| Summary: | The determination of wheat hardness by the evaluation of whole wheat grain would be of considerable value to the UK Milling Industry. Until now, accurate whole wheat grain hardness predictions by NIR spectroscopy have only been reported for North American wheats. By the evaluation of selected samples of UK and North American wheats this study showed that the prediction of whole wheat grain hardness by NTR spectroscopy depends only on the scattering properties of the sample and that there is no direct relationship with chemical composition. The scattering effect, in case of whole wheat grain reflectance and transmittance spectra, was found not to be multiplicative as in the case of ground wheat grain spectra. Empirical NIR spectroscopy calibrations are often performed without knowing what is measured or understanding the basis of the measurement. In other words the NIR spectrophotometer is often used as a "black box". Empirical calibrations were performed using three different software packages i.e. lnfrasoft International (ISI) Software, NIRSystems Spectral Analysis Software (NSAS) and UNSCRAMBLER. Successful NIR spectroscopy hardness measurements on ground wheat are based on light scattering. Separating the scattering effect from whole wheat grain spectra mathematically allowed predictions not significantly different to empirical calibrations, with the benefit of a theoretical explanation and fewer terms used. Although hardness predictions for whole wheat grain were not as accurate as in the case of ground wheat grain, it did prove to predict hardness with an acceptable accuracy with practical use as screening methods for grain trading. This study did not completely solve the problem of predicting whole wheat grain hardness by NIR spectroscopy, but new insights were provided which would hopefully encourage further work in this area and lead to a more complete fundamental understanding of the properties of whole wheat grain hardness using NIR spectroscopy.
|
|---|
