Discriminating stocking rates in a typical grassland using in situ spectral reflectance data

We investigated whether vegetation characteristics under different grazing intensities (measured by stocking rates, or SRs) could be discriminated using ASD field spectrometer data ranging from 350nm to 2,500 nm (a total of 2,151 wavelengths). Canopy spectral measurements and above-ground biomass under four different stocking rates (SR-0 sheep/ha, SR-3, SR-6 and SR-9) were sampled in situ from experimental plots in the Xilingol Sino-German experiment station, Inner Mongolia, China. Three types of measurements, i.e., canopy reflectance, the 1^st derivative of reflectance, and two categories of normalized difference vegetation indexes (NDVIs, including broad wavelength NDVI and narrow wavelength NDVI) derived from canopy reflectance, were tested for their ability to discriminate different stocking rates. Results show that vegetation canopy reflectance in different wavelength regions has variations in terms of its sensitivity to stocking rates and that different wavelength regions should be combined together to obtain better separation of stocking rates. Exploratory analysis indicates that both broad wavelength NDVI and narrow wavelength NDVI show strong negative correlation with stocking rates and strong positive correlation with above-ground biomass, indicating the possibility of applying NDVI to map grazing intensity at larger scales using satellite remote sensing imagery.