Oil and water analysis of sunflower seed by near-infrared reflectance spectroscopy

The applicability of NIR for oil and moisture analyses of sunflower seed was determined using a NIR spectrocomputer system. The method was compared with the wide-line NMR method for oil analysis and with the A.O.C.S. oven method for moisture analysis. The NIR was calibrated with 120 samples for oil (96 for calibration, 24 for prediction) and 63 samples for moisture (55 for calibration, 8 for prediction). Twenty-two sunflower seed samples were analyzed for oil and moisture by NIR and by methods used by industry. The oil contents of the samples by NMR and NIR were not significantly different. The overall mean oil contents and mean of the standard deviations for the samples were: NMR, 44.2%±0.35% and NIR, 44.34%±0.74%. A significant difference was found between the moisture values obtained by the oven-drying method and NIR. The average standard deviation for moisture by NIR was 0.57% compared with 0.07% for the oven-drying method. The variability of the oil content in one of the commercial seed samples was 1.52% oil as determined by NMR and 2.52% as determined by NIR. The advantages and disadvantages of both methods are discussed.