Abstract
Near-infrared spectroscopy (NIRS) calibration models for determining grain constituent contents at grain elevators were developed using rice and wheat grown in Hokkaido. The coefficient of determination (r2), standard error of prediction (SEP) and ratio of standard error of prediction to standard deviation (RPD) of the validation statistics were r2 > 0.99, SEP = 0.19 % and RPD = 25.2 for brown rice moisture content, r2 = 0.99, SEP = 0.09 % and RPD = 9.8 for milled rice protein content, r2 > 0.99, SEP = 0.34 % and RPD = 23.1 for wheat moisture content and r2 = 0.97, SEP = 0.31 % and RPD = 5.4 for wheat protein content. These results indicated that the accuracy of NIRS was sufficiently high for practical use at grain elevators.
