Tracing pistachio nuts' origin and irrigation practices through hyperspectral imaging.

Area of Science:

• Agricultural Science
• Spectroscopy
• Data Science

Background:

• Pistachio nuts are a high-demand global commodity due to their flavor and health benefits.
• Accurate assessment of pistachio origin, quality, and yield is crucial for agricultural optimization.

Purpose of the Study:

• To apply Hyperspectral Imaging (HSI) and Machine Learning (ML) for pistachio analysis.
• To determine pistachio geographic origin, irrigation practices, and predict quality and yield parameters.

Main Methods:

• Utilized HSI to capture spectral data from pistachios in Spanish orchards.
• Employed ML models including Partial Least Squares (PLS), Support Vector Machine (SVM), and Extreme Gradient Boosting (XGBoost).
• Analyzed spectral signatures for classification and regression tasks.

Main Results:

• Achieved >94% accuracy in classifying pistachio origin.
• Reached 99% accuracy in assessing water content and color pigments using PLS and SVM.
• Demonstrated high accuracy (92% with PLS) in identifying spectral signatures related to irrigation treatments.
• Successfully predicted yield (R²=0.89 with PLS) and blank nuts (R²=0.71 with PLS).

Conclusions:

• HSI and ML are effective tools for pistachio origin determination and quality assessment.
• Spectral analysis shows potential for optimizing pistachio production, sustainability, and yield forecasting.
• Distinct spectral signatures correlate with irrigation practices and nut quality parameters.