'eas without 
ISe a low 
ling degree 
calculated, 
y limited in 
as, disposal 
| allotments, 
nasked. 
Instrument 
EL 
out surface 
on surface 
t and wind 
jan surface 
ed for each 
> calculated 
1g. 
  
ate in °C 
morning) 
  
Stand. Dev. 
  
0,8 
  
0,9 
  
  
  
  
  
  
Figure 1 shows these temperature distributions in a graphical way: 
  
Yn 
Temperature [°C] 
  
50 
  
E3Midday temperature 
B3 Evening temperature 
Morning temperature 
  
  
  
40 
  
30 
20 
  
Bituminous road 
Concrete road 
Paved road Tile roof Bare soil 
Tarmac road 
  
  
Uncultivated field Grass 
Surface cover material 
  
    
Tree Tops 
Cultivated field Meadow Water 
  
  
Figure 1: Temperature Values of Different Surface Cover Materials 
The following conclusions can be drawn: 
- Traffic areas are characterized by high temperatures in the 
evening and in the morning with little standard deviation. In 
average the temperatures are 5 ?C higher than the 
temperatures of all other areas. Therefore it is possible to 
identify these areas by their temperatures. 
- Building complexes are typical to possess high heating 
during day time, and a fast and intensive cooling at night, 
dependent on their roof material. Even roofs with the same 
materials vary considerable in temperature, because of 
different exposition and age of materials. 
- Short-cut grass heats up extremely during day time and 
cools down rapidly after sunset, so the difference between 
midday and evening temperature is much higher than 
between evening and morning temperature. 
- Meadows heat up moderately and reach the lowest 
morning temperatures. 
- Temperatures of tree tops are not very high at midday and 
cool down heavily at night. 
: Water bodies show a very small variation of temperatures 
between day and night because of their high thermal store 
capacity. 
An exact and distinct separation of sealed and unsealed 
areas is not yet reached using the parameter temperature. 
The cooling characteristics of both types are, not as 
expected, partly too similar. 
Less precise rectification in certain stripes cause 
misclassifications in merged data. It is expected that the 
Separation between sealed and unsealed areas will not be 
reached sufficiently because the temperature ranges are too 
Similar and temperatures are influenced by external factors, 
|. €. relief, exposition, and local climate. Though it is quite 
491 
likely to determine traffic areas in the evening as well as 
in the morning data set, because of their significant 
higher temperature. This characteristical feature is very 
important to analyse the data with a multi-step method. 
4.4 Non-thermal Bands of ATM Scanner as an 
Instrument to identify Soil Sealing 
Apart from the quantitative methods sealing degree 
values of urban surfaces shall be obtained by performing 
a detailed land-use classification. Sealed surface classes 
will be separated for buildings and not built-up sealed 
areas, as unsealed surfaces will be divided into short cut 
grass, meadow, trees/shrubs, vegetationless areas and 
water surfaces. This classification will certainly deliver 
additional information about further urban ecological 
research subjects. 
First steps to produce this classification have just begun. 
A hierarchical classification is performed, structured in 
the following scheme: 
- Determination of areas covered with vegetation 
applying a threshold operator upon the vegetation 
index NDVI. 
- Subdivision of vegetation areas in grass, meadow and 
trees/shrubs with a Maximum - Likelihood - 
Classification (ML). 
- Separation of traffic areas (not built-up but sealed) 
using a threshold operation upon the morning 
temperature data set. 
- Separation of unsealed, vegetationless areas 
performing a ML - Classification. 
- Remaining pixels are mostly building areas. 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996