Coupling decision of water and nitrogen application in winter wheat via UAV hyperspectral imaging

Context

Improving water and nutrient use efficiency is essential for increasing crop yields and addressing global population growth. Optimal irrigation and nitrogen topdressing levels can enhance crop water and nitrogen use efficiency. UAV remote sensing has emerged as an efficient tool for optimizing water and nitrogen management due to its ability to monitor crop traits in real-time.

Objective

This study proposed a UAV-based hyperspectral imaging method to optimize water-nitrogen management in winter wheat.

Methods

By analyzing the interaction between nitrogen fertilizer and irrigation, a coupling decision model was developed for precise water-nitrogen application. Leaf water content (LWC) and chlorophyll content (SPAD) were estimated using machine learning algorithms combined with sensitive band selection methods, such as Successive Projections Algorithm (SPA) and Competitive Adaptive Reweighted Sampling (CARS).

Results

The SPA-Random Forest (RF) model performed best for LWC estimation (R² = 0.83, RMSE = 5.39 %), while the VIs-RF model was optimal for SPAD estimation (R² = 0.65, RMSE = 4.34 %). Conversion models linked LWC to soil water content (SWC) and SPAD to leaf nitrogen content (LNC), achieving R² values of 0.79 and 0.78, respectively.

Conclusions

The proposed water-nitrogen coupling model exhibited strong adaptability and stability during key growth stages by integrating hyperspectral inversion data with field measurements. This model enables dynamic water and nitrogen application rate adjustments across the growing period to achieve target yields, optimize application strategies, and enhance use efficiency.

Implications

The findings underscore the significant potential of UAV-based hyperspectral technology in optimizing water-nitrogen management. This method provides a reference for improving water-nitrogen use efficiency from the perspective of water-nitrogen coupling on yield.