though the means of the two images are about the same, the 
large SD of the 1993 image is responsible for its dominance in 
the PC1 image (Table 1). 
In the PC2 image for the two data sets, gray tones indicate 
insignificant or no NDVI changes (Figure 5). Such areas are 
observed mostly in the desert with shifting sands and barren 
soil that are incapable of supporting vegetation. The negative 
loading of the 1987 NDVI image translates to dark pixels and 
the positive loading of the 1993 NDVI image bright pixels. 
Thus, white areas denote increase in vegetation from 1987 to 
1993 and dark areas decrease in vegetation from 1987 1993. 
The prominent areas with increase in vegetation are observed 
in the north and east of Burgan oil field. However, there a few 
locations that show increase or decrease in NDVI values that 
are not related to vegetation. An example is the oil spill at 
Wafra oil field that appears as white on the PC2 image. The 
oil spill was present in February 1987, but was increased 
during the 1991 Gulf War and hence the white color. The oil 
spill shows up as black in both Figures 3 and 4 indicating no 
relations with vegetation. Also, some of the coastal sabkhas 
show up as white due to the increase in their water content 
from 1987 to 1993. The noticeable black/dark areas with 
decrease in vegetation from 1987 to 1993 are the oil lakes at 
Burgan oil field, the town of Ahmadi, sections of the 
experimental and Wafra farms, and some coastal sabkhas. The 
dark sections of the Wafra farms denote decreased plant vigor 
or non-cultivated areas from 1987 to 1993. 
4. CONCLUSION 
The ability to detect and monitor vegetation changes is crucial 
to detecting anomalous conditions that could enhance 
desertification or disrupt the plant ecosystem in arid 
environments. The technique offered by satellite remote 
sensing in global, regional and local vegetation analysis is 
second to none as minor changes can easily be detected. 
Desertification in an arid environment is a natural process that 
cannot be completely eliminated, but the ability to spot trend 
could be useful in early warning or remedial solutions. 
Analysis of Landsat TM images showed sensitivity to detecting 
vegetation changes in the arid environment of Kuwait. The 
TM NDVI images proved effective in mapping vegetation 
changes that occurred between the data sets acquired on 
February 4, 1987, and February 28, 1993. The NDVI images 
were used as input to a selective PCA procedure. Because the 
major spectral information mapped by the NDVI image is 
related to active green biomass, the change image (PC2) 
accounting for 19.82% of the total information between the two 
data sets represented vegetation increase between 1987 and 
1993, and conversely a decrease of 19.82% comparing the 
1993 and 1987 data sets. The observed vegetation variation 
was supported by rainfall data that increased by approximately 
three times between the two dates. Rainfall and irrigation as 
expected were therefore, the principal causes of vegetation 
variation in the study area. In the absence of abundant rainfall 
and irrigation the desert environment can only support a few 
desert plants. By virtue of the adverse effect of the burning of 
the oil lakes that occurred during the 1991 Gulf War, there was 
loss of vegetation in 1993 compared with what would have 
been observed with the high rainfall. 
Some authors have reported the ineffectiveness of the NDVI in 
desert areas with vegetation less than 20% (Choudhury gpg 
Tucker, 1987). Such observations were made from studies 
using AVHRR data sets with a resolution of 1.1 km/pixel 
However, Landsat TM data with a resolution of 30 m has 
proved to be quite effective in mapping vegetation changes in 
arid environments due to the high spectral and spatial 
resolutions. Some vegetation density would be missed because 
of the resolution of 30 m but the analysis gives a good order 
estimation of vegetation changes. 
A sensitive enhancement technique such as selective PCA that 
was used in this study has several important applications in 
arid environments. By comparing several years of satellite 
data, anomalous trends could be detected and, if possible, 
remedial measures applied. The technique could be used to 
map potential arable land in arid environments. The areas that 
manifest high seasonal or inter-annual vegetation variations are 
most likely to support plant cultivation in the presence of 
abundant water. Shifting sands and sandy area can only 
support limited plant life even with abundant rainfall In 
Kuwait, such studies a prerequisite in monitoring the resilience 
of the desert environment after the 1991 Gulf War. During the 
War, more than 50% of the desert environment were disturbed 
by the large-scale movement of troops, the burning of the oil 
wells, and post-war cleanup activities. 
5. REFERENCES 
Chavez, P.S., Jr, and A.Y. Kwarteng, 1989. Extracting 
spectral contrast in Landsat Thematic Mapper image data using 
selective principal component analysis. Photogrammetric 
Engineering and Remote Sensing, 55(3), pp. 339-348. 
Chavez, P.S., Jr, and D. MacKinnon, 1994. Automatic 
detection of vegetation changes in the southwestern United 
States using remotely sensed images. Photogrammetric 
Engineering and Remote Sensing, 60(5), pp. 571-583. 
Choudhury, B.J., and C.J. Tucker, 1987. Satellite observed 
seasonal and inter-annual variation of vegetation over the 
Kalahari, the Great Victoria Desert, and the Great Sandy 
Desert: 1979-1984. Remote Sensing of Environment, 23, pp. 
233-241. 
Dymond, J., P.R. Stephens, P.F. Newsome, and R.H. Wilde, 
1992. Percentage vegetation cover of degrading rangeland 
from SPOT. International Journal of Remote Sensing, 1311), 
pp. 1999-2007. 
Fritz, L.W., 1996. The era of commercial Earth observation 
satellites. Photogrammetric Engineering and Remote Sensing, 
62, pp. 39-45. 
Goward, S.N., B. Markham, D.G. Dye, W. Dulaney, and J 
Yang, 1991. Normalized difference vegetation index 
measurements from the Advanced Very High Resolution 
Radiometer. Remote Sensing of the Environment, 35, pp. 257- 
277. 
Halwagy, R., and M. Halwagy, 1974. Ecological studies on tle 
desert of Kuwait, Part IL. The vegetation. Journal of University 
of Kuwait (Science), 1, pp. 87-95. 
404 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996 
Sd 
sea 
Phi 
32) 
Sir 
tec 
of] 
To 
dyr 
Veg 
(11 
Tu 
of 
Ser