Assessment of forage quality in diverse pastures by sensing spectral reflection and height of swards

摘要

Forage quality of grassland swards is a key issue for animal nutrition. Its rapid change in time requires a permanent monitoring to support adequate management decisions. In the present study a field spectrometer (305-1,690 nm) and an ultrasonic distance sensor was applied, able to conduct undisturbed and repeated measurements under field conditions. To evaluate their potential for predicting crude protein (CP) and acid detergent fibre (ADF) content in above-ground biomass, the study was conducted ongrassland pasture plots with varying grazing intensities. The potential of two different sensors and the capability of a combination of the sensor data were evaluated against the background of sward structural and functional complexity. Highest prediction accuracies were achieved with hyperspectral calibrations using a partial least square regression method (range of Revalues: 0.64-0.85 for CP and 0.63-0.75 for ADF). A selection of best fit 2-band normalized difference spectral index (NDSI) from hyperspectral data reduced prediction accuracy, but could be improved by inclusion of ultrasonic sward height. The combination of spectral and distance sensor data mostly achieved or could even exceed calibration accuracies of exclusive hyperspectral data. Thus, sensor fusion allows a feasible option to assess forage quality under field conditions.