Summary
The benefits of F.NIRS technology to industry and research depend on accurate predictions. The performance of F.NIRS calibration equations for predicting faecal N concentration and faecal 13C was monitored over 3 years using selected faecal samples covering a wide range of pasture types, geographical locations and reference values. The performance of the faecal N calibration was considered to be excellent for each of four validation sample sets with a combined total of 434 samples. The performance of the faecal 13C calibration was variable, being most satisfactory for two of the validation sample sets (total of 125 samples) but much less satisfactory for the other two validation sets. The poorer validation statistics were due primarily to a relatively high proportion of faecal samples from pasture types and/or regions not represented in the calibration and these are detailed in the report. Validation sample spectra (434 for faecal N and 393 for faecal 13C) with matching reference values were added to the existing calibration sets and new calibration equations with enhanced robustness were developed from the expanded sets. The results of this project indicated that similar work focusing on F.NIRS calibration equations for predicting diet quality deserves high priority.
