Satellite imagery as a tool for monitoring species diversity: an assessment.
A landscape of 5×5.5 km in the Karnataka region of the Western Ghats of India was mapped into seven landscape element types, using field identification of types as well as supervised and unsupervised classification of satellite imagery. Plant communities distributed in these landscape element types were surveyed in the field using 246 quadrats of 10×10 m, in order to assess whether these types could be distinguished in terms of species composition. All angiosperms excluding grasses, which could not be identified accurately in the field, were recorded for this purpose. Landscape element types identified in the field harboured significantly distinctive sets of species of flowering plants, and were also by and large distinctive in terms of their species richness. Landscape element types could be identified accurately on the basis of supervised classification: the types thus demarcated harboured distinctive sets of flowering plants. Landscape element types coupled to satellite imagery could then be used to organize a programme of monitoring biodiversity. Unsupervised classification of satellite imagery did not permit classification of landscape element types with a high enough level of accuracy. In consequence, the demarcated landscape element types did not harbour significantly distinctive sets of species of flowering plants. Unsupervised classification is therefore not appropriate in a programme of monitoring biodiversity.