Vis/NIR soluble solids prediction in intact oranges (Citrus sinensis L.) cv. Valencia Late by reflectance

A commercially available instrument of near infrared spectroscopy, equipped with a post-dispersive reflectance configuration, was used for the development of partial least squares calibration models for soluble solids content and total acidity prediction in oranges (Citrus sinensis L.) cv. Valencia Late. Both internal validation and external validation were conducted and modified partial least squares regression was also tested. The best model for soluble solids content had 1-VR = 0.85, Rc2=0.91 and SECV = 0.51 in the internal validation, showing RMSEP = 0.51 in the external validation exercise. For TA and pH, the accuracy of the models developed was unacceptable.