841 
enth inter- 
ng of envirnment. 
n cultural 
o the archaeo- 
udy of world 
re (ed. ) . 
Symposium on Remote Sensing for Resources Development and Environmental Management / Enschede / August 1986 
Human settlement analysis using Shuttle Imaging Radar-A data: 
An evaluation 
apes: rural 
park system, 
erior. 
ial atlas of 
C.P.Lo 
University of Georgia, Athens, USA 
ockholm: Widlunds, 
ce-borne remote 
of Arizona 
c of Germany, 
eological 
orld cultural 
eere (ed.). 
graphy in anthro- 
e: Harvard 
ABSTRACT: The detectability of human settlements from Shuttle Imaging Radar-A images was determined with 
reference to the radar system geometry and physical and cultural characteristics of the environment in four 
specific geographic regions of the United States represented in five strips of images. The usefulness of the 
settlement area data directly measured from the images for population estimation was also evaluated. It was 
concluded that Shuttle Imaging Radar-A data could produce accurate population estimates of individual 
settlements and complement other forms of high-resolution space data in human settlement analysis. 
1 INTRODUCTION 
In recent years, high-resolution imagery obtained 
from space platforms which can be usefully 
employed in human settlement study becomes 
available. The most notable examples are Thematic 
Mapper data (ground resolution 30 m/pixel), SPOT 
data (ground resolution 20 m/pixel in the multi- 
spectral mode and 10 m/pixel in the panchromatic 
mode), NASA Large Format Camera photography 
(resolution 80 lp/mm), the Metric Camera 
Photography of the European Space Agency 
(resolution 40 lp/mm), and the Shuttle Imaging 
Radar-A data (ground resolution 40 m/pixel) 
(Doyle, 1984; Cimino and Elachi, 1982). Despite 
the relatively poorer spatial resolution of the 
Shuttle Imaging Radar-A (SIR-A) data, its 
employment in human settlement analysis is 
invaluable because of its all weather imaging 
capability which makes it suitable for use to 
monitor changes of the environment at any time and 
in different parts of the world. However, the 
interpretation of these radar image data is more 
complicated than that of photography or imagery 
obtained within the visible portion of the 
electromagnetic spectrum (0.4-0.7 ,um) . Research 
is required to examine problems associated with 
this type of space data and to evaluate 
objectively their utility in human settlement 
analysis. Human settlement analysis is taken here 
to refer to the identification of the shapes and 
sizes of individual settlements and their pattern 
of spatial distribution. This paper reports on 
some preliminary findings of such a research. 
2 NATURE OF THE SIR-A DATA 
The SIR-A data were acquired by the Space Shuttle 
Columbia on 12 November, 1981 with a side-looking 
synthetic aperture radar using horizontally 
polarized microwave radiation transmitted at 
L-band (1.278 GHz) from an altitude of 259 km 
(Cimino and Elachi, 1982; Ford et al., 1983). The 
depression angle of the antenna varied from 46° 
for the near range (southward edge of the film) to 
40° for the far range (northward edge), which 
produced a swath width of 50 km on the earth 
surface. The SIR-A data employed for human 
settlement analysis were optically processed and 
tilt corrected two-dimensional image film which 
was amenable to visual interpretation. The scale 
of the image was 1:500,000. Preliminary 
investigations carried out by the author in 
connection with settlement pattern analysis in the 
North China Plain has confirmed good planimetric 
accuracy of the SIR-A data as compared with 
1:250,000 scale topographic map (Lo, 1984). The 
quality of these radar data was rated excellent. 
3 OBJECTIVES AND METHODOLOGY 
The application of remotely sensed data to human 
settlement analysis involves a study of the 
following points: (a) the detectability of 
individual settlements, (b) the accuracy with 
which the shape of each settlement can be 
determined, (c) the accuracy with which the areal 
extent of each settlement can be delineated, and 
(d) the accuracy with which the population size of 
each settlement can be estimated. 
In the case of synthetic-aperture radar (SAR) 
data, the answers to these questions are clearly 
related to the nature of backscatter from the 
terrain, which is affected by the following 
factors: (a) the depression angle of the antenna, 
(b) the incidence angle of illumination, (c) the 
terrain slope, (d) the properties of the 
structures, (e) the orientation of the structures 
in relation to the illumination, and (f) the 
spatial resolution of the imaging system. 
In order to evaluate the significance of these 
various factors, five strips of the SIR-A imagery 
covering four distinct regions of the United 
States of America were selected. These were: (1) 
St. Joseph, Missouri, (2) Mobile, Alabama 
Hattiesburg, Mississippi, (3) Tallulah, Louisiana, 
(4) Louisville, Kentucky, and (5) Sterling, 
Colorado, which represented Interior Plains, Gulf 
Atlantic Coastal Plain, the Appalachian Plateau, 
and Great Plains in morphological regions (Fig. 
1). The settlements were visually detected and 
their areas measured using 1-mm square grids 
directly from the images. These results were then 
compared with those obtained from the U.S. 
Geological Survey 1:250,000 scale topographic 
maps. Visual comparison was also made on the 
shapes of these settlements. In addition, the 
areas of these settlements as measured from the 
imagery were correlated with the population data 
obtained from the 1980 census. 
Apart from these observations and measurements, 
the effects of depression angle, land-surface 
forms, soil types, vegetation cover, and land use 
on the detectability and size determination 
accuracy of these settlements from the SIR-A data 
were also investigated.