pd.  ——— BEE? A a“ % 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B-YF. Istanbul 2004 
also common landuse types. Related to the special climatical 
and soil facilities there are several native vegetation spots 
inserted into the cultivated fields as well. 
Data 
Aerial photos, hyper- and multispectral images were taken 
from the test site. Hyperspectral records were obtained using 
a new 80-channeled aerial spectrometer (Digital Airborne 
Imaging Spectrometer /DAIS 7915/. We took additional 
images of air and ground with a TETRACAM ADC wide 
band multispectral camera, which can sense green, red(635- 
667nm) and near infrared(835-870nm) bands. The size of 
pixels is approximately 5 by 5 meters. In 2002 the soil 
genetic and surplus water maps were constructed with GPS 
support. We also own the 3D digital relief model based on a 
contour-map with scale of 1:10000. The evaluation of the soil 
conditions was completed using an air-photo taken in 2000 
with scale of 1:10000 and with a weed and vegetation map of 
the area in the spring and the summer aspects taken in 2002 
as well as in 2003. 
Data processing 
Vector data were stored and processed by ArcGIS while the 
hiperspectral data were processed by ENVI. Object 
boundaries and actual land-cover classes were digitized as 
ROI (Region of Interest) by ENVI. The ROI files with the 
ground truth were used to validate the result of both 
classifications by calculating confusion matrices and the 
overall ^ accuracy. Furthermore after multispectral 
classification image processing was calculated by ArcGIS. 
Integrated GIS database and remote sensing data were 
georeferenced by Hungarian Georeferenced System (EOV ). 
RESULTS 
For evaluating the landuse types we reflection values of the 
visible, near and middle infrared ranges of the DAIS image 
(400-1800 nm) were considered. For studying the state of the 
vegetation the wavelength range between 400-2500 nm is the 
most suitable one (Zilinyi, 1990; Turner et al., 1999), which 
is can be explained most likely by the spectral character of 
the green leaves. In this range differences spectral profiles 
can be observed, with which help not only the soil from the 
vegetation, but the various native plant associations can be 
well separated (figure 1). 
Reflectance of analyzed classes 
  
  
  
  
  
  
450 
400 
350 —— 1. alfalfa 
8 300 
5 250 — 2, sugar beet 
3 200 
= 3. maize 
= 150 
100 4. maize 
85 (weedy) 
5 — 5. meadow 
SS DON (q qe d + ap BS — 6. bare field 
bands (400-1800nm) 
Figure 1. Reflectance of analyzed classes 
Our previous knowledge about occurred soil types, cultivated 
plant types with known number of crops provided us 
opportunity to study the reflectance values of different plant 
associations according to the above defined variables. 
87 
Through the field study bounded with GIS we sorted out 
homogeneous vegetation spots as teaching regions for 
supervised classification, by which the similar plant 
categories were identified (figure 2).