1 
Ali Farrag, Farrag 
  
27 ES | A a a —ÀÁ spas qu 27 IS'N 
       
     
      
  
  
500p: E 
-Asyut Cingpany - 
-Reftne 
nw ; Ll | Me V Z2 m 
498] pm eT Cr eed fA nas m c e] es 
  
  
  
  
620 625 
31 0Q0' E 31 05" 31 10° 
emu e mee SKM 
Figure 1- Location map showing the relative position of the cement factory and the surrounding environment 
3. METHODOOLOGOY 
A combination of image interpretation, unsupervised classification and image thresholding, techniques incorporation 
with ground measurements were applied in order to analyze the information contents of SPOT images. The images were 
analyzed at its original spectral condition and the process did not include radiometric correction, such as haze 
correction. This can avoid distortion of the information contents of the original image. IDRISI, grid-based geographic 
information and image processing, software was used as an image processing tool in this study. 
The process and analysis of each of SPOT images consisted of three steps. Firstly, the images were geometrically 
corrected. Secondly, a composite image was constructed and examined visually then unsupervised classification was 
carried out. Thirdly, the data of the green band which is the most sensitive band for such pollution was examined and 
analyzed. Then the obtained results were presented in form of maps and charts. These steps can be briefly explained as 
follows: 
STEP 1: Geometric correction of SPOT images is essential for carrying out quantitative analysis. The bands of each 
image were rectified to the Universal Transverse Mercator (UTM) coordinate system using a second order polynomial 
transformation. The measurement of the ground coordinates of the control points was carried out in the field by 
applying Differential Global Positioning System (DGPS). The residual Root Mean Square (RMS) error was 18 m (0.9 
pixel). Each of the three bands was individually geometrically corrected using the same set of control points. The 
resampling process of each band was carried out using nearest neighbor method (in order to avoid any smoothing to the 
original image data). A square window of 458 km”, corresponding to a part of the map shown in Figure 1, was extracted 
from the geometrically corrected image. The area represented by this window of the image contain several types of land 
cover as shown in the classified images given in plates 1 and 2 for 1995 image and 1998 image respectively. This 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000. 47