Mapping arid vegetation associations with HyMap imagery

Abstract

HyMap airborne hyperspectral imagery was applied to problems of discriminating variation in vegetation composition and mapping vegetation communities in a southern Australian semi-arid mallee and chenopod shrubland environment. Spectral absorption-depth methods were applied to chlorophyll, water, cellulose/lignin and oil/wax image spectral features to produce distribution and abundance maps of different vegetation components including trees, shrubs and dry plant litter. Surface soil and rock outcrops were mapped using clay, carbonate and iron oxide absorption features. Quantitative data on ground cover from field plots was significantly correlated with selected image endmembers, and served to verify the identity of the vegetation species mapped. Several image vegetation endmembers were combined to produce vegetation association maps that showed distributions similar to maps previously produced by conventional photo-interpretation and field sampling methods. The HyMap vegetation map however showed more detail about variation within communities and more precise definition of boundaries. Particular vegetation communities could be characterised by different endmember "profiles" or combinations and abundances of vegetation and soil end-members.