Spectral Mixture Analysis (SMA) was used to distinguish the fractional
abundance of green foliage, dry grass, and soil in Airborne Visible I
nfrared Imaging Spectrometer (AVIRIS) data. Three maximum likelihood c
lassifications were performed using topographic data only, SMA fractio
ns only, and both topographic and SMA fractions. The predictions were
compared to a held based vegetation map and to an aerial photograph of
the scene. The combined data set produced the highest correspondence
with the vegetation map for an overall correlation of 57 per cent for
five classes. Part of the difference was attributed to misclassificati
on in the held-based map.