Enhancing Nitrogen Nutrition Index estimation in rice using multi-leaf SPAD values and machine learning approaches.

Area of Science:

• Agricultural Science
• Plant Science
• Data Science

Background:

• Optimizing nitrogen fertilization is crucial for rice yield and environmental sustainability.
• Traditional nitrogen diagnostics can be labor-intensive and less precise.
• Developing efficient tools for in-season nitrogen assessment is vital for modern agriculture.

Purpose of the Study:

• To evaluate the efficacy of multi-leaf SPAD measurements combined with machine learning for improving nitrogen nutrition diagnostics in rice.
• To identify key leaf positions and statistical metrics that enhance prediction accuracy for Leaf Nitrogen Concentration (LNC) and Nitrogen Nutrition Index (NNI).

Main Methods:

• Collected SPAD values from the first to fifth fully expanded leaves across five locations and 15 rice cultivars at critical growth stages.
• Employed machine learning models, including Random Forest and Extreme Gradient Boosting, for LNC and NNI estimation.
• Incorporated statistical metrics (e.g., maximum, median SPAD values) alongside original SPAD data.

Main Results:

• Multi-leaf SPAD data integrated with machine learning significantly improved LNC and NNI estimation accuracy.
• The second fully expanded Leaf From the Top (2LFT) was the most critical predictor for LNC.
• The third fully expanded Leaf From the Top (3LFT) was pivotal for NNI estimation.
• Statistical SPAD metrics further enhanced predictive model performance.

Conclusions:

• Combining multi-leaf SPAD data with advanced machine learning offers a precise and effective method for rice nitrogen assessment.
• This approach supports enhanced nitrogen use efficiency and promotes sustainable rice cultivation through targeted management.
• The findings provide a foundation for developing practical tools for real-time nitrogen status monitoring in rice.