Kaur et al. (2025) Estimation of soil hydraulic parameters for fine-textured soils using HYDRUS-1D coupled with PEST

Identification

• Journal: Agricultural Water Management
• Year: 2025
• Date: 2025-11-08
• Authors: Ishmeet Kaur, Afua Adobea Mante, Ramanathan Sri Ranjan, Francis Zvomuya, Kayla Moore, Kurt Gottfried, Taras E. Lychuk
• DOI: 10.1016/j.agwat.2025.109947

Research Groups

• Department of Biosystems Engineering, University of Manitoba, Winnipeg, MB, Canada
• Department of Soil Science, University of Manitoba, Winnipeg, MB, Canada
• Agriculture and Agri-Food Canada, Brandon Research and Development Centre, Brandon, MB, Canada

Short Summary

This study utilized HYDRUS-1D coupled with PEST to estimate van Genuchten-Mualem soil hydraulic parameters for a fine-textured soil in Manitoba over four growing seasons, revealing the non-uniqueness of solutions and identifying stable parameter ranges through sequential calibrations.

Objective

• To evaluate the effectiveness of a one-dimensional domain setup within HYDRUS coupled with PEST to estimate soil hydraulic parameters for a fine-textured soil in southern Manitoba over four growing seasons.
• To identify the maximum likelihood parameter ranges based on repeated calibrations and simulation performance.

Study Configuration

• Spatial Scale: A one-dimensional soil profile of 1.3 m depth (131 nodes) at station MB5 (49.62°N, -97.96°W) in the Carman-Elm Creek area, Manitoba, Canada. The soil is classified as Gleyed Humic Vertisol (Red River soils).
• Temporal Scale: Four growing seasons (May to September) from 2016 to 2019. Calibration periods covered approximately 80 days, followed by validation periods until harvest (40-50 days).

Methodology and Data

• Models used:
  • HYDRUS-1D (version 5.03.0240) for simulating one-dimensional water flow using the Richards equation and van Genuchten-Mualem hydraulic models.
  • PEST (Parameter ESTimation) software for external parameter calibration and optimization using the Gauss-Marquardt-Levenberg method.
  • ROSETTA pedotransfer function for initial estimates of van Genuchten-Mualem parameters.
  • FAO Penman-Monteith model for reference evapotranspiration (ET₀) calculation.
  • Feddes water stress response model for actual transpiration.
• Data sources:
  • Field-measured volumetric soil water content (Stevens Hydra Probes II installed at 0.05 m, 0.20 m, 0.50 m, and 1.00 m depths, recorded at 15-minute intervals and daily averaged).
  • On-site weather station data (minimum and maximum temperature, average relative humidity, precipitation, wind direction and speed).
  • Solar radiation data from Manitoba Agriculture station at Elm Creek.
  • Crop characteristics (sowing/harvesting dates, Leaf Area Index (LAI) derived from Growing Degree Days (GDD)).
  • Data obtained from the Real-Time In-situ Soil Monitoring for Agriculture (RISMA) network stations (Agriculture and Agri-Food Canada and Environment and Climate Change Canada).

Main Results

• The coupling of HYDRUS-1D with PEST effectively estimated soil hydraulic parameters but revealed the non-uniqueness of solutions, as multiple parameter sets could produce comparable model performance.
• Sequential calibrations with 100 randomized initial parameter sets and performance-based filtering (NSE > 0.5, PBIAS within ±10 %, RSR < 0.7, R² > 0.5) successfully identified stable, maximum-likelihood parameter ranges.
• Model performance was satisfactory for 2016, 2017, and 2019 (Nash-Sutcliffe efficiency coefficient: 0.65 – 0.92; percent bias: -9.31 % – 1.48 %; ratio of root mean square error to standard deviation: 0.29–0.59; coefficient of determination: 0.68 – 0.96).
• Model validation for 2018 showed poor performance (NSE = -1.13, PBIAS = -18.51 %, RSR = 1.46, R² = 0.01), attributed to high intra-season weather variability (e.g., 0.0274 m rainfall in 3600 s, with 0.0217 m in 900 s on July 11, 2018) causing physical changes in soil structure not captured by the single-porosity model.
• Saturated hydraulic conductivity (Ks) in the top 0–0.1 m layer exhibited substantial year-to-year variation: 1.39 × 10⁻⁶ m/s (2016, 2019), 1.16 × 10⁻⁷ m/s (2017), and 2.43 × 10⁻⁷ m/s (2018).
• Ks in deeper layers (below 0.1 m) remained relatively stable, generally below 2.32 × 10⁻⁷ m/s across all years, indicating restrictive subsoil conditions.
• Strong compensatory interactions were observed between the residual volumetric soil water content (θr) and the pore size distribution index (n) (correlation coefficients ≈0.78–0.95), necessitating fixing θr values for improved calibration stability.

Contributions

• Demonstrated a robust methodology for estimating soil hydraulic parameters in fine-textured Vertisolic soils of cold-climate cropping systems by coupling HYDRUS-1D with PEST.
• Addressed the critical issue of non-uniqueness in inverse modeling by implementing sequential calibrations with randomized initial parameter sets and performance-based filtering to derive stable and physically meaningful parameter ranges.
• Quantified the significant temporal variability of saturated hydraulic conductivity in the topsoil, highlighting the influence of soil-plant-weather interactions and the need for site- and season-specific parameter characterization for effective agricultural water management.
• Provided locally adapted modeling insights for Manitoba's unique soil and climatic conditions, contributing to improved water management strategies in the region.

Funding

• Natural Sciences and Engineering Research Council (NSERC)
• Sustainable Canadian Agricultural Partnership Program
• University of Manitoba Graduate Fellowship

Citation

@article{Kaur2025Estimation,
  author = {Kaur, Ishmeet and Mante, Afua Adobea and Ranjan, Ramanathan Sri and Zvomuya, Francis and Moore, Kayla and Gottfried, Kurt and Lychuk, Taras E.},
  title = {Estimation of soil hydraulic parameters for fine-textured soils using HYDRUS-1D coupled with PEST},
  journal = {Agricultural Water Management},
  year = {2025},
  doi = {10.1016/j.agwat.2025.109947},
  url = {https://doi.org/10.1016/j.agwat.2025.109947}
}