Ma et al. (2026) Assessing integrated water status in drip-irrigated maize fields using UAV multispectral data and machine learning algorithms

Identification

• Journal: Agricultural Water Management
• Year: 2026
• Date: 2026-06-23
• Authors: Mingjie Ma, Jinghua Zhao, Ming Hong, Tingrui Yang, Qiuping Fu
• DOI: 10.1016/j.agwat.2026.110554

Research Groups

• College of Hydraulic and Civil Engineering, Xinjiang Agricultural University
• Xinjiang Key Laboratory of Hydraulic Engineering Safety and Water Disaster Prevention

Short Summary

This study proposes a comprehensive water estimation model for monitoring the soil–vegetation system in drip-irrigated maize farmland. The model integrates UAV technology with machine learning algorithms to achieve dynamic monitoring at the field scale.

Objective

• To explore the potential of UAV multispectral data for monitoring water status in drip-irrigated maize fields.
• To investigate the performance improvement of models through optimization methods based on hyperparameter tuning and recursive feature elimination.
• To evaluate the potential of water status monitoring models for maize fields constructed by integrating VWC with soil moisture indicators across different layers.

Study Configuration

• Spatial Scale: Field-scale (33.33 ha)
• Temporal Scale: Four distinct growth stages of maize: jointing, tasseling, filling, and milking

Methodology and Data

• Models used: XGBoost, Random Forest, Support Vector Machine
• Data sources: UAV multispectral data, field sampling (soil moisture content and vegetation water content)

Main Results

• The comprehensive moisture index (CMI) was constructed by integrating plant water content and soil moisture status across different soil layers using the coefficient of variation and entropy weight methods.
• The SSA-RFE-XGBoost model achieved the highest accuracy in predicting CMICV, with R² values ranging from 0.541 to 0.714.

Contributions

• This study provides a technical solution that integrates UAV technology with machine learning algorithms to achieve dynamic monitoring at the field scale.
• The proposed model optimization method (SSA-RFE-XGBoost) outperformed other models in predicting CMICV and CMIEWM.

Funding

• National Natural Science Foundation of China (Grant No. 52109251)
• Xinjiang Uygur Autonomous Region Science and Technology Plan Project (Grant No. 2022AB01155)
• Xinjiang Agricultural University Research Fund (Grant No. XJAU2019K01)

Citation

@article{Ma2026Assessing,
  author = {Ma, Mingjie and Zhao, Jinghua and Hong, Ming and Yang, Tingrui and Fu, Qiuping},
  title = {Assessing integrated water status in drip-irrigated maize fields using UAV multispectral data and machine learning algorithms},
  journal = {Agricultural Water Management},
  year = {2026},
  doi = {10.1016/j.agwat.2026.110554},
  url = {https://doi.org/10.1016/j.agwat.2026.110554}
}