The development and validation of the calibration model for the VIS-NIR spectrometer used for the evaluation of deoxynivalenol content in wheat grain directly during combine harvest.

The study is an attempt to build the calibration model for the VIS-NIR spectrometer used for measuring deoxynivalenol (DON) concentration in winter wheat grain at the stage of combine harvesting. In order to achieve this goal during the harvest season of 2011 the research material of 14.5 thousand radiation absorption spectrums and 283 grain samples was collected in field conditions. The low concentration of deoxynivalenol in natural grain samples did not allow a construction of the calibration model. Therefore, the PLS models (R≈0.9; RMSECV≈3–4 mg·kg) were developed on the basis of 60 samples prepared in laboratory conditions. The model set, which consisted of samples with the DON concentration of 0-32 mg·kg, was additionally supplemented with the results obtained for 58 samples from organic farming. The models were validated on the basis of the spectrums recorded in field conditions. The accuracy of prediction of the DON concentration by means of the model constructed on the basis of both sample collections was higher, but it was still insufficient. The study was a part of the development project No. R12 0073 06 entitled "Development and validation of the technology for separation grain stream during cereals selective harvesting ", financed by the Polish Ministry of Science and Higher Education.