International Archives of the Photogrammetry, Remote Sensin g and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004 
  
  
  
  
  
  
  
  
  
  
Figure 1. Study area: 
(Trabzon Province with its administration sections) 
2.2 Soil Data 
In Turkey, The General Directorate of Rural Services 
realizing its activities under the Ministry of Agriculture and 
Rural Affairs is responsible for production of soil related 
information. Produced by this institution, maps scaled 
1/100.000 were digitized. Land use types (LUT) are 8 classes 
on this map. Respect to soil damage and restrictions, these 
classes are range from 1. to VIII. Besides, agricultural lands 
are divided into 4 classes. Having I. degree importance, 
agricultural lands are named as absolute agricultural classes. 
These agricultural classes involve I. and II. degree classes 
and the other classes lands having needed irrigation except 
forest lands and moorlands. Having II. degree importance 
agricultural lands involve III. class land use capability lands. 
Having IIl. degree importance agricultural lands involve 
vineyards-gardens and special product lands which are in 
other than L., I. and IIL. class lands. IV. class lands are named 
as the other lands. These classes involve the lands except T., 
IT. and III. agricultural lands (Balci, 1996). 
2.3 Relief Data 
Digital Elevation Model (DEM) was digitized from 
1/100.000 scaled Standard Topographic Maps. Contours on 
these maps are drawn with a 50m elevation differences. DEM 
of the study area was created by using ArcView 3.2 and 
ArcGIS 8.2 GIS software’s. Slope Map (classified in 
percentage) and Elevation Figures were generated by using 
this DEM. Pixel dimensions of this Slope Map are 25x25m. 
When considered in view of slope, three fourth of the total 
arca of Trabzon province consist of areas with a slope rate of 
above 30%. Besides, areas with a slope rate of 50% or above 
constitute 40% of the total area of the province (Reis, 2003). 
2.4 Remote Sensing Imagery and Land Cover 
In the Eastern Black Sea Region of Turkey, the principle 
challenges in generating land cover types are rough 
topographic structure, rich in turns of plant species and 
scattered agricultural lands. Therefore, some difficulties were 
encountered in processing remotely sensed data in the region. 
In this study, a single date image of Landsat ETM+ (Path 
173; Row: 32) on October 19, 2000 were used to generate 
376 
land cover types. Using the image, after extracting an 
application area of approximately 120x90 km covering 
administrative boundaries of Trabzon province, some other 
required further studies were implemented on this area. 
Landsat ETM+ image has 6 multi-spectral bands with 30m 
resolution, 2 thermal bands with 60m resolution and a 
panchromatic band with 15m resolution. 
For the study, a total of ten land cover classes arc chosen. 
These are pasture, water, deciduous, coniferous, mix wood, 
green tea, hazelnut, rocky, settlement and agriculture. 
Training area data for these classes were obtained using 
mobile GPS in two different seasons (for May&September). 
Supervised classification process was realized. During the 
classification process each of the following algorithms were 
tested: Minimum Distance, Mahalanobis Distance and 
Maximum Likelihood with several thresholds. It was seen 
that the Standard Maximum Likelihood Method provides the 
best result with multi-spectral bands (123457) when the 
supervised approach is used. Afterwards, to analyze this 
classification, the overall accuracy percentage, user’s and 
producer’s accuracy and the Kappa Coefficient that provide 
the statistical evaluation of these accuracy values were 
calculated. The overall accuracy of this classification was 
84.68% and the Kappa Coefficient was 0.829. The hazelnuts 
producer's and the user's accuracy were realized respectively 
87.9 94 and 83.5 94. 
3. DETERMINATION OF THE POTENTIAL 
HAZELNUT AREAS 
Turkey state was brought into force a law numbered 2844 in 
1983, named as “The act about planning of the hazelnut 
production and determining of the planting areas”, for 
gaining thc most yield from the hazelnut plants and balancing 
the world markets harvest of the hazelnuts. The last 
regulation was done in January 2002 for constricting the 
hazelnut areas and enhancing the hazelnut productivity. 
According to this regulation the harvesting of hazelnut can be 
done the areas as follows; 
a) in the areas of which maximum elevation is up to 750 m. 
b) in Ill.class agricultural areas and slope is more than 6%. 
c) in IVth or upper class of the land use ability areas 
(Turkey official gazette, 2002). 
Because of the resolution of land use map, obtained from 
satellite image, is 28m the analysis of Potential Hazelnut 
Areas (PHA) is done in grid format of which pixel dimension 
is 28m. In this analysis procedures all data coverage are 
transformed into a pixel dimension in 28m. The criterions 
given above are used to determine the PHA in GIS context. 
The spatial analysis module in ArcView for carrying out this 
aim was used (Esri, 1996). According to this analysis, it is 
assigned that there is total 101.384 ha PHA in Trabzon 
province which is 21.896 of entire the province. 
4. DISCUSSION 
In Trabzon, there is an area of 164.370 hectares between the 
elevations of 0-750 m, which is used as a criterion in 
determining the general PHA. While 101.384 hectares of this 
area is assigned as PHA, it is determined without taking into 
consideration of the existing land cover. When the current 
land cover is considered so as to determine the PHA, it will 
be seen that the PHA will become less than the calculated 
one obtained from Landsat ETM+ satellite image. According 
to the results of analyses with land cover map, as given on 
Inter: 
the T 
overl 
whicl 
consi 
areas 
conif 
extra 
47.20 
is st: 
deteri 
calcu 
menti 
was s 
24,10 
Table 
When 
subtre 
PHA 
that t] 
in T 
agrici 
  
GIRESUN 
  
V 
dE 
TT WW mn 
eo 
bd