Exploring spectral discrimination of grass species in African rangelands

The spectral reflectance from eight species of rangeland grasses in the Mas ai Mara Nature Reserve, Kenya, was measured using a laboratory-based spectr ometer. There were statistically significant differences in the spectral re flectance between species-a result which is encouraging for future work on identifying, classifying, mapping and monitoring rangeland ecosystems from hyperspectral imagery. To date, hyperspectral imagery has been available only through airborne sca nners, but the European Space Agency and the United States' National Aerona utics and Space Administration (NASA) both plan satellite missions. The sec ond part of this paper describes the acquisition and analysis of hyperspect ral data (CASI) coincident with ground plots. In these plots, the mix of gr ass species varied from pure (monospecific) patches through to mixes of fou r to five different species. Evidence is presented indicating that some spe cies may be identified on the image, based on the laboratory-obtained spect ra.