Determination of green herbage ratio in grasslands using spectral reflectance: methods and ground measurements

In this study, the suitability of spectral vegetation indexes for predicting green ratio (the percentage of green biomass with respect to the total phytomass) has been tested with the Italian Alps and New Zealand South Island grasslands. Considering three different datasets, green ratio (GR) was found to be negatively correlated with visible bands, while it was positively correlated in the NIR region (in total, R>0.80 in the 745–950 nm interval). GR proved to be more predictable than biomass and phytomass, both using hyperspectral single narrow bands and band ratios. Considering three different datasets, GR–index correlations were found to be linear and did not involve saturation problems. Many vegetation indices have been tested; they were well correlated with GR. Green Normalized Difference Vegetation Index (NDVIgreen) was the most stable index, with high values of R2 in all areas, low standard deviation and without significant differences in the slopes and intercepts of the linear correlations of the three datasets