Abstract
A database of 352 near infrared (NIR) spectral data from Iberian pig liquid fat was used for the prediction of palmitic, stearic, oleic and linoleic fatty acids content. The results showed the excellent accuracy of NIR equations with coefficients of determination (r2) reaching values close to 1 and standard error of cross-validation (SECV) values similar to the standard error in the reference method (SEL). The possibility of transferring the calibrations developed in one instrument to another NIR instrument, following the Shenk and Westerhaus standardisation algorithm, was examined. Two monochromator instruments were optically matched using four different standardisation sets of unsealed samples (Set A: one sample of Iberian pig liquid fat with a similar fatty acids composition to the average of the calibration set; Set B: 20 samples of Iberian pig liquid fat covering the range in optical density and fatty acid composition of the calibration set; Set C: oleic acid; Set D: linoleic acid. The standardisation completed with Set A and Set B produced spectra on the satellite instrument which resembled spectra scanned on the master instrument. The standard error of the differences (SED) between the values predicted by the master and the satellite after standardisation were similar to, or even lower than, the standard error of prediction (SEP) values for the same samples scanned on the master instrument.
Get full access to this article
View all access options for this article.
