POPULATION GROWTH AND LAND COVER CHANGE IN EGYPT: EVALUATING THE UTILITY OF
ARCHIVAL AND RECENT DMSP/OLS DATA SETS FOR URBANIZATION STUDIES
Lawrence, William T., Associate Professor, Natural Sciences, Bowie State University, Maryland USA; Marc L. Imhoff,
Scientist, Biospheric Sciences, NASA Goddard Space Flight Center, Maryland USA; David Stutzer, Research Associate, Natural
Sciences, Bowie State University, Maryland USA and Norman Kerle, Graduate Student, USRA, USA and Germany.
Commission VII Symposium, Working Group No. 5
KEY WORDS: urbanization, land cover, land change, DMSP/OLS, Landsat, nighttime images
ABSTRACT:
Currently it is extraordinarily difficult to locate and track change in urban land cover classes at the global scale using
anything but high resolution remotely sensed data. Even if global coverage were available from high resolution sensors [< 60
m resolution], data management, acquisition and analysis would pose almost insurmountable difficulties using standard
classification approaches. To find more feasible approaches for much needed urbanization studies, we have developed
successful techniques employing the nighttime imaging capability of the U.S. Air Force Defense Meteorological Satellite
Program’s Operational Line Scanner. DMSP/OLS data was originally deployed for meteorological forecasting based on
nighttime cloud patterns, but also offers evidence of Man’s presence through the “city lights” emitted by populated areas
around the globe. After three years of working with various types of DMSP/OLS data, we have developed successful techniques
for extracting accurate urbanization and population-related information from the raw data and derived products. This study
looks at DMSP/OLS and high resolution Landsat data for Egypt to estimate the areal extent of urban lands, and to see if
population growth can be monitored through time using archival DMSP/OLS data sets. We found that our technique, using a
modified multi-orbit DMSP/OLS product, provided an estimate of urban land cover at 3.7% of the total land area for Egypt.
This number, based on “city lights” type analysis, is very close to other values derived with non-remote sensing techniques.
We also found that over 30% of the soils most appropriate for agriculture are now apparently under urban land cover; probably
unavailable for agricultural uses. Historical analysis of single orbit DMSP/OLS data to estimate population change proved
futile due to unknown instrument gain, noise, and non-digital data sources.
1. INTRODUCTION
There is a tremendous need to have current maps of the
location and density of populations across the globe. In
addition to population information, it is useful to know the
level of urbanization, or the presence of infrastructure related
to human occupation. These kinds of data are of vital use to
planners, management, many government and non-
governmental organizations and the scientific community.
In addition, the presence and level of development of
infrastructure can be used to determine the impact of human
habitation on the natural and agricultural productivity of a
region. These data, when used with soils, productive
potential or other ecological data or models, can help gauge
the extent to which human Occupation and concomitant
urbanization reduce the access to arable soils or displace
native vegetation communities, both of which have
potential impacts on economic, global change and
biological sustainability issues [Daily and Ehrlich, 1992;
Ehrlich and Ehrlich, 1992; Kates et al., 1992; Myers and
Simon, 1994; Raven, 1991].
Traditional methods for acquiring these types of data include
traditional census, field observations, aerial photography
and interpretation of images from sensors on a variety of
satellite platforms. Census data has been by far the most
useful for populations studies where quantification is
required, but image analysis has been effective for locating
the infrastructure related to human Occupation. Drawbacks to
census include the infrequency of large scale sampling,
usually carried out at intervals in the developed world,
difficulties and expense in the process, and the absence of
Current data for-much of the world. Remotely sensed data
International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998
also has drawbacks, including difficulty in interpretation,
large quantities of data to acquire and manage, and the cost
and/or lack of access to appropriate world-wide.data sets.
In response to these difficulties we have spent several years
developing ways to determine population and urbanization
measures from a data source that collects nighttime images
of the Earth. The source of these data, the Operational Line
Scanner on the U.S. Department of Defense's [DoD] Defense
Meteorological Satellite Program, has proven valuable to us
for a variety of studies. The DMSP/OLS nighttime images
are acquired using a photomultiplier with a visible-nearIR
[400-1100 nm] broadband sensor with a nominal spatial
resolution of 2.6 km [Kramer, 1994]. The data is acquired to
support weather forecasting missions for the U.S. Air Force
based on the observation of moon-lit clouds. The
DMSP/OLS images have had long-term use in the scientific
community for the detection of snow and ice as well as for
delineating urbanized areas world-wide [Croft, 1977; Welch,
1980; Welch and Zupko, 1980]. The early use of DMSP/OLS
was novel but due to problems with registration, unknown
instrument gain, glint on waterbodies, cloud cover and
ephemeral light sources such as lightning and fires the
information was difficult to use quantitatively. Thanks to a
collaboration between the DoD and the U.S. National
Oceanographic and Atmospheric Administration [NOAA]
National Geoscience Data Center [NGDC] the DMSP/OLS
data are now being archived and advanced products being
made available to the user community.
Our research has looked at many of the DMSP/OLS products
available from NOAA NGDC. In this paper we use historical
data [early data was recorded on thermal printers], single
443
