441 
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008 
The surface reflectance is obtained by using the coefficients in 
Table 1. with the following formula (Gonzales, et al., 2006): 
Y = Xa x {measured radiance) - Xb 
Acr "% + XcxY ) 
where Xa, Xb, Xc = coefficients simulated by 6S 
Acr = atmospherically corrected reflectance 
After atmospherically correcting the image data, the spectral 
backscatter response curves are derived for the different surface 
materials measured at the field sites to use as reference spectra 
in the further processing of the Proba CHRIS image. 
Next, the minimum noise fraction (MNF) transform is applied 
to the image data. This reduced the dimensionality of the 
dataset while retaining a small number of noise-free 
components. Consequently, the computational requirements for 
subsequent processing have been reduced. The first 16 
transformed bands showed high variance in the MNF 
Eigenvalue plot. The data is reduced from 62 spectral bands to 
16 transformed bands. 
The reference spectra are processed to provide an average 
spectrum for minimizing the noise in the data. This data is used 
for the end member selection. In order to match the reference 
spectra with the image spectra derived from the reference sites 
spectral angle mapper is used. The spectral angle mapper 
determines the spectral similarity between the two spectra by 
calculating the angle between the spectra and treating them as 
vectors in n-D space. It is relative insensitive to illumination 
and albedo effects when applied to calibrated reflectance data. 
This technique compared the angle between the end member 
spectrum vector and each pixel vector in the 16-D space of the 
transformed data. The smaller angles are representing closer 
match to the reference spectrum. The not matching pixels are 
not classified. 
After the classification process a land cover map of the £ankiri- 
Eldivan area is produced. This interpreted map of the study area 
is used to verify the lithological units. Finally, a geological map 
is produced. In the verification part, the laboratory spectra of 
the samples collected in the field and the geological map of the 
study area are used. 
CONCLUSION 
Remote sensing is a very useful tool for regional geological 
mapping. This study covered the utilization of Proba CHRIS 
hyperspectral image for lithological mapping of Eldivan- 
£ankiri part of Rankin Basin. Additionally, the pros and cons 
using hyperspectral data instead of multispectral data are 
examined. 
The Proba CHRIS has 62 spectral bands with narrow bandwidth 
which increases its discrimination ability. But it has some 
disadvantages over a multispectral image data. The processing 
time takes more time and effort. 
Unfortunately, Proba Chris has no SWIR (shortwave infrared) 
bands. The SWIR bands are useful for lithological 
discrimination and are widely used in geological applications. 
Other than this inherent constrains, the area is imaged in the 5 
acquisition angles at each acquisition. The coverage of these 
different angles varies. The best fitting areal coverage has been 
the -55 acquisition angled image data. But the spatial resolution 
has increased than the nadir acquisition that caused trouble in 
identifying features during geometric correction and spectral 
end member selection. 
A hyperspectral remote sensing study is nothing without a field 
spectroscopy study. Although great effort has been given for the 
planning of field study there have been some setbacks at the 
field and processing of image data due to the unstable weather 
conditions. So, it is important to collect as many measurements 
as possible from diverse sites of the study area not to end up 
with no data. 
The Proba CHRIS showed excellent results in land cover 
mapping but not as good result in lithological mapping. 
Nevertheless, it has been a valuable experience both the field 
spectroscopy study and the processing of Proba CHRIS image 
data. 
REFERENCES 
Bertels L., et al., 2006. HyperTeach, Training in Imaging 
Spectroscopy - Theory. VITO, pp. 56-69. 
Cutter, M., A., 2006. HDFclean V2 software. 
http://earth.esa.int/object/index. cfm?fobjectid=4409 (accessed 
22 Apr. 2008). 
Dellaloglu, A.A., Tiiysiiz, O., Kaya, O.H. and B., Harput, 1992. 
Kalecik (Ankara) - Eldivan - Yaprakli ((fanktri) - iskilip 
((forum) ve Devrez (fayi arasindaki alanin jeolojisi ve petrol 
olanaklari. TP AO Rap. No. 3194, Turkey. 
Gonzales, L., et. al., 2006. Msixs software. http://www- 
loa.univ-lillel.fr/Msixs/msixs_gb.html (accessed 22 Apr. 2008). 
Kaymakqi, N., 2000. Tectono-stratigraphical evolution of the 
(fanktri Basin (Central Anatolia, Turkey). Ph.D.Thesis, Utrecht 
University, Netherlands. Geologica Ultraiectina, no. 190, 248 
pp. 
Tiiysuz, O. and A.A., Dellaloglu, 1992. (fankiri Havzasimn 
Tektonik birlikleri ve jeolojik evrimi. Tiirkiye 9. Petrol Kongresi 
Kitabi, Turkey. 
Tiiysuz, O., 1993. Karadenizden Orta Anadolu'ya bir 
jeotravers. Tiirkiye Petrol Jeologlan Biilteni, 5(1), pp. 1-33. 
Tiiysuz, O., Dellaloglu, A. A. and Terzioglu, N., 1995. A 
magmatic belt within the Neo-Tethyan suture zone and its role 
in the tectonic evolution of northern Turkey. Tectonophysics 
243, pp. 173-91. 
Van der Meer, F., and S., De Jong, 2001. Imaging Spectrometry: 
Basic Principles and Prospective Applications. Kluwer 
Academic Publishers, Dordrecht, Netherlands, pp. 44-63.