sas, and 
M). 
d extent 
is), and 
data is 
ydwater. 
jd. This 
spectral, 
its were 
nmental 
ge from 
specific 
3,4,5,7, 
leu, 25 
reas for 
Yt. 
ference 
mely, a 
NDSAT 
ed to a 
Potcoava, Mariana - Camelia 
The flight modelling procedure used for airborne geometric correction has two attractive features: 
1) the rectified imagery is registered to the topographical map (or DEM); 
2) the flight line parameters allow one to determine the three dimensional orientation of the sensor target line of sight 
for each corrected pixel. 
Since the DEM is automatically part of the integrated data set, it was possible to incorporate information related to 
topography in image classification. For this purpose, elevation, slope, and aspect channels were utilised as additional 
features. 
The NDVI derived from LANDSAT data is used to estimate different vegetation state. Since the pixels of the images 
may be totally or partially by clouds, which obviously triggers errors in correctly estimating the characteristics of the 
water surfaces, an obligatory stage of pre-processing satellite data was to identify and eliminate them. The 
recommended procedure in view to eliminate the cloud-contaminated pixels is based on computing the vegetation 
normalised difference index (NDVI), with relation: 
_ TM4- TM3 
TM4+ TM3 
NDVI (1) 
Thus, the areas with clouds have an NDVI ranging between -1 and 0, whereas the vegetation have an NDVI as close to 
1 as possible. Through eliminating these values of the NDVI index specificated for cloud contaminated pixels, water 
bodies might be identified easier. 
In addition, during the resembling process of the geometric correction of SAR data, radiometric corrections for range 
attenuation and antenna pattern were included. Lee-Sigma filtering of the SAR imagery was used to substantially reduce 
speckle. 
5. RESULTS 
5.1 Futures 
The purpose of this work is to show how the combined use of Landsat TM data with ERS-1 SAR data can contribute 
towards a deeper knowledge of the area under consideration. 
1) Landsat TM 
The city is star-shaped, with a historic centre in the middle from which the suburbs extend outwards. The presidential 
palace in the city centre can be made out, on the right of which is the main road leading to the palace and an area under 
construction. The national airport is to the north. The main road systems consists of three concentric circles around the 
historic centre in which the railway station, to the south, is easily visible. A series of roads fanning out like the spokes 
of a wheel connect the suburbs to the centre. 
The river Dimbovita, a tributary of the Danube, which crosses the northern area of the city, shows evidence of 
meandrification, which has given rise to some small lakes in the eastern section of the image. Two small lakes are 
present in the centre and lower area of the image, while an artificial, circular-shaped humid area, is visible in the south. 
Vegetation visible in the image is mainly agricultural. The fields are of different sizes and are rectangular. In the north, 
two larger areas are used for tree plantations. 
Different band combinations (figure 2) allow examination of various features of the area shown in this image: 
- Bands 3,2,1 (red, green, and blue) - this band combination allows us to show urban areas and roads, in white. The lake 
sited in the southern 
Portion of the image shows different colour due to the sedimentation content. 
- Bands 4,5,3 (near infrared, medium infrared, red) - this band combination is useful for distinguishing different types of 
agricultural vegetation, which appears in different shades of green, orange and ochre. This combination is also useful to 
obtain further information about roads, which show up in a lighter blue than urban areas; 
- Bands 5,3,1 (medium infrared, red, blue) - this band combination clearly shows urban areas in pale blue on a brown 
background which is farmland; 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000. 1187