Description and Evaluation of a near Infrared Reflectance Spectro-Computer for Forage and Grain Analysis1
- J. S. Shenk,
- I. Landa,
- M. R. Hoover and
- M. O. Westerhaus2
Near infrared (NIR) reflectance has been proposed as a procedure to determine the nutritive value of forages and feedstuffs. A Neotec model 6100 scanning monochromator interfaced to a PDP-11 minicomputer was used for this study. Instruments of this type are called spectro-computers. Ten computer programs developed at University Park, Penn. were used to operate and test the instrument. The scanning range of the spectro-computer was from 1,100 to 2,500 nm. Wavelength accuracy was ± 1.0nm, wavelength precision ± 0.01 nm, stray light <0.10%, and system peak to peak and root mean square noises were 0.33 × 10-3 and 0.05 × 10-3, respectively, expressed as log (1/R). This high level of monochromator performance permits the spectro-computer precision to compare favorably with that of the standard laboratory procedures.
Two hundred forage samples were used to test the accuracy of the spectro-computer system. These samples represented a broad array of species and mixtures, stages of maturity, and harvest locations. Protein and in vitro dry matter disappearance were predicted best with NIR data treated as log of 1/R, resulting in standard errors of prediction of 0.96 and 3.18%, respectively. Other quality parameters were predicted better when the log (1/R) NIR data were transformed to the second derivative. They were acid detergent fiber ± 1.99%, neutral detergent fiber ± 2.27%, lignin ± 1.13%, cellulose ± 1.27%, Ca ± 0.16%, P ± 0.04%, and K ± 0.37%.Further study of the prediction of protein in 90 Canadian wheat (Triticum aestivum L.) samples resulted in a prediction error of 0.17%. We believe this spectro-computer system is suitable for use in forage and grain analysis.Please view the pdf by using the Full Text (PDF) link under 'View' to the left.
Copyright © . .