capacity under frozen conditions.

Contributions

• Presents a novel, numerically efficient, semi-analytical coupled thermal and mass transport model that accurately simulates near-surface soil ice content and midwinter melt events in mineral soils.
• Fills a critical gap between computationally intensive physically-based continuum models and low-fidelity empirical models, offering a practical solution for regional-scale hydrologic modeling.
• Demonstrates high computational expediency, making the model suitable for integration into operational forecasting tools and large-scale hydrologic models where computational cost is a limiting factor.
• Extends the applicability of a proven interface model from organic soils with permafrost to intermittently freezing mineral soils without permafrost.
• Provides a framework and recommendations for incorporating the model's outputs (e.g., bulk temperature, unfrozen water content) into existing hydrologic models to improve predictions of hydraulic conductivity, infiltration, and storage capacity in cold regions.
• Offers improved estimates of over-winter streamflow and flood potential, particularly relevant in regions experiencing increased frequency of midwinter freeze/thaw events due to climate change.

Funding

• NSERC (Natural Sciences and Engineering Research Council of Canada)
• CIFAR (Canadian Institute for Advanced Research)
• Government of the Northwest Territories (through partnership agreement with Wilfrid Laurier University)
• Cold Regions Research Centre
• ArcticNet (through support of the Dehcho Collaborative on Permafrost (DCoP))

Citation

@article{Devoie2025Modelling,
  author = {Devoie, Élise and Lara, Renato Pardo and Berg, Aaron and Quinton, William L. and Craig, James R.},
  title = {Modelling near-surface ice content and midwinter melt events in mineral soils},
  journal = {Environmental Modelling & Software},
  year = {2025},
  doi = {10.1016/j.envsoft.2025.106816},
  url = {https://doi.org/10.1016/j.envsoft.2025.106816}
}