Authors:D.P. K. Thulasidas, S. Buddhan, E. M. Muralidharan and R. C. Pandalai
Abstract:Fourier Transform Near-infrared (FT-NIR) spectroscopy was successfully used to predict the silica content of Ochlandra travancorica. FT-NIR spectra were collected from powdered samples of reed bamboo with a fibre-optic probe in diffuse reflectance mode. NIR PLSR model of silica from the cross validation was developed by using spectra of 40 samples where silica content was determined by wet lab procedures. The precision and accuracy of the NIR PLSR model were analysed by comparing the statistical data of cross validation, calibration and test set validation. A good regression coefficient (r2cv=0.96) was obtained from the cross validation with FT-NIR spectroscopy for prediction of the silica in O. travancorica. The robustness of the PLSR model was proved with the similar values of coefficient of determination (r2), RMSECV, RMSEE and RMSEP. The silica content of unknown samples was predicted using the PLSR model. The mean value of silica content varied from 3 to 5% among the populations from different geographic locations. However, there was no statistically significant difference in silica content between the populations.
Keywords:Silica content, ash, FT-NIR spectroscopy, PLSR model, Ochlandra travancorica