Zhao et al. (2026) Identification of key growth stages for cotton yield formation and SPAD response to irrigation thresholds using UAV multispectral data

Identification

• Journal: Agricultural Water Management
• Year: 2026
• Date: 2026-08-12
• Authors: Xin Zhao, Cong Shi, Haireati Gulibiya, Zhipeng Huang, Hang Wang, Wenqing He
• DOI: 10.1016/j.agwat.2026.110687

Research Groups

• Institute of Environment and Sustainable Development in Agriculture, Chinese Academy of Agricultural Sciences
• Institute of Western Agricultural, Chinese Academy of Agricultural Sciences
• Xinjiang Key Laboratory of Agricultural Green Materials and Soil Health

Short Summary

This study utilized UAV multispectral data to identify growth stages associated with cotton yield variation and evaluated the response of SPAD values to different lower-limit irrigation thresholds. The results indicate that a flowering-boll lower irrigation threshold range of 60–70% FC may provide a practical reference for balancing canopy physiological status, yield formation, and water-management performance.

Objective

• Identify key growth stages associated with cotton yield variation using UAV multispectral data.
• Evaluate the response of SPAD values to different lower-limit irrigation thresholds.

Study Configuration

• Spatial Scale: Field-scale experiment in Laolonghe Modern Agricultural Demonstration Zone, Xinjiang, China.
• Temporal Scale: Cotton growing season (2024).

Methodology and Data

• Models used:
  • Random Forest (RF) yield model
  • XGBoost
  • RFR
  • CNN-BiLSTM-Attention model for SPAD estimation
• Data sources:
  • UAV multispectral data
  • Ground-truth data (SPAD, above-ground biomass, plant height, stem diameter)

Main Results

• The RF yield model achieved moderate predictive performance (R² = 0.40, RMSE = 771.96 kg ha⁻¹, MAE = 671.31 kg ha⁻¹).
• SHAP analysis revealed that variables with high contributions were mainly concentrated during the flowering and boll-forming stages.
• The CNN-BiLSTM-Attention model performed best for SPAD estimation (R² = 0.89, RMSE = 0.03, RPD = 3.08).

Contributions

• This study provides a theoretical basis and technical support for the remote sensing monitoring of key physiological indicators and optimization of irrigation thresholds.
• The results indicate that UAV multispectral data can support the monitoring of cotton SPAD and irrigation threshold responses.

Funding

• This research was funded by the National Key Research and Development Program of China (Grant No. 2017YFD0300400).
• The authors acknowledge the financial support from the Chinese Academy of Agricultural Sciences (CAAS) and the Xinjiang Uygur Autonomous Region Science and Technology Department (Grant No. 2021AB01003).

Citation

@article{Zhao2026Identification,
  author = {Zhao, Xin and Shi, Cong and Gulibiya, Haireati and Huang, Zhipeng and Wang, Hang and He, Wenqing},
  title = {Identification of key growth stages for cotton yield formation and SPAD response to irrigation thresholds using UAV multispectral data},
  journal = {Agricultural Water Management},
  year = {2026},
  doi = {10.1016/j.agwat.2026.110687},
  url = {https://doi.org/10.1016/j.agwat.2026.110687}
}