Page 269 - Digital Analysis of Remotely Sensed Imagery

Image Enhancement 231

<=10%
11–20%
21–30%
31–40%
41–50%
51–60%
61–70%
71–80%
81–90%
91–100%

FIGURE 6.20 Distribution of grass cover density derived from an NDVI image
that is produced from Landsat TM3 and TM4 in conjunction with 68 in situ
density samples. (Source: Zha et al., 2003a.) See also color insert.

Relatively easy to derive from multispectral remote sensing
data, NDVI has found wide applications in quantifying vegetative
cover on the ground (Bryceson, 1989), monitoring land surfaces and
vegetation canopies, and estimation of leave area index, in addition
to estimating grass cover, vegetation biomass, and quantifying per-
centage grass cover. Multitemporal NDVI data are routinely used to
study vegetation health and seasonal variations. With a number of
concurrently collected in situ samples, it is possible to convert radio-
metrically calibrated satellite data into absolute cover densities on
the ground via this index (Zha et al., 2003a). Hence, NDVI is effective
at detecting and visualizing the quantity of grassland biomass
(Fig. 6.20), revealing potential and current status of grassland,
from which it is possible to quantify grassland degradation. If com-
bined with the moving standard-deviation index, NDVI proves a
powerful index in monitoring degradation patterns in a semiarid
heterogeneous landscape (Tanser and Palmer, 1999).

6.6 Image Transformation
Image transformation differs from all the image processing tech-
niques covered previously in that the input image is made up of more
than two bands. It is also possible to output multiple bands from
image transformation, their exact number depending upon the specifi-
cation by the analyst. These newly created bands represent the reor-
ganized information content of the raw bands. Images are transformed