Non-destructive measurement of soluble solids content of three melon cultivars using portable visible/near infrared spectroscopy.

In this study, a non-destructive method using visible/near infrared (Vis/NIR) spectroscopy was investigated to predict the soluble solids content (SSC) of intact melons (Cucumis melo L.) cv. 'Manao', 'Jinhongbao', 'Xizhoumi'. A set of 360 samples (120 melons of each cultivar) was used to develop the calibration model, and two location (stylar-end and equatorial locations) models were investigated independently. The samples' spectra were obtained by a portable Vis/NIR photo-diode array spectrometer operated in reflectance mode. Multiplicative scatter correction (MSC), first derivative and Savizky-Golay (SG) smoothing in turn were applied to the obtained spectra. The region from 750 to 950 nm was selected to develop NIR models combined with the partial least squares (PLS) regression method. The results indicated that the stylar-end of the intact melon was the proper location to evaluate the SSC in the intact melon due to its suitable and exclusive physiological structure. A competitive adaptive reweighted sampling (CARS) algorithm was used to select effective wavelengths. Results showed that the CARS algorithm had great potential for simplifying the variables. Furthermore, another 195 samples were used for external prediction to evaluate the CARS-PLS model's accuracy and stability, which resulted in a high determination coefficient (R 2 p = 0.83) and a low root mean square error (RMSEP = 0.73 ºBrix). Image 1 • A universal NIR model for SSC testing among three cultivars was examined. • Some spectral pre-treatments were helpful to develop a universal model. • Stylar-end proved more appropriate to predict SSC of three cultivars melons. • CARS algorithm reliable for effective wavelengths selection for universal PLS model. [ABSTRACT FROM AUTHOR]