International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004 
  
  
Figure 5. Histogram depicting the occurrence (special extend) 
of each class. 
  
  
  
  
  
  
  
  
  
Thematic Class | ID " em ia H a St. dev 
Lake 7 10.6 91 26.7 
Cultivated 3 5 | 3602 1614 283.3 
Cultivated 2 4 | 3706 1597 286.5 
Cultivated 1 3 [| 409.1 1601 274.3 
Cultivated 4 6 | 425.1 1601 316.2 
Barren 2 481.1 1599 389.9 
Mixed Forest | 8 | 481.6 1618 413.9 
Fir Forest | | 9788 1610 357.9 
  
  
  
  
  
  
  
Table 2. Elevation statistics for the thematic classes (presented 
in increasing mean elevation order) 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
Thematic Class | ID en "co m St. dev 
Lake 7 1.52 28 3.68 
Cultivated 1 3 14.00 43 7.82 
Cultivated 3 5 14.02 44 8.27 
Cultivated 2 4 14.13 43 7.70 
Cultivated 4 6 15.85 41 10.65 
Barren 2 16.06 45 9.88 
Mixed Forest 8 17.14 45 9.41 
Fir Forest 1 23.20 43 6.95 
  
  
Table 3. Gradient statistics for thematic classes (presented in 
increasing mean gradient order) 
The table 2 indicates that fir forest terrain class presents the 
greatest height, while mixed forest and barren terrain classes 
follow. The cultivated landcover types are developed in lower 
elevation. These finding are in accordance with the local 
geomorphometric conditions of the study area (high mountain 
and limited plains). The most interesting finding of Table | was 
the relative high elevation of the barren terrain class. 
The table 3 indicates that fir forest and mixed forest present the 
higher gradient values, followed by the barren terrain class. The 
explanation given is that the forests are developed on the 
highest areas which are quite steep in order to be protected by 
human activity. Human activity (cultivated lands) is limited to 
plain areas or to areas with lower gradient value. The high 
gradient of barren terrain class indicates that these polygons are 
of high risk relative to flash floods. That is why the relative 
496 
distribution of barren polygons with high gradient/height values 
will be identified and studied in more detail in the following 
section. 
2.6 Title 
A connected components labelling algorithm is applied to the 
barren terrain class ( Figure 6). 
  
  
  
  
  
Figure 6. Barren class 
A total of 1314 objects (adjacent pixels of the barren terrain 
class) were found with various size dimensions as it can be 
observed in figure 7 
  
[~ Graph Type —— r- Mode — 34 j- Summary Statistics ——— 
| E i idth - ; a 
| @ BarGraph H # NonCumdative |] Class width: 1 dí: 23030 
Ï © Line Graph i | Displaymin: 1 Actual min : 0 
| "io ion I A | Displaymax: 1314 Actual max: 1314 
Le || € Cumulative | 
| € Area Graph il 
| Mean: 572.403 Standard dev: 358.1068 
Histogram ot 02avmnoedafos forepits. ccifore 
  
-—1..:.100 7200 300. 400 500 600 700 $800 900 1,000 1,00 1200 1300 | 
Figure 7. Object dimension for the barren class 
Mean gradient and mean elevation as well as roughness (the 
standard deviation of elevation) and local relief (minimum — 
maximum elevation) were computed for each object of the 
barren terrain class (see Table 4). Elevation is expressed in 
meters and gradient in degrees. 
  
  
  
  
  
  
Size in Mean Mean 
ID i : ; 
pixels elevation gradient 
309 1327 1373.2 29.51 
1 958 978.9 13.71 
677 925 432.0 16.98 
446 796 896.2 23.30 
763 730 581.8 27.13 
  
  
  
  
  
  
Inte