Baioni et al. (2026) A regionally based method to identify lithology-specific hydraulic conductivity distributions in shallow aquifers using catchment-scale effective values

Identification

• Journal: Journal of Hydrology
• Year: 2026
• Date: 2026-03-01
• Authors: Elisa Baioni, Hélène Hivert, Jean Marçais, Nicolas Cornette, Ronan Abhervé, Enzo Maugan, Clément Roques, Alexandre Boisson, Jean-Raynald de Dreuzy
• DOI: 10.1016/j.jhydrol.2026.135264

Research Groups

• Univ Rennes, CNRS, Geosciences Rennes - UMR 6118, Rennes, F-35000, France
• Inria, Rennes, F-35000, France
• Univ Rennes, CNRS, IRMAR - UMR 6625, F-35000 Rennes, France
• INRAE, UR Riverly, Villeurbanne, F-69625, France
• INRAE, UMR SAS 1069, Institut Agro, Rennes, F-35000, France
• Centre for Hydrogeology and Geothermics, Univ of Neuchatel, Neuchatel, 2000, Switzerland
• BRGM - DAT Bretagne, Rennes, F-35000, France

Short Summary

This paper introduces a novel method (HCDM) to infer lithology-specific hydraulic conductivity distributions in shallow aquifers using catchment-scale effective values. Validated with synthetic data and applied to 113 catchments in the Armorican Massif, the method demonstrates high predictive accuracy, with 85% of modeled conductivities falling within a 90% confidence interval of observed values.

Objective

• To infer a probability distribution of hydraulic conductivity for each lithological class from catchment-scale effective hydraulic conductivity values, assuming these are linear combinations of underlying lithology conductivities.

Study Configuration

• Spatial Scale: Catchment scale (ranging from several kilometers to several tens of kilometers), hillslope scale (approximately 1 kilometer), regional scale (Armorican Massif, France). Aquifer thickness is typically on the order of several tens of meters (e.g., 30 meters assumed for the HS1D model). Geological map resolution is 1:1,000,000, and the Digital Elevation Model (DEM) resolution is 25 meters.
• Temporal Scale: River flow time series spanning several decades (30 to 60 years) were used for calibration.

Methodology and Data

• Models used:
  • Hydraulic Conductivity Decomposition Model (HCDM): A novel method developed to infer lithology-specific hydraulic conductivity distributions.
  • HS1D (one-equivalent hillslope hydrological model): Used to derive catchment-scale effective hydraulic conductivities from river flow data.
  • SURFEX (hydrometeorological model): Used to derive recharge time series for HS1D.
  • SAFRAN (atmospheric reanalysis): Provided consistent meteorological inputs to force SURFEX.
  • MATLAB quadprog function: Employed as a numerical solver for the optimization problems within HCDM.
• Data sources:
  • River flow data from 113 gauging stations in the Armorican Massif, France (sourced from DREAL via the HydroPortail database).
  • French national geological map (1:1,000,000 scale, Chantraine et al., 1996) for lithological composition.
  • French Digital Elevation Model (DEM) BD ALTI (25 m resolution, IGN) for catchment delineation.
  • Synthetic datasets for method validation and performance assessment.

Main Results

• The HCDM successfully estimates lithology-specific hydraulic conductivity distributions from catchment-scale effective values, demonstrating robust convergence and high predictive accuracy.
• For the Armorican Massif application, 85% of the modeled effective hydraulic conductivities for 113 catchments fall within a 90% confidence interval of the observed values.
• Synthetic tests show that using approximately five catchments per lithology is sufficient to achieve 75% predictive success, with mean and standard deviation parameters estimated within 6% and 30%, respectively. With 80 catchments, prediction accuracy rises to 80%, and parameter estimation errors reduce to 3% and 15%.
• Estimated mean (mj) and standard deviation (sj) of log-normally distributed hydraulic conductivities (K, in meters per second) for the five lithological classes in the Armorican Massif are:
  • Brioverian schists: mj = -7.68, sj = 0.70
  • Primary schists: mj = -6.92, sj = 0.49
  • Paleozoic plutonic: mj = -6.29, sj = 0.49
  • Proterozoic plutonic: mj = -6.19, sj = 0.29
  • Metamorphic: mj = -6.01, sj = 0.38
• Brioverian schists are significantly less conductive than primary schists, and schists are generally less conductive than metamorphic and plutonic rocks.
• The differentiation between Paleozoic and Proterozoic plutonic rocks does not yield significantly different hydraulic conductivity distributions, suggesting general lithology is a stronger control than age in this context.
• The estimated hydraulic conductivity values are consistent with ranges reported in the literature for crystalline aquifers.
• The method is computationally efficient, with processing times ranging from 0.1 seconds for 2 lithologies to 0.4 seconds for 10 lithologies on a personal laptop.

Contributions

• Development of a novel, computationally efficient, and generalizable Hydraulic Conductivity Decomposition Model (HCDM) for inferring lithology-specific hydraulic conductivity distributions from catchment-scale effective values.
• Provides an effective tool to expand regional lithology-specific hydraulic conductivity datasets, particularly valuable for characterizing ungauged catchments and improving regional hydrogeological models.
• Demonstrates that lithology is a key controlling factor of hydraulic conductivity at plurikilometric scales in crystalline environments, a finding often difficult to establish with local measurements.
• Offers a practical framework for characterizing hydraulic conductivity at the catchment scale, a critical but challenging scale for both groundwater and surface water modeling.
• The method can be adapted to estimate other hydrogeological parameters, such as porosity, where linear averaging applies.

Funding

• Région Bretagne (SAD postdoctoral fellowship program)
• Aqui-Aqua project, SAD 2023

Citation

@article{Baioni2026regionally,
  author = {Baioni, Elisa and Hivert, Hélène and Marçais, Jean and Cornette, Nicolas and Abhervé, Ronan and Maugan, Enzo and Roques, Clément and Boisson, Alexandre and Dreuzy, Jean-Raynald de},
  title = {A regionally based method to identify lithology-specific hydraulic conductivity distributions in shallow aquifers using catchment-scale effective values},
  journal = {Journal of Hydrology},
  year = {2026},
  doi = {10.1016/j.jhydrol.2026.135264},
  url = {https://doi.org/10.1016/j.jhydrol.2026.135264}
}