Effect of spectrum measurement position variation on the robustness of NIR spectroscopy models for soluble solids content of apple

In this paper, the influence of variation of spectrum measurement position on the near infrared (NIR) spectroscopy analysis of soluble solids content (SSC) of apple was studied. The spectra were collected around stem, equator and calyx positions for each apple. Partial least squares (PLS) was used to develop compensation models of SSC for each measurement position separately (local position models) and for the full data set containing all positions (global position model). The results indicated that the influence of measurement position on the spectra affected the prediction accuracy of SSC. Compared with the local position models, the global position model was well suited to control the prediction accuracy of the calibration model for SSC with respect to the variation of spectrum measurement position. Next, competitive adaptive reweighted sampling (CARS) was used for the robust global position model to select the most effective wavelengths (EWs). It indicated that the global model established with effective wavelengths (EWs-global position model) achieved more promising results, with r(p) and RMSEP values for three measurement positions being 0.977, 0.977, 0.955 and 0.409, 0.386, 0.486 degrees Brix, respectively. Moreover, the local position models based on these effective variables (EWs-local position models) were more accurate than the models built with full range spectrum. The overall results indicated that the EWs-global position model could make the variation of spectrum measurement position a negligible interference for SSC prediction. (C) 2015 IAgrE. Published by Elsevier Ltd. All rights reserved.