A Study of Nitrogen Deficiency Inversion in Rice Leaves Based on the Hyperspectral Reflectance Differential

Area of Science:

• Agricultural Science
• Remote Sensing
• Plant Physiology

Background:

• Nitrogen deficiency significantly impacts rice yield and quality.
• Accurate and non-destructive methods for diagnosing nitrogen status are crucial for precision agriculture.
• Hyperspectral imaging offers potential for real-time plant health assessment.

Purpose of the Study:

• To develop a hyperspectral reflectance difference inversion model for rapid and accurate diagnosis of nitrogen deficiency in cold land japonica rice.
• To establish a method for precision fertilization that optimizes nitrogen application without compromising yield.
• To investigate the relationship between nitrogen content and spectral reflectance differences.

Main Methods:

• Collected hyperspectral data from field experiments on japonica rice.
• Utilized the principle of minimum fertilizer application at maximum yield to establish standard N content and spectral reflectance.
• Applied discrete wavelet multiscale decomposition, successive projections algorithm, principal component analysis, and iteratively retaining informative variables (IRIVs) for dimensionality reduction.
• Developed and compared inversion models including Partial Least Squares Regression (PLSR), Extreme Learning Machine (ELM), and Genetic Algorithm-Extreme Learning Machine (GA-ELM).

Main Results:

• The Genetic Algorithm-Extreme Learning Machine (GA-ELM) model, combined with discrete wavelet multi-scale decomposition, demonstrated optimal performance in data modeling and training.
• The GA-ELM model achieved a coefficient of determination (R²) above 0.68 for both training and validation datasets.
• The model exhibited root mean square errors (RMSEs) below 0.6 mg/g, indicating high predictive accuracy, stability, and generalizability compared to PLSR and ELM models.

Conclusions:

• The hyperspectral reflectance difference inversion model, particularly the GA-ELM approach, provides a reliable method for diagnosing nitrogen deficiency in japonica rice.
• This technique supports precision fertilization strategies, enabling efficient nitrogen management and potentially reducing fertilizer use.
• The study offers a valuable tool for enhancing rice cultivation practices through advanced spectral analysis.