Non-Destructive Assessment of Table Grape Quality Properties Using FTIR Spectroscopy and Chemometrics

This study establishes Fourier Transform Infrared (FTIR) spectroscopy coupled with partial least squares regression (PLSR) as a robust, non-destructive method for predicting key table grape quality properties: total soluble solids (TSS), titratable acidity (TA), TSS/TA ratio, dry matter, firmness, and vitamin C. The performance of three preprocessing strategies (raw spectra, Savitzky-Golay smoothing, and multiplicative scatter correction) was systematically compared. Notably, PLSR models built from raw spectra consistently outperformed preprocessed ones, achieving exceptional accuracy for all properties. Prediction results (R²pred; RMSEpred) were as follows: TSS (0.89; 1.08 °Brix), TA (0.90; 0.14%), TSS/TA (0.94; 0.33), dry matter (0.91; 0.88%), firmness (0.90; 0.14 N/mm²), and vitamin C (0.91; 0.57 mg/100 g). This performance surpasses many previous NIR/FTIR studies, particularly for vitamin C, which has historically been difficult to predict. Principal component analysis (PCA) indicated that the first two components explained 24–30% of the spectral variance, reflecting the complexity yet discriminative power of FTIR fingerprints. This work demonstrates that minimal spectral preprocessing can unlock the full potential of FTIR-PLSR for rapid, accurate table grape quality assessment, paving the way for its integration into real-time grading and precision viticulture systems.