the 
"ts. 
are 
ids 
the 
ral 
lon 
| in 
ion 
res 
igh 
eld 
via 
and 
1go. 
ural 
for 
ulti- 
ases 
ise, 
ase 
ures 
are 
a to 
e of 
Fabing, Aline 
model. The main advantages of such methods are the richness of the thematic content and the possibility to describe the 
field processes. But the main limitations are related to the difficulty to quantify the reliability of the information. 
1.5.1 The spatial indicators of forest cover change 
The criteria we chose were population change (i.e., migration, increase, location), accessibility schemes (i.e. road 
network), the land use practices (i.e. logging, agriculture (i.e. shifting cultivation, fuel wood production)), socio- 
economic, political and cultural information i.e. (environmental perceptions) for the human factor (Thumerelle, 1996); 
physical and climatic information for the natural factors. 
The in-depth analysis of such criteria allows the definition of spatial indicators of forest cover change, which were used 
as input in the spatial model. These indicators might be direct and indirect and have to be obtained through remote 
sensing and photogrammetric data as well as through ground research and inquiries. Multivariate analysis requires the 
input of these factors, which consists in selecting those spatial variables that are most correlated with the observed 
changes. The methodology for the identification and analysis of these factors is adapted from a hierarchical structure 
approach in order to compare these results with other researchs (FAO, 1997, TREES, 1998). The figure 2 resumes the 
approach we followed. 
  
Global spatial database 
  
  
Photogrammetric and remote 
sensing data 
Forest cover proportion 
Savannah cover proportion 
Land use proportion 
Socio-economic indicators 
from additional database 
Population change 
Fuel wood exploitation 
Ecology perception and ethnicity 
Cartographical database 
Topography 
Hydrological network 
Protect areas 
Road network 
Human settlement 
Rural poverty indicators Logging concession 
  
  
  
  
  
  
  
Analysis of processes 
  
  
Spatial model to obtain the: 
Deforestation risks map 
Different scenarios 
  
  
  
Figure 2: Theoretical approach for the establishment of the deforestation risks map. 
RESULTS AND DISCUSSION 
As results we obtain maps where areas with recent deforestation indicate different speed of forest cover change (high, 
medium, low) and different forest status (dense, fragmented, low density). These results show where natural forest 
transgresses and in what quantities; how much of this forest has disappeared since 1950 under population and urban 
pressures. We verified these results with field research as physical sampling, inquiries about fuel wood and agricultural 
exploitation and the general perception of this problem. 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000. 423