anbul 2004 
| 
  
em 
JI 
[US 
vl 
Cn 
3.91 
3.52 
5.03 
  
).39 
1.47 
DER 
8.72 
0.34 
  
rithm 
study zones 
71% of the 
es. Table 2 
dates, year 
ease of the 
' "dunes of 
ovement of 
. It appears 
hus by the 
n 43% of 
the lagoon 
'etland zone 
c Figure 3, 
ous erosion 
it 
year 98 
2.17 
0,90 
6,86 
26%] 
0,67 
3,34 
5,97 
0,24 
11,19 
6,15 
4,92 
10,39 
2.69 
3,62 
0,43 
2,46 
  
  
0,54 
e 
of images 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004 
£8 
  
  
g 
* 
         
   
  
  
SPOT Image 
Copyright 1998 © 
© SPOT Image 
Copyright 1989 
  
2. Classification of the image 
Spot 11-03-1989 
1. Example of image to be 
classified Spot 11-03-1989 
3.Classification of the image 
spot 06-11-1998 
    
  
  
  
  
Wetland 1 Flooded zone Ergue Zone Hu Deposit of dune 
Wetland 2 Turbid Water 1 Old dune BE Vegetation 
@ Wetland 3 @  Turbid Water 2 | Nouakchotian soil Limpid water 
Figure 5: Evolution of the classes factors of risks on the area of Nouakchott 
The flooded zones in the lagoon are those due to water of the referents representing the units of landscape, it would be 
resurgences. It is appearing in the medium of the "zone of necessary to have a larger base of images on the area so that all 
construction". Figure 5 shows the evolution of each unit of the cases of reflectances are presented at the referents map. 
landscape at the three dates. Vegetation surfaces did not grow in More the base of training is rich; more the map of referents is 
this area. exact. 
2 
In this case of application, we are limited to 3 images of the 
% same season and a fourth one for the validation. The spectra of 
from one season to another. 
19 
i the calculated referents are relatively close for the three images. 
8: Bees The map of the referents generated is thus valid only for this 
1 e . season, because reflectances of the units of landscapes changes 
3 
The physical properties of certain units of landscape, like the 
fixed dunes and the city, do not change appreciably, therefore 
their spectra are stable. It was thus easier to characterize them. 
+ À . s SIS CN J CCF ME x 
Figure 5. Evolution of the classes on 3 different dates AKNOWLEGEMENTS 
This work is the result of collaboration between PRODIG 
laboratory (UMR 8586 CNRS) and LODYC laboratory, Paris 6 
university. We thank all the team for their availabilities in 
particular Yagoub Meziane. Photo interpretation is made with 
4.1 Discussions and prospects 
The method of self organizing map classification is based on the 
characterization of the units of landscape from the reflectance 
values. However, these values vary for the same unit according 
to the season and other physical parameters, solar radiance and 
climatic conditions. To arrive at a more exact characterization 
the support of the geomorphologic team and physicists of 
507