ul 2004 
fic heat 
Jemand 
method 
it atlas, 
ethods. 
atlas is 
present 
ilding 
ology 
  
p from 
>Y 
'ding to 
iown in 
e has a 
-family 
ore the 
uilding 
on. The 
jlldings 
  
  
arge 
nore 
imily 
ouse 
24,4 
68,9 
40,6 
60,4 
  
o2 
\O 
> 
» 
  
© 
5, 
75,8 
No 
  
  
86,4 
1g types 
and no 
because 
in their 
as to be 
nternational Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004 
Inte 8 S | / 
  
done manually using aerial photographs. In any case, with this 
statistical methods there is no information about the spatial 
distribution of the heat demand. 
One thing one should remember is, that this building typology 
only considers residential buildings. So it can only used in 
residential areas and not for instance in industrial areas. 
A settlement typology describes the combination of different 
building types within a certain area, e.g. a city centre consist of 
blocks of great more family houses and a suburb consist of one- 
family houses or row-houses. 
Some settlement types are more suitable for local and district 
heating. For instance, in dense areas it is very expensive to lay 
the pipelines, so these areas are not suitable for this kind of 
heating. 
4. IDENTIFYING BUILDING TYPES 
From the laser scanning data we only obtain geometric 
information about the building. 
The main task is to create a link between the geometry and the 
building types. This is a classification task, that relates object 
attributes with a certain object class. To this end, different 
classification methods can be applied. Using supervised 
classification, one starts with training data, which contain 
classified examples. In our case, in a first step the building 
volumes are combined with a heat atlas which contains relevant 
data to identify building types, which were used as training 
data. 
We have to decide which attributes should be used. The most 
characteristic attributes are the height, length, width and area. 
An analysis delivers an enhanced building typology which 
includes geometric values. This enhanced building typology 
then can be applied to other regions 
Below different methods to determine the building type are 
described. 
4.1 Approximate values 
In this method we start with approximate values. These values 
can be obtained from experience. From statistical data we can 
obtain some information about certain building types, c.g. 
number of floors, number of apartments and the size of 
apartments. Besides the length and width become greater from 
one-family house to the high tower building. 
  
  
  
  
  
  
  
building area height [m] | length [m] | width [m] 
type [m?] 
one-family 90-115 3-7.5 13-15 7-8.5 
house 
row-house 70-110 3-7.5 10-12 7-8 
small more- | 90-150 6-11 10-17 10-11 
family 
house 
large more- | 140-260 10-18 14-24 10-16 
family 
house 
| tower block | 400-900 28-45 25-65 20-30 
  
  
  
  
  
  
Table 2: Initial values for building type classification. 
717 
   
Another way to determine the values is to manually select 
buildings which types are known. The Table 2 shows the initial 
values. 
With these values from the training data set all buildings of the 
whole data set are classified. After the classification a visual 
inspection takes place. Then, the values are iteratively 
improved until all buildings are correctly classified. 
  
  
+ 
> 
EN > 
Result of the classification with initial values. 
Brown buildings are classified as large more- 
family houses. Yellow buildings are classified 
as small more-family houses. Red buildings are 
not classified. 
Figure 2: 
Figure 2 shows the first classification with the approximate 
values. There are small and large more-family houses in this 
area. Many small more-family houses (red) are not assigned to 
the right building type (yellow). This is due to the fact that the 
initial values for the area and the height were a little too small. 
The large more-family houses are already assigned to the 
correct class (brown). 
c 
  
Result of the classification with adapted values. 
Brown buildings are classified as large more- 
family houses. Yellow buildings are classified 
as small more-family houses. Red buildings are 
not classified. 
Figure 3: 
In figure 3 are the results for the adapted values from table 3. 
Now most buildings are correctly classified. Only garages, 
subterranean garages and commercial buildings are not 
assigned to a class.