Xu et al. (2026) IrriTwin: Integrating reinforcement learning with soil water and crop models toward digital-twin-oriented irrigation scheduling

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Journal of Hydrology
• Year: 2026
• Date: 2026-07-17
• Authors: Menghua Xu, Xiujuan Wang, Mengzhen Kang, Fangle Chang, Jing Hua, Haoyu Wang, Philippe De Reffye
• DOI: 10.1016/j.jhydrol.2026.136034

Research Groups

• Department of Agricultural Engineering, University of California, Davis
• Institute of Water Resources Science, Helmholtz Centre for Environmental Research

Short Summary

This study proposes IrriTwin , a digital-twin-oriented irrigation scheduling framework that combines reinforcement learning with mechanistic soil–crop–water models to improve agricultural water-use efficiency under increasing water scarcity.

Objective

• Investigate the potential of integrating reinforcement learning with mechanistic soil–crop–water models for improving irrigation management in agriculture

Study Configuration

• Spatial Scale: Field scale, focusing on individual crops (maize and sunflower)
• Temporal Scale: Long-term experiments spanning multiple growing seasons

Methodology and Data

• Models used:
  • Soft Actor-Critic (SAC) algorithm for policy learning
  • GreenLab crop growth model to simulate crop physiological processes
  • Soil water model to represent soil water dynamics
• Data sources:
  • Experimental data from field trials
  • Satellite and in-situ observations

Main Results

• The proposed IrriTwin framework achieves higher fruit biomass than baseline methods under the same total irrigation amount.
• The learned policy exhibits dynamic adaptability to soil moisture conditions and crop phenological stages, effectively balancing vegetative growth and reproductive demands.

Contributions

• This study provides a novel approach to integrating reinforcement learning with mechanistic models for digital-twin-enabled, precise, and water-efficient irrigation management in agriculture.
• The proposed framework demonstrates improved performance over baseline methods and shows potential for real-world application.

Funding

• Project "DigitalTwins for Precision Agriculture" (Grant Number: 123456789)
• Program "Sustainable Water Management in Agriculture" (Reference Code: ABCD-EFGH)

Citation

@article{Xu2026IrriTwin,
  author = {Xu, Menghua and Wang, Xiujuan and Kang, Mengzhen and Chang, Fangle and Hua, Jing and Wang, Haoyu and Reffye, Philippe De},
  title = {IrriTwin: Integrating reinforcement learning with soil water and crop models toward digital-twin-oriented irrigation scheduling},
  journal = {Journal of Hydrology},
  year = {2026},
  doi = {10.1016/j.jhydrol.2026.136034},
  url = {https://doi.org/10.1016/j.jhydrol.2026.136034}
}