Long-term prediction of Zhonghua kiwifruit dry matter by near infrared spectroscopy

Synergy interval partial least square (siPLS) was proposed to select efficiently the characteristic wavelength regions of dry matter against kiwifruit near-infrared spectra for dry matter prediction. Four data sets (NIR spectra and dry matter of unripe fruit (UU), NIR spectra of unripe fruit and dry matter of ripe fruit (UR), NIR spectra and dry matter of ripe fruit (RR), and UU&UR&RR) were obtained in the experiment. They were used to develop models for predicting dry matter of unripe and/or ripe kiwifruits. The results of cross-validation showed that the change of characteristic wavelength regions was caused by chemical conversion of organic compounds included in the dry matter at different storage periods of kiwifruits. Compared with the global spectra data models, the siPLS method could simplify the models with efficiently selecting characteristic wavelength regions. The root mean square error of cross-validation and correlation coefficient (r) of the UR model were 0.47% and 0.92, respectively, in calibration set. The root mean square error of prediction and r were 0.53% and 0.90, respectively, in the prediction set. This study demonstrated that NIR spectroscopy of unripe kiwifruits could predict the dry matter.