Rooij (2025) Fitting the junction model and other parameterizations for the unsaturated soil hydraulic conductivity curve: KRIAfitter version 1.0

Identification

• Journal: Geoscientific model development
• Year: 2025
• Date: 2025-10-08
• Authors: Gerrit H. de Rooij
• DOI: 10.5194/gmd-18-6921-2025

Research Groups

• Department of Soil System Science, Helmholtz Centre for Environmental Research – UFZ, Halle, Germany

Short Summary

This paper introduces and evaluates a novel "junction model" for unsaturated soil hydraulic conductivity (UHCC) that simplifies the conceptualization of water domains (films or capillaries) and documents the associated Fortran fitting code, KRIAfitter. The junction model often provides good fits to observed data with fewer parameters compared to existing multi-domain models, offering a more parsimonious alternative, though some complex soil behaviors may still require models with more parameters.

Objective

• To introduce and evaluate a new "junction model" for unsaturated soil hydraulic conductivity that assigns liquid water exclusively to either film or capillary domains, thereby sidestepping the complexities of averaging parallel domain conductivities.
• To document KRIAfitter version 1.0, a Fortran code designed to fit the parameters of the junction model and other existing conductivity models to observed unsaturated hydraulic conductivity data.

Study Configuration

• Spatial Scale: Laboratory-scale measurements on soil samples (13 diverse soils).
• Temporal Scale: Steady-state relationships for soil hydraulic properties under isothermal conditions (typically 20 °C).

Methodology and Data

• Models used:
  • Junction model (JUV): A new model for UHCC, where liquid water is entirely in films (dry range) or capillaries (wet range), joined at a critical matric potential.
  • RIA parameterization: A sigmoidal junction model for the soil water retention curve (SWRC).
  • Other UHCC models: Unweighted additive model (ADV), arithmetic mean model (AMV), geometric mean model (GMV), harmonic mean model (HMV), and Kosugi model with capillary water only (KGV).
  • Vapour flow model: De Rooij's (2024a) modification of Peters' (2013) model.
  • Fitting software: KRIAfitter version 1.0 (Fortran code) for UHCC, RIAfitter version 2.0 for SWRC.
  • Optimization algorithm: Shuffled Complex Evolution (SCE) algorithm.
• Data sources:
  • Observed soil water retention and unsaturated hydraulic conductivity data for 13 soils, primarily from Weber et al. (2019) and the UNSODA database (e.g., UNSODA soil 4142).
  • Laboratory measurements, with matric potentials ranging from 0.0 meters to -1000 meters of water column (equivalent to -100000 cm H2O).

Main Results

• The junction model (JUV) and the unweighted additive model (ADV) consistently provided good fits to unsaturated hydraulic conductivity data across a range of soils.
• The junction model, with up to six fitting parameters, often achieved comparable performance to other multi-domain models with one fewer parameter, suggesting that models with six or more parameters may be overparameterized for many soils.
• For certain soils (e.g., Rehovot sand), the additive model's extra parameter was necessary to accurately reproduce abrupt transitions in the hydraulic conductivity curve.
• Geometric mean (GMV) and harmonic mean (HMV) models generally underperformed, often underestimating conductivity in the wet range.
• The classical Kosugi model (KGV), despite its simplicity, performed remarkably well and often closely tracked the junction model.
• Based on the corrected Akaike's information criterion (AICc), the junction model was the most successful, requiring an average of 4.25 parameters for optimal fits, compared to 6 parameters for the lowest root mean square error (RMSE).
• Parameter correlations were highly dependent on the specific input data, with no universal patterns observed, necessitating case-by-case examination.
• The Fortran code KRIAfitter 1.0 was successfully developed and documented, providing a robust tool for fitting these models.

Contributions

• Introduction of a novel and parsimonious "junction model" for unsaturated soil hydraulic conductivity, which offers a physically plausible alternative to existing multi-domain models by avoiding the problematic averaging of parallel domain conductivities.
• Development and comprehensive documentation of KRIAfitter 1.0, a Fortran-based software tool that facilitates the fitting of the new junction model and five other established conductivity models to observed data.
• Extensive evaluation of the junction model against existing models using data from 13 diverse soils, demonstrating its effectiveness and highlighting its advantage in terms of parameter parsimony for many soil types.
• Providing insights into the potential for overparameterization in complex conductivity models and the importance of considering model complexity (e.g., via AICc) in addition to fit quality (RMSE).

Funding

• Helmholtz Centre for Environmental Research – UFZ (covered article processing charges).

Citation

@article{Rooij2025Fitting,
  author = {Rooij, Gerrit H. de},
  title = {Fitting the junction model and other parameterizations for the unsaturated soil hydraulic conductivity curve: KRIAfitter version 1.0},
  journal = {Geoscientific model development},
  year = {2025},
  doi = {10.5194/gmd-18-6921-2025},
  url = {https://doi.org/10.5194/gmd-18-6921-2025}
}