Prediction of fermentation parameters in grass and corn silage by near infrared spectroscopy

Prediction models were developed for determination of lactic acid (Lac), acetic acid (HAc), pH, NH(3) N, and ethanol (EtOH) in grass and corn silages by near infrared (NIR) spectroscopy. Generally, Lac, pH, and NH(3) N concentrations in original sample material could be determined with a better accuracy when NIR measurements were performed on dried material rather than on wet material. The difference could only partly be explained by the repeatability error of NIR measurements for compounds stable with respect to drying. The HAc concentration was determined slightly more accurately when measurements were performed on wet material. The respective ratios between analytical concentration range expressed as standard deviation and the root mean square error of cross validation (RMSECV) were 4.9, 2.0, 3.7, and 3.1 for Lac, HAc, pH, and NH(3) N measured on dried grass silage, and 3.3 for EtOH measured on wet grass silage. The corresponding standard deviation:RMSECV for corn silage were 4.7, 1.9, 2.4, 2.9, and 4.0. The content of Lac was affected only slightly by drying at 80 degrees C, whereas the effects on NH(3) N and HAc were more pronounced, depending on silage type.