Settlement patterns and distributions in the north China plain were examined by Lo
(1984). Using SIR-A imagery attained at scales varying from 1:500,000 to 1:130,000
enlargements, Lo found that settlements, in general, stood out against a dark background
of surrounding terrain. Areas were grouped into population size groups, the smallest being
those below 10,000 population. He also commented on the lack of conformity Lo the
Christallerian structure of central places. In a later study (Lo, 1986), he reported that
there was much more detection variability across four topographic regions of the United
States with accuracies ranging from 60 to 100 percent. Causes related to azimuth angle,
soils, and surrounding land cover. SIR-A imagery of north China was also studied by
Jiyuan, et.al. (1986). They found residential areas to be one of the "obvious” features
because walls of buildings and metallic structures strongly backscattered the radar signal.
However, when the imagery was density sliced, a detection accuracy of only 50 per cent
was attained for residential buildings.
Endlicher (1982) and Endlicher and Kessler (1982) commented in general terms on the
visibility of intra-urban landcover types in and around Cologne, Germany, using Seasat
5AR imagery. Examining airborne X- and L-band imagery of Freiburg, Germany, in a
later study, Endlicher and others were able to distinguish among older urban areas, modern
housing estates and office blocks, and house and garden areas of the city (Trevett, 1986).
Unfortunately, although each of these reports were useful and informative, they provided
only qualitative, descriptive data for the European environs. Harris (1985) stated that
towns and villages in his Tunisian study area were visible as white circular patches against
a dark background of flat to undulating topography. However, he also stated that
settlements were "not so clear" on the Seasat imagery due to its low incidence angle.
Sieber (1985) examined the statistical properties of multi-polarized X- and L-band
airborne SAR-580 imagery of villages selected as typical of Bavaria and showing different
stages of historical settlement. He did not state how many pixel values or urban areas
were used but did determine that additional data sets were needed as house orientation
and type, house density, roof type, and the vegetation present all affected the radar return.
Bartholome (1983) and Dowman and Gibson (1984) concluded that one of the factors
preventing a confident, detailed settlement classification with airborne X-band imagery
was the presence of trees within the settlements. However, Endlicher, et.al. (1984), also
using X-band imagery, stated that trees and vegetation helped to accent settlements and
differentiate intra-urban land covers.
Kessler (1986a) examined three different passes of SIR-B data over a small area near
Freiburg, Germany, to assess the visibility of land use classes. Among his findings were
the ease of settlement detection in the flat river valleys and the difficulty of separating
one land use class from another in the mountain forested areas. In a related work,
(Kessler, 1986b) he concluded that the bright signal return from settlements was similar to
that from river valley corn and certain slopes in a volcanic vineyard area near