Amiri et al. (2025) Modeling root water uptake of landscape groundcovers with HYDRUS-1D and particle swarm optimization

Identification

• Journal: Agricultural Water Management
• Year: 2025
• Date: 2025-11-19
• Authors: Zahra Amiri, Amir Haghverdi
• DOI: 10.1016/j.agwat.2025.109972

Research Groups

• Department of Environmental Sciences, University of California, Riverside, CA, USA
• Department of Civil, Environmental, and Architectural Engineering, University of Colorado, Boulder, CO, USA

Short Summary

This study coupled HYDRUS-1D with Particle Swarm Optimization (PSO) to calibrate root water uptake (RWU) parameters for four groundcover species, demonstrating that simultaneous optimization of RWU and soil hydraulic parameters significantly enhances the accuracy of soil water content (SWC) simulations compared to a two-step approach.

Objective

• To optimize root water uptake (RWU) parameters of four landscape groundcovers using the Particle Swarm Optimization (PSO) algorithm.
• To assess the capability of the PSO algorithm to calibrate soil hydraulic and RWU parameters via two inversion approaches (two-step vs. simultaneous).
• To evaluate soil water balance components of selected groundcover species.

Study Configuration

• Spatial Scale: Four drought-tolerant groundcover species (Acacia redolens, Arctotis acaulis, Chrysanthemoides incana, Lippia nodiflora) planted in 12 plots (3.05 m × 3.05 m) with three replicates. Soil profile simulated to a depth of 100 cm, with measurements at 10 cm, 30 cm, 50 cm, and 75 cm soil depths.
• Temporal Scale: Calibration periods: November 2, 2020 – March 31, 2021 (bare soil for soil hydraulic parameters) and April 24 – October 31, 2022 (groundcovers for RWU and combined parameters). Validation period: May 1 – October 31, 2023 (groundcovers).

Methodology and Data

• Models used:
  • HYDRUS-1D (for simulating one-dimensional water movement, using the van Genuchten-Mualem model for soil hydraulic properties and the Feddes RWU model).
  • Particle Swarm Optimization (PSO) algorithm (for parameter calibration).
  • Extended Fourier Amplitude Sensitivity Testing (EFAST) (for global sensitivity analysis).
  • Generalized Likelihood Uncertainty Estimation (GLUE) (for uncertainty analysis).
• Data sources:
  • In-situ Time Domain Reflectometer (TDR) sensors: Soil water content (SWC) at 10 cm, 30 cm, 50 cm, and 75 cm depths (30-minute intervals).
  • AccuPAR LP-80 and WinRHIZO system: Leaf Area Index (LAI).
  • Core sampler and WinRHIZO system: Root Length Density (RLD) down to 70 cm.
  • California Irrigation Management Information System (CIMIS) weather station: Daily air temperature, reference evapotranspiration (ETo), precipitation, total solar radiation, and relative humidity.
  • Weathermatic ET-based smart irrigation controller: Irrigation amounts.

Main Results

• Scenario I (Two-step optimization): Showed satisfactory SWC simulation at shallow depths (10 cm and 30 cm) but significant deviations at deeper layers (50 cm and 75 cm) during validation, resulting in Kling–Gupta efficiency (KGE) values for SWC ranging from 0.64 to 0.84 and coefficients of determination (R²) from 0.49 to 0.79. However, soil water storage and actual evapotranspiration (ET) were accurately estimated (KGE > 0.85, R² > 0.85, RMSE for ET 0.9–1.3 mm/day).
• Scenario II (Simultaneous optimization): Substantially reduced the deviation between simulated and observed SWC, achieving KGE and R² values above 0.90 for all groundcovers (except Arctotis acaulis: KGE = 0.86, R² = 0.80) during validation. RMSE for SWC was below 0.018 cm³/cm³ (except Arctotis acaulis: 0.024 cm³/cm³). Soil water storage and actual ET also showed excellent agreement (KGE > 0.83, R² > 0.85, RMSE for ET 0.9–1.3 mm/day).
• Sensitivity and Uncertainty: Soil hydraulic parameters, particularly shape parameters (α, n) across soil layers, were the most influential on model performance. Root water uptake (RWU) parameters (h1, h2) had moderate influence, while h3h, h3l, and h4 were the least influential under well-watered conditions. Uncertainty analysis showed notable reductions in confidence intervals for soil hydraulic shape parameters (up to 50.2%), but negligible reductions for RWU parameters (< 6%).
• Soil Water Balance: Simulated actual transpiration (Ta) was lowest for Lippia nodiflora (765.9 mm during calibration, 447.0 mm during validation) and approximately 50% higher for other species. Actual evaporation (Ea) contributed 35% of actual ET for Lippia nodiflora, but less than 7% for other species. Deep percolation (DP) was insignificant, less than 3.6% of total applied water.

Contributions

• First application of the Particle Swarm Optimization (PSO) algorithm with HYDRUS-1D for determining root water uptake (RWU) parameters for landscape groundcovers.
• Demonstrated that simultaneous optimization of RWU and soil hydraulic parameters significantly enhances HYDRUS-1D performance, particularly for accurate simulation of soil water content (SWC) dynamics at all depths, compared to a two-step approach.
• Provided species-specific calibrated Feddes RWU parameters for four drought-tolerant groundcover species (Acacia redolens, Arctotis acaulis, Chrysanthemoides incana, and Lippia nodiflora).
• Offered a robust framework for urban hydrologic modeling, supporting improved irrigation management and the evaluation of urban vegetation's role in stormwater management, green roof efficiency, and urban runoff mitigation.

Funding

Not explicitly stated in the provided text.

Citation

@article{Amiri2025Modeling,
  author = {Amiri, Zahra and Haghverdi, Amir},
  title = {Modeling root water uptake of landscape groundcovers with HYDRUS-1D and particle swarm optimization},
  journal = {Agricultural Water Management},
  year = {2025},
  doi = {10.1016/j.agwat.2025.109972},
  url = {https://doi.org/10.1016/j.agwat.2025.109972}
}