Li et al. (2026) UAV-based multispectral inversion models for soil moisture content across the whole growth cycle of mung bean using SHAP interpretation

Identification

• Journal: Agricultural Water Management
• Year: 2026
• Date: 2026-08-19
• Authors: Yihui Li, Fushun Wang, Xinxin Wang, Renqiang Chen, Huiyan Gao, Xiaoling Wang, Tida Ge, Zhipeng Yue, Zehao Zhou, Hongquan Liu
• DOI: 10.1016/j.agwat.2026.110705

Research Groups

• College of Urban and Rural Construction, Hebei Agricultural University, Baoding, Hebei 071001, China
• College of Information Science and Technology, Hebei Agricultural University, Baoding, Hebei 071001, China
• Hebei Provincial Key Laboratory of Agricultural Big Data, Baoding, Hebei 071001, China
• Agricultural Technology Innovation Center in Mountainous Areas of Hebei Province, Baoding, Hebei 071001, China
• National Research Center for Agricultural Engineering Technology in Northern Mountainous Areas, Baoding, Hebei 071001, China
• North China Key Laboratory of Water Saving, Ministry of Agriculture and Rural Affairs, Baoding, Hebei 071001, China
• College of Horticulture, Hebei Agricultural University, Baoding, Hebei 071001, China
• State Key Laboratory for Quality and Safety of Agro-Products, Ningbo, Zhejiang 315211, China
• Key Laboratory of Biotechnology in Plant Protection of MARA, Ningbo, Zhejiang, 315211, China
• Zhejiang Key Laboratory of Green Plant Protection, Institute of Plant Virology, Ningbo University, Ningbo, Zhejiang 315211, China

Short Summary

This study proposes a synergistic optimization strategy for soil moisture monitoring in mung bean using UAV multispectral imagery at different flight altitudes and machine learning algorithms. The results show that the optimal model varies across growth stages and soil layers.

Objective

• Investigate the relationship between UAV multispectral imagery, flight altitude, and soil moisture content in mung bean fields.
• Develop a synergistic optimization strategy for soil moisture monitoring using machine learning algorithms.

Study Configuration

• Spatial Scale: Field-scale (8.65 m × 6.77 m)
• Temporal Scale: Entire growth period of mung bean (76 days)

Methodology and Data

• Models used:
  • Random Forest (RF)
  • Extreme Gradient Boosting (XGBoost)
  • Multiple Linear Regression (MLR)
  • Lasso
  • Ridge
  • Support Vector Regression (SVR)
  • Multi-Layer Perceptron (MLP)
• Data sources: UAV multispectral imagery, soil moisture content data collected using destructive sampling

Main Results

• The optimal model varies across growth stages and soil layers.
• At the seedling stage, the optimal model for the 0–20 cm layer was 50 m XGBoost (R2 = 0.674), and for 20–40 cm was 30 m RF (R2 = 0.709).
• At the branching stage, the 0–20 cm optimum was 50 m RF (R2 = 0.860), and the 20–40 cm optimum was 50 m XGBoost (R2 = 0.768).
• At the flowering stage, 30 m XGBoost was optimal for both 0–20 cm (R2 = 0.887) and 20–40 cm (R2 = 0.906).
• At the maturity stage, the 0–20 cm optimum was 50 m XGBoost (R2 = 0.838), and the 20–40 cm optimum was 30 m XGBoost (R2 = 0.654).

Contributions

• This study proposes a synergistic optimization strategy for soil moisture monitoring in mung bean using UAV multispectral imagery at different flight altitudes and machine learning algorithms.
• The results show that the optimal model varies across growth stages and soil layers, providing new insights into the relationship between UAV multispectral imagery, flight altitude, and soil moisture content.

Funding

• This research was funded by the National Natural Science Foundation of China (Grant No. 52177101).
• Hebei Provincial Key Laboratory of Agricultural Big Data (Grant No. HBKLAB2021001).

Citation

@article{Li2026UAVbased,
  author = {Li, Yihui and Wang, Fushun and Wang, Xinxin and Chen, Renqiang and Gao, Huiyan and Wang, Xiaoling and Ge, Tida and Yue, Zhipeng and Zhou, Zehao and Liu, Hongquan},
  title = {UAV-based multispectral inversion models for soil moisture content across the whole growth cycle of mung bean using SHAP interpretation},
  journal = {Agricultural Water Management},
  year = {2026},
  doi = {10.1016/j.agwat.2026.110705},
  url = {https://doi.org/10.1016/j.agwat.2026.110705}
}