International Archives of the Photogrammetry, Remote Sensing and Spatial 
Rational Polynomial Coefficient (RPC) file for 
orthorectification; 
-  Panchromatic photomosaic, realized by USA Army, at a 
nominal scale of 1:25,000, dated July, 20 1943; 
-  Panchromatic orthophotomosaic, realized by EIRA (Ente 
Italiano Riprese Aeree), at a nominal scale of 1:20,000, 
dated year 1965; 
-  Topographic map at a scale of 1:50,000 (Quadrant Susah, 
by USA Army, dated year 1964). 
      
    
geo s 581000 82000 SRM s44000 ASA ë 
= ee "vd 
3637 
3636000 
3630000 
2,250 3,000 
Meters 
581000 £82000 SRM £54000 S85000 
  
  
2 LX 
578000 £79000 SHON) 
Figure 2. QuickBird panchromatic satellite imagery 
3.2 QuickBird Image pre-processing 
The satellite scene has been purchased standard type and it 
represents a subset of 8x8 Km“. As a consequence the fusion 
between the panchromatic and the multispectral imageries (at 
lower spatial resolution) it must be done. The Resolution Merge 
between them has been executed by using the Principal 
Component method on Erdas Imagine 8.7 software. Moreover 
the standard type scene is only rectified and it has to be 
orthorectified in order to make it available for geological 
interpretation, by removing the effects of geometric distortion, 
dues to the orography and to the acquiring pushbroom manner. 
Considering the high spatial resolution (70 cm), accurate GCPs 
must be collected and a proper Digital Elevation Model (DEM) 
had to be created. 
survey and DGPS 
3.3 Fieldwork (geological 
measurements) 
Geologic and geomorphologic survey has lead to the production 
of preliminary fieldwork cartography at a scale of 1:10,000 and 
the measurement and the analysis of a complete stratigraphic 
column located along the road connecting Shahhat to Susah 
(Figure 3). 
During the fieldwork DGPS measurements necessary for the 
orthorectification of the satellite data have been gathered. Two 
Leica System 530 instruments were utilized for measuring 83 
GCPs collected in RTK modality all over the scene. Because of 
the lack of known reference points, the accuracy of the 
measurements has been verified, during DGPS post-processing, 
Information Sciences, Vol XXXV, Part B7. Istanbul 2004 
by crossing the data with simultaneous records taken in the 
closest. permanent reference stations of Noto (SR-IT) and 
Lampedusa (AG-IT); 60 cm mean displacements has been 
observed. 
     
  
  
  
     
    
  
Al Abraq 
Formation 
Darnah 
Formation 
  
\ 
      
     
    
  
     
Al Bayda 
Formation | x Marly Limestones rich 
in macro-foraminifers 
(nummulites s.l; 
    
   
  
  
  
  
i 
i 
(Algal Limestone / I 
Member) l 
| Darnah 
| Formation [5 
Marly Limestones with red seaweed 
(Shahhat Mari { 
Early Oligocene Member) \ i 
E T Pu 4 
Late Eocene 
(Priabonian) 
  
Figure 3. Stratigraphic column measured along the road 
connecting Shahhat to Susah 
3.4 QuickBird Image processing 
The orthorectification of the satellite data has been carried out 
by using the DGPS GCPs and the DEM created from the 
topographic map. This last was produced by digitising 20 
meters interval contour lines, spot heights and hydrography 
from the topographic map at a scale of 1:50,000. According to 
National Imagery and Mapping Agency (NIMA, 2004) from a 
50,000 scale topographic map, a Level 2 DEM can be created. 
The spatial resolution of such a DEM is approximately 30 
meters and an RMSE of one-half contour interval is the 
maximum permitted (USGS, 2004). According to Kolbl (2001) 
the orthorectification of high resolution satellite images can be 
obtained from a DEM with an accuracy of approximately £10 
meters. It means that the created DEM could be utilized for 
QuickBird image geometric correction (Toutin and Cheng, 
2002). 
Then, initially the orthorectification of the image has been 
performed by using à rational “non-rigorous” method by means 
of the RPC file, the GCPs and the DEM. The accuracy of the 
correction was assessed by using 21 checkpoints among the 83 
collected during DGPS survey; it averagely results of 
approximately 6 meters with a minimum of 2 and a maximum 
of 14 meters. The rational method doesn’t allow a perfect 
orthorectification of the satellite image especially in non-flat 
areas; the lack of the “basic” type full scene and of the image 
acquisition geometry can only be mediated by the polynomials 
(Volpe and Rossi, 2003). Finally the satellite scene could be 
correctly orthorectified in respect of the terrain only by using 
polynomial functions of high order, the so-called rubber 
sheeting method. 
This operation required all the available GCPs because high 
polynomial functions fit locally to the GCPs but not the area 
between GCPs. This choice made impossible the positional 
accuracy assessment by checkpoints but, at the same time, it 
made allowable the internal imagery deformation. 
1102 
Inte 
3.5 
The 
cou 
visi 
cou 
geo 
ster 
sec 
PCI 
arti 
the 
ima 
valu 
al, 
pixe 
DP 
whe 
The 
PCI 
with 
paraı 
= US 
th 
N 
3.6: 
After 
the cl 
obser 
anagh 
photo; 
Leica 
8.7 so 
Metric 
focal | 
knowr 
formu 
HS : R 
where 
The ca