important. 
riedness 
lso from 
generate 
he pixel 
ance of 
. weakness 
pear but 
cts in one 
| weel dif- 
'ecognition 
ircely ap- 
i maximally 
lor of 
the pixel 
total 
rous clus- 
nan assis- 
] this ap- 
iminating 
of spetral 
point out 
related 
  
3. Case studies 
The following examples are dealing with two urban areas of Belgium, 
namely Charleroi and Brussels. These cities have very different structures 
indeed. The former is essentially a conurbation made of the coalescence 
of previous villages overrun by industry, including a just three century old 
hardcore of tertiary (military) origin. The latter is a multifunctional capital 
progressing on a radial growth started from an old nucleus and spreading into 
the surrounding countryside or peripherical districts featuring former suburbs 
and recent outskirts. 
It appeared to be interesting to examine to what extent the satellite 
enables the analysis of the morphological and possibly functional structure of 
both cities by means of land use. In this order two LANDSAT recordings went 
into multispectral examination : March 22, 1973 and May 17, 1976 (LANDSAT -1). 
Besides the interest of observing seasonal change effects on urban and sub- 
urban open spaces phenology, the 3 years period between two recordings could 
be useful to the analysis of the urban land evolution. 
a. Built up areas quantification 
  
This study calls for supervised methods. The unsupervised classifi- 
cation indeed leads to two important misunderstandings, between built up and 
open spaces. On the one hand, certain areas covered with recent concrete ap- 
pear in the same class than certain bare grounds ; on the other hand, old 
thickly settled parts merge with water areas on the other extremity of reflec- 
tance classes (pools or canals). 
In this connexion, the high frequency of class 
6 is noticeable in the Brussels periphery. By big patches or small groups, 
this class splits the rural space into a lot of separated pieces. It is mo- 
reover associated with other classes in various rates having to be stated. 
The comparison using aero-sketchmaster towards topographical maps 
and aerial photographs shows it is concerned with a characteristical component 
of residential settlement on private plots ; most of the time the house lies 
in front of the plot with a rather large yard behind it. The relative density 
of the settlement and its regularity have an effect on reflectance and seem to 
concede to this land use its relatively specific signature. The association