M 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004 
  
Figure 1. Image SPOT V - Iraq, Assur ruins 
(Copyright SPOT Image, CNES 2002) 
e Radiometric analysis: with the available tools the multi- 
spectral image, which is aimed at drawing the land’s use 
map and identifying the old construction material present 
on the surface, is classified. Actually some agricultural 
works (with plough) crumble the superficial part of the 
buried structures, and consequently the surrounding land 
has some traces of construction materials which could be 
recognised through the different spectral firma. 
» Interpretative analysis: it is a process developed by the 
researcher on the basis of his/her own knowledge of the 
historical and human settlements context of the area. The 
working platform is the GIS/SIT one, and a wide database 
with information collected from a bibliography and a 
survey is available. In this phase the data received from the 
satellite image could be positioned both on the spatial and 
temporal dimension. 
How relevant is the geometric accuracy of this analysis ? It is 
surely quite relevant, but not so much as we could actually 
imagine. Let’s analyse the reasons of this assertion. 
The study of land is normally carried out at a medium-large 
scale. On 1:5.000 scale, for example, the attended map's 
accuracy in cartography is 1 meter, therefore it could be 
compared with the pixel of the panchromatic Ikonos band. (Fig. 
2) 
  
Figure 2. Image Ikonos: ancient human evidence (wall) 
(Copyright Space Imaging 2001) 
However if we work with multi-spectral data, the pixel on the 
ground measures 4 metres, therefore the localisation of each 
pixel overcomes the defined planimetric tolerance. 
In our applications we have verified the validity of the pan- 
sharpening between the  panchromatic band and the 
multispectral ones, both from a metric and radiometric side. 
Actually it has been applied in the surveys at 1:5.000 scale, and 
also 1:2.000 scale with results that we consider interesting. 
Even if we are aware of the possibility of some errors, it is often 
necessary the use of the satellite image over the limit of its own 
resolution: therefore the georeferentation, which is acceptable 
for a working scale 1:5.000, could not be acceptable for a larger 
working scale. 
Which data (historical and not) could be considered for the 
comparison with information received from the study of the 
satellite image? 
In many cases these data are characterised by a relevant 
topographic imprecision, due to direct measurements carried out 
with a large approximation and often without a cartographic 
framework of reference. 
Information collected more recently is more reliable, even if 
most of the available material is characterised by a position data 
rather imprecise. 
Therefore our question is: how has to be the level of accuracy of 
the GIS/SIT system which includes all this information ? 
Most probably the medium planimetric error could not only 
referred to the working scale (with the often used criteria of 
error due to graphicism), but it will be compared by the 
accuracy which is specific for each group of information. 
In this perspective the georeferentation of the satellite image 
maintains a relevant role, but the accepted tolerance could be 
   
   
    
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
   
   
   
  
  
   
    
   
    
   
   
   
   
   
   
   
   
    
  
   
    
    
   
   
    
  
  
   
[nternati 
HE 
higher th 
Followin 
which € 
standard: 
happen 
Mesopot 
survey € 
reference 
The avai 
rare: the 
necessar: 
measurel 
restitutio 
to acquir 
Consider 
studies € 
slight le 
georefer 
rectificat 
the loca 
dimensic 
studies. ( 
  
d 
Informa! 
verified, 
directly 
("navig; 
often ab: 
The pre 
cases the 
informat 
ortho-pr 
It is un 
paramet 
on a sol 
control | 
In a rec 
before. 
archaeo| 
panchro