at 
ly 
of 
he 
nd 
on 
ed 
he 
€, 
ge 
ge 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B-YF. Istanbul 2004 
  
The satellite images were corrected for the influence of 
atmosphere and topographic relief. These data were geo-coded 
by the help of rectified SPOT data according to the European 
Datum (ED50) and a UTM projection to combine and analyse 
with the other data. In the end, the atmospherically, radio 
metrically and geometrically corrected / geo-coded images were 
supplied by ERDAS software. 
The purpose of digital land cover classification is to link the 
spectral characteristics of the image to a meaningful 
information class value, which can be displayed as a map so 
that resource managers or scientists can evaluate the landscape 
in an accurate and cost effective manner (Weber and Dunno, 
2001). In this study, the Maximum Likehood supervised 
classification algorithm was used (Lillesand Kieffer, 2000). The 
best way to compare images from different dates is to classify 
the two images separately and to compare the statistical results. 
The classification of satellite images were utilised in land cover 
analyses and in determining change between the land use before 
the fire and that after the fire. 
In order to obtain effective and more accurate conclusions, 
mathematical operations in the GIS analysis were formed. The 
input information on forest fire influencing factors indicates the 
weights in the fire risk in an area. The factors were analysed in 
the following order of importance: vegetation type, slope, 
aspect, distance from roads and settlements. First classes 
represent high risk places and last classes represent the minor 
risk place. Each class has different weights. 
The vegetation types were classified according to the moisture 
context that has an influence on breaking out forest fire. For 
example, the vegetation type that is very dry is the most 
flammable whereas the fresh type is inflammable. 
Slope influence behaviour of fire was evaluated the second 
highest weight. Fire travels most rapidly up-slopes and least 
rapidly down-slopes. Slope classes were created according to 
this rule. Aspect was assigned equal weight with slope. Since 
the sunlight is much more reflected on the slopes in the south, 
fire breaks out fast and spreads in the south sides. 
Distance from roads and settlements were evaluated the third 
highest weight. The risk factor decreases farther from these 
places. It means that a zone close to these places were evaluated 
a higher rating. 
Water bodies areas do not affect the forest fire risk. These 
zones have no weights in determination of fire rating class. 
The equation used in GIS to determine forest fire risk places is 
shown in equation I. 
. RC=T*Vr+ 5%(S+4) + 3%(Dp+Dy) (1) 
In this equation, RC is the numerical index of forest fire risk 
zones where Vr indicates vegetation type with 5 classes, S the 
slope factor with 5 classes, A the aspect variable with 4 classes, 
D, and Ds indicate distance factor from road and settlement. 
Finally, based on these analysis carried out, a fire risk zone map 
was produced. 
35 
3.3 GIS 
Geographic Information System (GIS) has also developed 
functions such as analyzing available information and using 
them as a decision and a support system as well as it compiles 
the information as a whole and stores it. 
In that way, GIS analyses show the spatial distribution of the 
observed forest condition and also may help to find cause 
parameters for the observed — and classified — forest decline 
phenomena. The link between remote sensing results and an 
efficient forestry GIS can therefore work as a tool for an 
operational and practically oriented monitoring system for 
forest damage assessment and management. It may play an 
important role as a planning tool for forest and land-use affairs 
in a broader sense (Faber et al., 1994). 
Parameters Weight Classes Factors Fire Rating 
classes 
Vegetation 7 Very dry 5 Very High 
Dry 4 High 
Moist 3 Medium 
Fresh-like 2 Low 
Fresh 1 Very Low 
Slope 5 >% 35 5 Very High 
% 35-25 4 High 
% 25-10 3 Medium 
% 10-5 2 Low 
<%5 1 Very Low 
Aspect 5 South 5 Very High 
West 4 High 
East 3 Medium 
North 2 Low 
Distance from 3 < 100m 5 Very High 
roads 
100-200m 4 High 
200-300m 3 Medium 
300-400m 2 Low 
> 400m l Very Low 
Distance from 3 « 1000m 3 Very High 
settlements 
1000- 4 High 
2000m 
2000- 3 Medium 
3000m 
> 3000m 2 Low 
Table ! The weight of parameters in determination of fire risk 
areas.