Measuring pasture biomass with a spectroradiometer
The spatial variability of pasture biomass, and nitrogen content are important characteristics to monitor if feed supply to grazing animals is to be optimized. Diffuse spectral reflectance from cereal crops in the visible and near infrared range has been used elsewhere for the purpose of estimating crop biomass and nitrogen status. Variable cloud cover in temperate regions prevents successful employment of these techniques on crops and pastures. This study evaluates an ASD FieldSpec PRO modified (with an artificial light source) for the collection of weather-independent reflectance spectra from grazed dairy pastures.
During the first week of May 2006 reflectance measurements were taken, using an ASD FieldSpec PRO with illumination provided by a 50 Watt lamp with a tungsten-quartz-halogen bulb, from fifty four sites on dairy pastures. Simultaneous readings of grass height using a Fillip’s folding plate pasture meter were obtained to estimate the kg dry matter per hectare. The reflectance data was first smoothed, and then two spectral transformation methods were applied: derivatives and band depth normalization of continuum removed spectra (BDn). The statistical method used to explore relationships between the two datasets was partial least squares regression.
The best prediction of pasture biomass was obtained using the first derivative of the reflectance (R2 = 0.96, R2 predicted = 0.81 and RMSE = 117 kgDM/ha), followed by the reflectance (R2 = 0.93, R2 predicted = 0.75 and RMSE = 135 kgDM/ha) and the BDn (R2 = 0.72, R2 predicted = 0.59 and RMSE = 173 kgDM/ha). The potential to predict pasture N status is also discussed. The results obtained demonstrate the potential of the method to estimate biomass in kg dry matter per hectare; but indicate how the newly-modified instrument can be improved. The improved instrument is currently being thoroughly tested.