Forage quality calibration and validation of portable near infrared reflectance instruments using dry, ground forage samples

Authors: BERZAGHI, P - University Of Padua; CHERNEY, J - Cornell University; Casler, Michael

Submitted to: Computers and Electronics in Agriculture
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/22/2021
Publication Date: N/A
Citation: N/A

Interpretive Summary: Forage analysis by near infrared reflectance (NIR) spectroscopy has had many advancements since it began in the 1970’s. There have been steady improvements in instrumentation, computers, and computational methods for developing calibrations, making NIR the most used technique to routinely analyze samples for forage producers, plant breeders, animal nutritionists, livestock producers, and feed companies. Routine use of hand-held and portable NIR instruments to analyze fresh forage samples would represent a huge labor and time saving advancement to the industry. This study compared three different portable instruments with a typical laboratory NIR instrument, using a wide range of alfalfa and grass samples. The most inexpensive hand-held device had the poorest performance, often unacceptable by industry standards. The two devices that had intermediate prices generally had anywhere from acceptable to equal performance compared to the standard laboratory NIR method, likely due to their use of a wider range of spectral wavelengths. These results have significant potential to support broader and more frequent use of hand-hand NIR devices for scientific uses as well as on-farm applications.

Technical Abstract: Forage analysis by near infrared reflectance (NIR) spectroscopy has had many advancements since it began in the 1970’s. There have been steady improvements in instrumentation, computers, and chemometric algorithms for developing calibrations, making NIR the most used technique to routinely analyze samples for forage producers, plant breeders. animal nutritionists and cattle farmers and feed companies. This study compared the performance of prediction of three different portable instruments compared to a typical laboratory NIR instrument, using a wide range of alfalfa and grass dried, ground forage samples. Laboratory instrument scans were replicated with a reduced spectral range to match the range of each portable instrument. Alfalfa tended to have better calibration and validation statistics than the grasses in this study. Portable instruments evaluated did not scan the upper portion of the spectral range (1652 – 2498mn), which had some negative impact on forage calibration. The SCiO instrument scanned a very narrow range (740 – 1070nm) and although it had comparable results to the laboratory instrument constrained to same wavelength range, most major peaks related to forage quality parameters are outside this range. The expensive laboratory instrument had the best performance as expected, while the very inexpensive SCiO instrument had much greater error of predictions to the point that, for most traits, the prediction would not be considered reliable. However, the AuroraNir and the NIR-S-G1 digital light processing NIR may provide an alternative to expensive laboratory equipment, while still providing sufficiently accurate predictions. This study was conducted using dried and ground forage samples under typical laboratory conditions. Some portable instruments have the potential to be used for on-farm analysis of wet, coarsely chopped forage, and this option must be evaluated in future studies.