ırated pixels
; - The
"AO soils
ff et al.,
A NGDC
ts in close
stated that
ur analysis
he analysis
| and cover
> soil types
n in Table
; in Egypt.
r of area
by far the
ing to our
> under the
hypothesis,
area. Note urban, agricultural and desert land cover types.
removing them from potential agricultural uses with its
concomitant negative impact on potential productivity in
this region. The second most urbanized soils are Regosols
and Xerosols, which are generally a young, shallow or sandy
soils, or arid desert soils respectively. Neither of these are
well suited to agriculture, but their use is entirely appropriate
to urban land use, thereby sparing more useful soils. The
FAO soils map is very course resolution and has some
registration faults. We suspect this registration error, along
with possible realignment of channels and glint of urban
light onto adjacent water bodies is responsible for the
apparent occupation of 417 km? of water by urban land
cover. The "city lights" product was not customized for
Egypt. We used the average threshold of >89% that proved
effective in our study of the United States [Imhoff et al.,
1977a,b]. Further refinement of this threshold will provide
better accuracy. We have used FAO soils-based fertility
indices in the U.S. [Imhoff et al., 19772] to get a better idea
of the occupation of the most ideally suited soils by urban
lands. The Fertility Capability Classification index [FCC,
Buol et al., 1974, Sanchez et al., 1982] was used in the
previous study. We are now working on a modeling
approach to further evaluate the impacts of urbanization on
global productivity.
Table 2. FAO Soil types under urban land cover as
determined using the "city lights" data set [Imhoff et al.,
1997a,b]. Area measurements are in km?.
FAO Soil Order Total Area Urbanized '96 % Area
Yermosols 572562 7053 1.2
Lithosols 158754 2619 1.6
Dunes/Sand 149156 1275 0.9
Fluvisols 55250 17993 32.6
Solonchaks 31334 1657 5.3
Regosols 21667 5155 23.8
Water 4759 417 8.8
Rock 2230 0 0.0
Xerosols 1096 233 23.3
Totals 996809 36423 3.4
International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998
M tes ^ 3 RE E
% A EEROA ^ uos
Figure 5. 1987 Landsat Thematic Mapper image of Cairo
4. CONCLUSIONS
4.1 DMSP Historic and Single Orbit Data
Due to contrast, variable contrast, registration and gain
issues the historical DMSP/OLS archive is difficult to use in
more than a qualitative fashion. Further work may prove
productive in refining this product. Further work would be
useful as no digital data exists prior to mid-1990's from the
DMSP/OLS program. Likewise, the single orbit digital
DMSP/OLS product presents similar difficulties with
unknown gain. Our work in attempting to provide a historic
measure of urban land cover change in Egypt using these data
was unsuccessful. :
4.2 *City Lights" Enhancement of DMSP/OLS
Once we discovered the method for ‘thresholding’ the NOAA
NGDC ‘stable lights’ product [Imhoff et al., 1997a,b] we
found that we could eliminate or significantly reduce the
problems of misregistration, cloud cover, glint of light into
adjacent water bodies, etc. The "city lights" product has
proven to be an accurate representation at least in total area
of urban land cover in our initial study in the U.S. [Imhoff et
al., 1997a] and in this study where we find 3.7% of Egypt
under urban land cover types.
4.3 Landsat MSS and Thematic Mapper Data
High resolution daylight sensors will always be highly
suited to determination of areal extent and location of urban
land use types. Our work showed both MSS and TM to be
useful in this regard, although very light infrastructure,
which may not preclude the use of lands in agriculture, may
be interpreted as 'urban' classes in classification schemes.
Our analysis showed that as of 1987 perhaps as much as
12% of the area under study was in an urban class. This
number cannot be compared directly with our whole country
analysis because it only included the heavily urbanized area
of Cairo and the Nile delta - a 14,400 km? area where we had
Landsat multitemporal coverage. Problems with high
resolution data will always include cost of acquisition, high
data volumes, low revisit frequency, cloud cover and
difficulty in interpretation.
4.4 FAO Soils under Urban Land Cover Classes
Our fusion of "city lights" urban land cover classes with
FAO soils showed a disturbingly high occupation of the
most prevalent agricultural soil in Egypt [Fluvisols] -
almost 33%. This occupation follows the pattern found in
the U.S. by Imhoff et al. [1997a] where respectively 15, 8, 7
and 2% of the best soils were covered by urban land cover in
the four most highly productive agricultural states in the
U.S. [California, Illinois, Florida and Wisconsin].
Fortunately in the U.S. only about 2.9% of the total land
area is urbanized and much of the country has vast areas of
highly fertile soils. In places like Egypt where the arid
climate greatly reduces the potential for agriculture, an
occupation rate of 3.7% may significantly compromise
continued agricultural productivity, especially if conversion
of the most appropriate agricultural soils to urban land use
continues,
447
