î 
885 
rt (left) and 
îre readily 
î Navy pier, and 
such as parks, 
>ity 
iy was 
in the early 
in of Skylab's 
t was done by 
'iment is 
90B 
'ban land use 
¡idential, 
1 commercial 
:ation 
:ansitional 
>ed by Lins 
:ies, 
Imini strati on/ 
either 
1 Urban or Built-Up 
11 Residential 
111 Single Family Housing 
112 Multiple Family Housing 
12 Commercial and Services 
121 Commercial 
122 Institutional 
13 Industrial 
14 Transportation and Utilities 
141 Road, Highway, Railroad 
142 Utilities 
17 Other Urban or Built-Up Land 
171 Urban Grassland 
172 Golf Course 
Colwell and Poulton (1985) achieved equally 
good success in using SPOT simulation data for 
identification of urban features. The authors note 
that detailed image interpretation "depends very 
largely on resolution detail that preserves the 
integrity of building shape and allows 
detection of major and minor streets and secondary 
roads." 
Welch tried computer-assisted multi spectral 
classification of SPOT data, but with unsuccessful 
results. Interestingly, successful 80 m Landsat MSS 
classification results cannot be duplicated with the 
SPOT data because of the greatly increased texture 
and complex image information level of the latter 
(Ballut and Nguyen, 1984). 
Microwave data collected at lower frequencies 
has the highly desirable characteristic of 
pentrating clouds. Bryan (1982) describes 
applications of 25 m resolution Seasat SAR for urban 
mapping. He was able to distingish urban land cover 
types which are similar to categories researchers 
have been able to detect using multispectral data 
collected in the visible and infrared portions of 
the spectrum. A future source of 30 m resolution 
radar data is the planned European Space Agency's 
Remote Sensing Satellite (ERS-1) C-Band SAR 
(Duchossois, 1984). ESA plans to launch ERS-1 in 
April 1989. 
Figure 8. Digitally enhanced SPOT simulator visible 
red image (Band 2, 610-680 nm) of Santa Cruz 
collected in June, 1983. Street patterns are 
clearly visible in the 20 m resolution image, as 
well as the piers in the elongated small boat harbor 
which occupies a dredged slough. 
Table 1. Empirical Evaluation of Satellite Imagery 
Collected over Urban Areas. 
IMAGE 
GROUND 
RESO 
LUTION 
PRIMARY IMAGE 
CHARACTERISTICS FOR 
IMACE INTERPRETATION 
TONE/COLOR 
TONE/COLOR 
♦ RELATIVE LOCATION 
+ SOME TEXTURE/PATTERN 
TONE/COLOR 
♦ RELATIVE LOCATION 
♦ DETAILED TEXTURE/ 
PATTERN 
+ SHAPE/SHADOW 
• URBAN VS NON-URBAN 
i URBAN LAND COVER TYPES 
i MAJOR TRANSPORTATION/ 
COMMERCIAL ARTERIES 
i LAND CLEARINC DETECTABLE 
RESIDENTIAL HAS SOME 
TEXTURE/PATTERN OF STREETS 
URBAN/RURAL FRINGE DISTINCT 
i VERY LARGE BUILDINCS 
DETECTABLE 
i URBAN LAND USE/LAND COVER 
FEATURES DETECTED AND 
DELINEATED WITH HIGHER 
CONFIDENCE 
i NEW CONSTRUCTION (LAND 
SCRAPINC) EVIDENT 
i DETAILED LEVEL ll/lll URBAN 
LAND USE MAPPING POSSIBLE 
i LARGE AND MEDIUM SIZED 
STRUCTURES ( AND SHADOWS) 
DETECTABLE 
i ALL TRANSPORTATION FEA 
TURES EVIDENT 
• VEGETATION HAS DISTINCTIVE 
TEXTURE 
■a (LFC) has 
n high 
Doyle, 1984 & 
: the LFC 
is that shape 
eristies 
on, 
use and land 
s (several 
and unique 
ate detection 
and 
tification of 
ransportation 
SPOT data, 
ectral data 
I ize 
imagery -- 
nd 20 m 
ts that 
80 percent 
II urban 
on of the 
son et. al, 
Figure 7. Large Format Camera photograph of central 
Boston collected 7 October 1984 by Shuttle Mission 
41-G. With an altitude of 231 km, ground resolution 
is about 9.5 m. Large aircraft, ships, piers, 
buildings (including their shadows) and details of 
Logan International Airport are all visible within 
the image. 
3 SPATIAL RESOLUTION ANALYSIS SUMMARY 
A summary of the utility of the sensors for 
urban mapping is listed in Table 1. As the spatial 
resolution of the data increases, more image 
characteristics are available for image 
interpretation. With very low resolution data, 
tone/color is available to the interpreter, but only 
urban/rural differentiation may be made. Moving to 
low resolution, the analyst uses tone/color, along 
with relative location of features, to distingish 
major urban land use and land cover types, major 
transportation arteries and commercial strip 
development, and land clearing for new 
construction. A degree of image texture/pattern is 
available when using medium resolution data, 
permitting urban land cover types--particularly at 
the urban-rural fringe--to be detected and 
delineated with higher confidence than using low 
resolution data. Finally, the high resolution data 
from space reveal shape and shadow of urban 
features; when that image interpretation aid is 
added to tone/color, relative location, and detailed 
texture/pattern the interpreter can produce detailed 
Level II/I 11 maps.