Feasibility of filter-based NIR spectroscopy for the routine measurement of olive oil fruit ripening indices

The aim of this work was to apply and validate a NIR spectrometer based on a discrete filter system for the rapid measurement of the moisture, oil, sugar, and phenolic compounds contents of olive oil fruits. The batches of olive oil fruits were collected during seven crop seasons from several farms located in Tuscany and then they were crushed into olive paste. The water content was measured gravimetrically, oil content was measured using the Soxhlet method, sugar content was measured enzymatically and phenolic compound content was measured by HPLC. NIR spectra were recorded from 1400 to 2400nm at 19 selected wavelengths. Calibration and validation models were processed using PLS regression. In PLS models built for moisture, oil, and sugar contents, the r2 in calibration ranged between 0.90 and 0.93 with low standard error of calibration (SEC) values (i.e. 2.5, 3.6 and 4.0, respectively). For these parameters the standard error of prediction (SEP) and the standard error of laboratory (SEL) values proved to be comparable (i.e. 2.4 vs. 3.2, 6.0 vs. 4.4, and 6.7 vs. 4.7, respectively). Instead, the calibration and validation results concerning the phenolic compounds were not satisfactory, probably because the necessary wavelengths in the section of absorbance from 1100 to 1400nm were not covered. Practical applications: NIR spectrometers based on discrete filter systems may be interesting since they are cost-saving compared to the more sophisticated FT-NIR and NIR-AOTF instruments (i.e. the cost is approximately half). Our study also showed that with this instrument it was possible to build some effective models for predicting moisture, oil, and sugar contents in olive paste. The results obtained for moisture and oil contents are comparable with those obtained with other spectrometers, and a predictive model was obtained for sugar content for the first time. Instead the tool did not prove suitable for obtaining predictive models for total phenolic compounds or oleuropein contents. The additional use of a filter-based NIR spectrometer is therefore to be suggested to rapidly monitor olive fruit ripening on the basis of moisture, oil, and sugar contents.