Estimating Leaf Nitrogen Accumulation Considering Vertical Heterogeneity Using Multiangular Unmanned Aerial Vehicle...

Area of Science:

• Agricultural Remote Sensing
• Plant Physiology
• Precision Agriculture

Background:

• Leaf nitrogen accumulation (LNA) is crucial for crop yield and quality.
• Vertical nitrogen heterogeneity in crop canopies complicates accurate LNA estimation.
• Unmanned Aerial Vehicle (UAV) and hyperspectral data offer potential for non-destructive LNA assessment.

Purpose of the Study:

• To develop and validate models for estimating wheat LNA that account for vertical canopy heterogeneity.
• To compare the accuracy of models incorporating vertical information against those that do not.
• To identify the optimal data acquisition strategy and modeling approach for LNA estimation.

Main Methods:

• Collected UAV multispectral and near-ground hyperspectral data at various view zenith angles.
• Divided wheat canopies into upper, middle, and lower layers to quantify LNA per layer.
• Developed linear regression (LR) and random forest regression (RF) models, including those considering vertical heterogeneity (LNA_Sum) and those that do not (LNA_non).

Main Results:

• Models incorporating vertical heterogeneity (LNA_Sum) significantly improved LNA estimation accuracy compared to LNA_non models.
• The optimal estimation scheme involved combining upper, middle, and lower leaf layers using the normalized difference red edge index.
• The RF-LNA_Sum model achieved the highest accuracy, with validation relative root mean square errors of 19.3% (UAV-measured) and 17.8% (simulated data).

Conclusions:

• Accounting for vertical canopy nitrogen distribution is essential for accurate LNA estimation using remote sensing.
• UAV-based hyperspectral and multispectral data, when analyzed with models considering vertical heterogeneity, provide a robust method for LNA assessment.
• The developed RF-LNA_Sum approach offers a promising tool for precision nitrogen management in wheat production.