orbital swath digital data, and the NGDC ‘stable lights’ data
product in our analysis [Elvidge et al. 1997a]. The ‘stable
lights’ product is a global product that uses multiple orbits
of the DMSP/OLS that are manually screened for clouds and
mapped into a single product. For any one location,
multiple cloud-free orbits are collected and a number,
representing the fraction of time the location was lit out of
all observations, is presented as the location or pixel digital
value. Values range from 0 to 100% lit for any location. The
‘stable lights’ data, when appropriately processed, can be
used to delineate urban areas world-wide.
We have had excellent results in delineating urban areas in
the coterminous U.S. [Imhoff et al., 1997a, 1997b] using a
thresholding technique that eliminates ephemeral light
sources, misregistration, and water glint from the NOAA
NGDC ‘stable lights’ data set. Our approach has been used to
map urbanization on soil resources in the U.S. [Imhoff et al.,
1997a] and to compare urban area from the U.S. with that
from the thresholded ‘stable lights’ data [Imhoff et al.
1997b]. The two estimates were within 5% over the total
7.7 X 106 km? of the U.S. with 2.7% of the area urbanized.
The approach of Imhoff et al. [1997a,b] is being used by
Gallo and Owen [1998] to differentiate between urban,
suburban and rural weather stations in the U.S. Similar
approaches are being used to refine the data set and use
newer ‘radiance controlled’ data for studies of population
[Sutton et al., 1997], human settlement [Elvidge et al.,
1997b] and human energy use [Elvidge et al., 1997c].
The purpose for this research work was an interest in the
complex problems of urban expansion and population
growth in the arid regions of the Mid-east. The work was
stimulated by an initiative sponsored by Headquarters of the
U.S. National Aeronautics and Space Administration. The
issues of land occupation, access to arable land, water rights,
and rapid population growth in this area made remotely
sensed indicators of population density and urban
infrastructure of great interest. This project was an initial
‘proof of concept’ effort to see if historical trends in
population or urban areas could be evaluated with DMSP/OLS
product and/or high resolution daytime data, if soils
resources were in fact heavily urbanized, and if our approach
would work in areas other than the continental U.S. where
they were developed. Further work will build on this initial
project and will focus the same types of analysis on Israel
and Jordan.
2. METHODS
2.1 Data Sets Used
2.11 DMSP/OLS
DMSP/OLS ‘historical’ non-digital products -
DMSP/OLS data prior to the mid-90’s were only acquired in a
‘hard copy’ format. These data were digitized from a paper
print using a desktop scanner. Usable ‘historical’ data sets
included images from 10/77, 06/84, 07/86, 01/88 and
02/90. The resultant files in raw binary format were co-
registered using prominent features such as water ways and
coastlines, and then registered with the DCW regional map.
DMSP/OLS single swath products - A single orbital
swath of DMSP/OLS nighttime data was acquired for 10/96.
This data set was subset [it contained an entire orbit of the
Earth] to just cover the area under study. The data was
registered with the other DMSP/OLS nighttime images.
Enhanced DMSP/OLS composite orbit “city
lights” product - We used the ‘stable lights’ DMSP/OLS
product available from NOAA NGDC, but enhanced it with
out ‘thresholding’ technique to remove registration errors,
ephemeral light sources and light glint into water bodies
[Imhoff et al., 1997a,b]. We call this enhanced urban land
cover product "city lights".
2.12 Landsat Images: Thematic Mapper [TM] and
MultiSpectral Scanner [MSS] - 1976 MSS and 1987 TM data
were acquired for the U.S. Geologic Survey EROS Data
Center. A block of four adjacent scenes were required to
cover the lower Nile, Cairo and Nile Delta areas that were .
under study. The MSS scenes were 189/038, 189/039,
190/038 and 190/039 [WRS I]. The comparable TM scenes
were 176/038, 176/039, 177/038 and 177/039 [WRS II].
These data were acquired to test a process of classification of
daytime images for extraction of areas under urban-type land
cover.
2.13 FAO Soils Map: The UNESCO FAO Digital Soils
Map for the World was acquired from FAO and subset for this
project. The map is produced at a scale of 1:5 million, is of
course resolution, but the only global soils database
available. It is based on a paper product. (FAO-UNESCO,
1975, 1992].
2.2 Data Registration
Digital Chart of the World [DCW, ESRI, Redlands CA USA]
was used as the base map for registration of all the image
products. Albers Equal Area projection was used. Ground
control points were extracted from the DCW for registering
TM, MSS and DMSP/OLS historical and single orbit data
sets. No registration was required for either the UNESCO
FAO Soils Map or the DMSP/OLS ‘stable lights’ data as they
are distributed in geographic coordinates.
2.3 Analysis
Due to the wide variance in spatial coverage between the data
sets used in this analysis, we chose several areas in common
for our urban land cover study. For the Landsat
multitemporal [1976-1987] land use study, this area was
4000 X 4000 TM 30m pixels, approximately 120 X 120 km
or 14,400 km?. The geographic bounds were N319 10'
20.71", E309 26' 19.21" [upper left corner] and N29» 38"
40.38", E319 57' 59.07". This area covered Cairo and most
of the Nile delta region of Egypt. For the examination of
soils covered by urban land use based on the "city lights"
data the entire country was used.
2.31 DMSP/OLS
DMSP/OLS ‘historical’ non-digital products -
These non-digital products were scanned and then a manual
process used to 'threshold' the the wide range digital values
until only solid blocks of lighted pixels were visible. This
was an iterative process that removed moonlit desert,
444 Intemational Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998
OR d a. And ims uh "C 6 SZ BET V A
— — ; ge ed > ba
©
Ww
