Stručni rad
https://doi.org/10.33128/k.61.2.6
Developing near-infrared (NIR) spectroscopy calibration for the prediction of forage nutritive value
Marina Vranić
; Agronomski fakultet Sveučilišta u Zagrebu, Zagreb, Hrvatska
Krešimir Bošnjak
; Agronomski fakultet Sveučilišta u Zagrebu, Zagreb, Hrvatska
Siniša Glavanović
; Belupo Inc., Istraživanje i razvoj, Koprivnica, Hrvatska
Tomislav Mašek
; Veterinarski fakultet, Sveučilišta u Zagrebu, Zagreb, Hrvatska
Sažetak
The aim of this article is to present the basics of the development of calibration models for the estimation of the physical, chemical and / or biological parameters of forage by NIR spectroscopy. Compared to conventional chemical methods, NIR spectroscopy is a multianalytical, environmentally friendly and fast method, easy to apply and interpret results. For the use of NIR spectroscopy for the prediciton of the nutritional value of forages, developed calibration equations are necessary. The samples used to develop applicable calibrations must be similar enough to belong to the same calibration set but also sufficiently different to cover the expected physical, chemical and / or biological variability of samples of unknown chemical composition. Calibration development consists of a range of activities from collecting / selecting samples and analyzing them by classical chemical / physical and / or biological methods to pre-processing and scanning, developing a calibration equation and finally applying and controlling the developed calibration prediction model. A developed calibration with a degree of correlation (R2) closer to 1.00, a low standard error of calibration (SEC), a standard error of cross-validation (SECV), and a standard error of prediction (SEP) can be applied in practice. Further research is focused on the use of mobile NIR spectroscopy for daily on-farm application.
Ključne riječi
forage; NIR spectroscopy; developing calibration
Hrčak ID:
243382
URI
Datum izdavanja:
20.8.2020.
Posjeta: 1.924 *