Li et al. (2026) Harnessing multimodal UAV data and ensemble learning to improve estimation accuracy of winter wheat water content

Identification

• Journal: Agricultural Water Management
• Year: 2026
• Date: 2026-08-06
• Authors: Xinyu Li, Xiaohai Fang, Xiaobo Gu, Zhikai Cheng, Penglin Li, Yang Xu, Gengmin Jiang, Yunhui Niu, Yadan Du
• DOI: 10.1016/j.agwat.2026.110681

Research Groups

• Key Laboratory of Agricultural Soil and Water Engineering in Arid and Semiarid Areas, Ministry of Education, Northwest A&F University, Yangling, Shaanxi 712100, China
• School of Information Engineering, Shaanxi Vocational College of Agriculture and Forestry, Yangling, Shaanxi 712100, China
• College of Civil Engineering and Architecture, Nanyang Normal University, Nanyang 473061, China
• Jinghui Irrigation and Drainage Management Station, Lintong District, Xi'an, Shaanxi 710000, China

Short Summary

This study employed multimodal UAV data and ensemble learning to improve the estimation accuracy of winter wheat water content. The results demonstrated that multimodal data fusion significantly enhanced the accuracy of PWC estimation.

Objective

• Evaluate the potential of UAV-derived multimodal data for field-scale PWC estimation
• Compare ensemble learning against conventional machine learning models to identify the most accurate and stable estimation approach

Study Configuration

• Spatial Scale: Field scale, with a study area of 72 plots across 24 treatments
• Temporal Scale: Winter wheat growth period (2024–2025)

Methodology and Data

• Models used:
  • Single machine learning algorithms: partial least squares regression (PLSR), ridge regression (RR), support vector regression (SVR)
  • Bagging-based ensemble models: random forest (RF), extra trees (ET), bagged decision tree (BDT)
  • Boosting-based ensemble models: eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), Categorical Boosting (CatBoost)
  • Fusion ensemble models: Stacking, Averaging, Bagged-Boosting
• Data sources: Multispectral and thermal infrared imagery from UAVs

Main Results

• The results demonstrated that multimodal data fusion significantly enhanced the accuracy of PWC estimation.
• Ensemble learning outperformed single machine learning algorithms in terms of R2 values (0.781–0.831) and RMSE (0.017–0.020 g⋅g−1).
• Among the ensemble models, Averaging achieved the best performance under multimodal conditions (R2 = 0.831, RMSE = 0.017 g⋅g−1).

Contributions

• This study contributes to the development of a reliable and robust method for estimating winter wheat PWC at the field scale.
• The results provide insights into the potential of UAV-derived multimodal data for improving crop water status estimation.

Funding

• This research was supported by the Key Laboratory of Agricultural Soil and Water Engineering in Arid and Semiarid Areas, Ministry of Education, Northwest A&F University.

Citation

@article{Li2026Harnessing,
  author = {Li, Xinyu and Fang, Xiaohai and Gu, Xiaobo and Cheng, Zhikai and Li, Penglin and Xu, Yang and Jiang, Gengmin and Niu, Yunhui and Du, Yadan},
  title = {Harnessing multimodal UAV data and ensemble learning to improve estimation accuracy of winter wheat water content},
  journal = {Agricultural Water Management},
  year = {2026},
  doi = {10.1016/j.agwat.2026.110681},
  url = {https://doi.org/10.1016/j.agwat.2026.110681}
}