任泽岭 et al. (2026) AquaCrop-derived surrogate model for crop yield response to monthly irrigation deficit

Identification

• Journal: Agricultural Systems
• Year: 2026
• Date: 2026-09-26
• Authors: 任泽岭, Binquan Li, Jing Liu, Kuang Li, Zhongmin Liang
• DOI: 10.1016/j.agsy.2026.104995

Research Groups

• State Key Laboratory of Water Cycle and Water Security, Hohai University, China
• College of Hydrology and Water Resources, Hohai University, China
• Key Laboratory of Hydrologic-Cycle and Hydrodynamic-System of Ministry of Water Resources, Hohai University, China
• State Key Laboratory of Water Cycle and Water Security, China Institute of Water Resources and Hydropower Research, China

Short Summary

The study develops a high-order polynomial surrogate model derived from AquaCrop to efficiently map monthly irrigation deficits to nonlinear crop yield losses, facilitating its integration into reservoir operation frameworks.

Objective

• To develop a computationally efficient agricultural surrogate model that accurately represents the nonlinear relationship between monthly water deficits and crop yield responses to replace oversimplified linear agricultural objectives in reservoir operation optimization.

Study Configuration

• Spatial Scale: Hetao Irrigation District, Upper Yellow River Basin, China.
• Temporal Scale: Monthly intervals throughout the growing season.

Methodology and Data

• Models used: AquaCrop (process-based model for generating training data), High-order polynomial ridge regression (surrogate model), and Standardized Precipitation Evapotranspiration Index (SPEI).
• Data sources: Simulated data generated via a locally calibrated AquaCrop model using a multi-round Latin Hypercube Sampling (LHS) method to cover the full 0–100% irrigation deficit gradient for spring wheat, spring maize, and sunflower.

Main Results

• Accuracy: The surrogate model achieved a coefficient of determination ($R^2$) greater than 0.93 across dual held-out test sets.
• Efficiency: Computational efficiency was increased by over five orders of magnitude compared to the original process-based model.
• Precision: The model maintained a yield-loss bias of less than 2%.
• Operational Impact: The application of surrogate-optimized schemes delayed the yield-loss threshold by 6.7–14.7 percentage points.

Contributions

• Provides a high-fidelity yet computationally lightweight agricultural loss function that allows for the direct embedding of complex crop yield responses into long-sequence iterative reservoir operation frameworks, overcoming the limitations of both computationally expensive process-based models and inaccurate linear approximations.

Funding

• Not specified in the provided text.

Citation

@article{任泽岭2026AquaCropderived,
  author = {任泽岭 and Li, Binquan and Liu, Jing and Li, Kuang and Liang, Zhongmin},
  title = {AquaCrop-derived surrogate model for crop yield response to monthly irrigation deficit},
  journal = {Agricultural Systems},
  year = {2026},
  doi = {10.1016/j.agsy.2026.104995},
  url = {https://doi.org/10.1016/j.agsy.2026.104995}
}