A linear mixture model based on calibrated, atmospherically corrected Probe
-1 hyperspectral imagery was compared with three vegetation indices to test
its relative ability to measure small differences in percent green vegetat
ive cover for areas of sparse vegetation in arid environments. The goal of
this research was tp compare multispectral and hyperspectral remote sensing
approaches for detecting human disturbances of arid environments. The norm
alized difference vegetation index (NDVI) was tested using both narrow and
broad band-widths. Broadband NDVI provided results (r(2) = 0.63) similar to
NDVI derived from individual hyperspectral channels (r(2) = 0.60). While t
he soil-adjusted vegetation index (SAVI) was designed as an improvement to
NDVI for sparse vegetation, in this study SAVI performed significantly wors
e than NDVI (r(2) = 0.51). The modified soil-adjusted vegetation index (MSA
VI) provided an insignificant improvement overt NDVI (r(2) = 0.64). Linear
mixture modeling provided significantly better results, r(2) of 0.74. Cross
-validation was used to test the significance of differences between the va
rious methods and to determine the standard error associated with each meth
od. Results suggest that any improvements provided by adjusted vegetation i
ndices over NDVI may be strongly dependent on those adjustments being deriv
ed from local conditions. The use of a linear mixture model with multiple s
oil endmembers appears to provide the best method for quantifying sparse ve
getative cover. Though present in small amounts, a single plant species, Kr
ameria erecta, was strongly correlated with residuals of the mixture model.
Inclusion of a spectral endmember for this species increased the r(2) of t
he fit with percent green cover to 0.86. However, it is not clear if the ex
plained variation was actually due to K. erecta or a correlated phenomena.
Problems were also identified with the use of multiple vegetation endmember
s. (C) Elsevier Science Inc., 2000.