884 
Figure 6. B&W reproduction of September 18, 1973 Skylab-3 S-190B image of Chicago's O'Hare Airport (left) and 
Loop (right). Many details of the airport, surrounding industrial parks, and residential areas are readily 
discernable in the 25 m resolution O'Hare image, while details of Chicago Harbor, Grant Park, the Navy pier, and 
Meigs Field can be detected in the Loop image. N 
scenes collected two years aparty ^‘s described by 
Todd (1977). 
2.3 Medium Resolution Data 
Advancing from the Landsat 80 m MSS to 40 m RBV 
results in significant improvement in identification 
and delineation of urban features. Using the MSS / 
data, residential areas have a mottled texture and 
tone/color which is often confused with land cover 
categories being mapped. But street patterns are 
evident within RBV imagery, which gives residential 
areas a characteristic texture/pattern (Figure 4). 
Of interest is the greatly improved confidence of 
residential mapping at the urban-rural fringe (Lauer 
& Todd, 1981). 
Two studies -- Lauer & Todd (1981) and Snyder 
(1982) report taking advantage of both the high 
spatial resolution of the R8V and the multi spectral 
attribute of the MSS data. In the first study, the 
MSS and RBV data were spatially registered, the MSS 
data were resampled to the smaller pixel size of the 
RBV data, and then color composite images were 
created for interpretation.!! The authors report that 
the RBV data can be used alone, without the MSS, for 
urban land cover mapping, but that the "combined 
RBV/MSS color composite image is easier and quicker 
to work with than the MSS qr RBV image alone." In 
his analysis of MSS and RBV imagery collected over 
Soviet cities, Snyder (1982) found that the RBV 
provides "better delineation of boundaries," but 
MSS gives "better categorical accuracy." 
The 40 m RBV data was only a forerunner of the 30 
m Thematic Mapper (TM) data, which became available 
with the launch of Landsat-4 in July, 1982. Both 
Toll (1985) and Quattrochi (1983) discuss the 
significant advantages of TM over the MSS. 
Bernstein et. al. (1984) notes that high-contrast, 
linear features as narrow as 7.6 m (about 0.25 
pixel) can "be easily discerned." 
Similar to the RBV data, residential areas 
exhibit a characteristic texture/pattern which 
greatly assists in detection and delineation (Figure 
5). The urban-rural fringe can be outlined, and 
confidence in identification of other features is 
increased. For example, a hint of within-feature 
texture/ pattern aids in mapping industrial/ 
commercial areas, transportation/port facilities, 
aTrd densely vegetated urban land uses such as parks, 
golf courses, vacant land, and low density 
residential, 
High Resolution Data 
Relatively high resolution photography was 
collected from space over urban areas in the early 
1970 1 s by Skylab (Figure 6). Evaluation of Skylab's 
S-190B Earth Terrain Camera photography was done by 
Welch (1974) and an urban mapping experiment is 
described by Lins (1976). The 25 m S-190B 
photography was interpreted to yield urban land use 
categories, including single-family residential, 
multifamily residential, industrial and commercial 
complexes, highways and other transportation 
facilities, improved open space, and transitional 
areas. Several additional classes mapped by Lins 
such as retail trade, education facilities, 
religious facilities, and government/administration/ 
services were probably identified with either 
ancillary data or field work. 
More recently, the Large Format Camera (LFC) has 
been flown on the Shuttle, and has taken high 
resolution photographs of urban areas (Doyle, 1984 & 
1985). The 9.5 m spatial resolution of the LFC 
photograph shown in Figure 7 illustrates that shape 
and shadow are additional image characteristics 
which can be used in image interpretation, 
permitting detailed (Level I I/I 11) land use and land 
cover interpretation. Residential areas (several 
types may be identified) have detailed and unique 
texture and pattern, which allows accurate detection 
and delineation. Shapes of structures and 
transportation features assists in identification of 
industrial, commercial, services, and transportation 
land use patterns. 
Similar resolution is available with SPOT data, 
although SPOT also has the 20 m multi spectral data 
(Figure 8). SPOT analysts commonly utilize 
digitally enhanced SPOT color composite imagery -- 
combinations of the 10 m panchromatic and 20 m 
multi spectral data. Welch (1985) reports that 
"classification accuracies in excess of 80 percent 
can be realized for selected level I I/I 11 urban 
classes." Following is the Urban portion of the 
classification system used (after Anderson et. al, 
1976) to analyze SPOT simulation data: