Tang et al. (2025) Reproductive stage superiority in irrigation scheduling: UAV spectral mechanisms validated by field canopy architecture for soybean yield prediction

Identification

• Journal: Field Crops Research
• Year: 2025
• Date: 2025-11-13
• Authors: Zijun Tang, Youzhen Xiang, Junsheng Lu, Tao Sun, Wangyang Li, Xueyan Zhang, Zhijun Li, Fucang Zhang
• DOI: 10.1016/j.fcr.2025.110230

Research Groups

• Key Laboratory of Agricultural Soil and Water Engineering in Arid and Semiarid Areas of Ministry of Education, Northwest A&F University, Yangling, China
• Xinjiang Research Institute of Agriculture in Arid Areas, Urumqi, Xinjiang, China
• College of Natural Resource and Environment, Northwest A&F University, Yangling, Shaanxi, China
• Department of Hydrology and Atmospheric Sciences, The University of Arizona, Tucson, AZ, USA

Short Summary

This study aimed to improve soybean yield prediction accuracy for precision irrigation management using UAV multispectral imaging. It found that the full pod stage (R4) is the most sensitive window for prediction, and a multi-source data fusion framework with XGBoost achieved high accuracy (R² = 0.83) by combining vegetation indices, texture features, and texture indices.

Objective

• To systematically analyze the sensitivity of soybean seed yield to various physiological and growth indices measured at different phenological stages using UAV multispectral imaging, aiming to improve yield prediction accuracy for precision irrigation management.

Study Configuration

• Spatial Scale: Field-scale experiments in northwestern China.
• Temporal Scale: Two consecutive growing seasons (2021–2022).

Methodology and Data

• Models used: Extreme gradient boosting (XGBoost) algorithm.
• Data sources: Unmanned aerial vehicle (UAV)-based multispectral imaging, field canopy architecture measurements.

Main Results

• The full pod stage (R4) was identified as the most sensitive phenological window for soybean yield prediction, coinciding with peak canopy cover and chlorophyll content.
• The ratio texture index (RTI, defined as DIS1/HOM3) showed the strongest correlation with yield (R = 0.69) among individual features.
• A three-source data fusion framework combining vegetation indices (VIs), texture features (TFs), and texture indices (TIs) with an XGBoost algorithm achieved optimal yield prediction performance at the R4 stage.
• The integrated model yielded a coefficient of determination (R²) of 0.83, a root mean square error (RMSE) of 280.80 kg ha⁻¹, and a mean relative error (MRE) of 6.32 % on the validation set.
• This multi-source XGBoost model improved R² by 31.7 % and reduced RMSE by up to 17.5 % compared to a model based solely on spectral VIs (R² = 0.63).

Contributions

• Provides a theoretical basis for precise field management in arid areas.
• Establishes a technical framework for remote sensing monitoring of crop yield, particularly highlighting the importance of reproductive stages and multi-source data fusion for enhanced prediction accuracy.

Funding

• Not specified in the provided text.

Citation

@article{Tang2025Reproductive,
  author = {Tang, Zijun and Xiang, Youzhen and Lu, Junsheng and Sun, Tao and Li, Wangyang and Zhang, Xueyan and Li, Zhijun and Zhang, Fucang},
  title = {Reproductive stage superiority in irrigation scheduling: UAV spectral mechanisms validated by field canopy architecture for soybean yield prediction},
  journal = {Field Crops Research},
  year = {2025},
  doi = {10.1016/j.fcr.2025.110230},
  url = {https://doi.org/10.1016/j.fcr.2025.110230}
}