International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004 International 
ntern 
Hera SEE 
In ASTE 
spatial re 
One SWI 
different 
operation 
SWIR pb 
(15 m) b: 
in more 
SWIR cl 
channels, 
correctio: 
  
Marmara Sea 
   
    
E NU CER i 
Mediteranean Sea: 
  
The next 
atmosphe 
check th 
the same 
be attain: 
   
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
a UTM( 
3l grour 
resamplii 
process, 
General 
DEM (di 
y and Z « 
and + 0. 
were obt 
1.097 pi: 
Followir 
performé 
atmosph 
correctic 
Sample locations SWIR 11% 
::*] Volcanic rocks (andesite, dacite, trachyte) 10 km The nui 
7] Other rocks (metamorphic, sedimentary, ophiolitic rocks) = 
affected 
Figure 1. Location of samples on the geological unit map simplified from 1/100 000 scaled geological map for spec 
effect ar 
3. METHODOLOGY covers only VNIR and SWIR spectral region, TIR of the i 
spectral region of ASTER did not be used. subtract 
During the study, data analyses were carried out using (Crane, 
Erdas Imagine and PCI Geomatica image analysis Kaowt ASTER DEM Topographic used du 
software. In order to find out kaolinite and alunite ocourances Data (1/250 000) map image c 
minerals, two different image processing techniques were (hdf) (1/100 000) as lake 
performed. These techniques are called as band ratioing e | image c 
and spectral indices techniques (Yamaguchi and Naito, as brieh 
2003). After applying these techniques, results were area e 
compared by using selected sites as ground truth. During Ortho- DN to DN al 
the study, an ASTER scene taken at September, 20, 2000 rectified æ| Relative 
was handled to analyze. Figure 2 summarizes the ASTER Dat) Rofloctance E Cm 
algorithm which followed during the study. Soon 
value. 
3.1. Preprocessing 
Band Spectral 3.2. Bai 
; 3 ; Ratios Indices 
ASTER image has 14 different spectral bands which are 
3 VNIR (visible-near infrared), 6 SWIR (short wave Each ol 
infrared) and 5 TIR (thermal infrared) channels differen 
(Yamaguchi et. al, 1998). Each spectral region has | : is a kin 
different spatial resolutions such as 15 m (VNIR), 30 m unit m: 
(SWIR) and 90 m (TIR). The most of the image portion. 
processing methods use same pixel size for all channels region. 
due to pixel based arithmetic operations. In order to Result iati: 
perform image processes, pixel size of different spectral 
regions of ASTER image arranged as 15 m cell size for in 0.6 
all regions by using PCI Geomatica software. This study Figure 2. Schematic diagram of the study. eit 
his in 
techniq 
the ver 
sensing 
78