Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

NIR reflectance and interactance modes were used as rapid techniques to predict the glucose and sucrose ratio in two common baking and chipping industry potato varieties (Russet Norkotah, and Frito Lay) for the growing/storage season of 2009-2010. Three different storage temperatures were used (4, 7, and 10 Celsius) to develop uniform and broad constituent distributions. Both glucose and sucrose ratio were measured using the Megazyme sucrose/D-glucose assay procedure as reference values. Partial least squares regression (PLSR) and radial basis functions artificial neural networks (RBFNN) methods were used for building prediction models. Strongly correlated models for glucose from the NIR interactance mode for sliced and whole tubers were obtained for Russet Norkotah with correlation coefficients of 0.92, 0.95 respectively, whereas those values were 0.75 and 0.83 for Frito Lay. For reflectance mode, R values were 0.88 and 0.92 for Russet Norkotah and 0.71 and 0.70 for Frito Lay. In general, the sucrose PLSR models showed poor correlation for Russet Norkotah and Frito Lay with the highest R obtained from the reflectance model for Russet Norkotah for sliced samples yielding an R value of 0.66. Using the RBFNN increased the efficiency of the sucrose model for Russet Norkotah yielding an R value of 0.96 in NIR reflectance for sliced samples. Similar R values were obtained for whole tubers in reflectance mode and VIS/NIR interactance modes for both sliced and whole samples. Data fusion from different systems was tried and results of both PLSR and RBFNN models showed no, or limited, increase in correlation.