2 on the 
-ANDSAT 
notos. 
Delta and 
ric data. 
e ERS- 
he 1975 
parallel 
organic 
Water 
on the 
> great 
  
ng line of 
e Remote 
over the 
at deltaic 
  
3. PRELIMINARY ACTIONS AND PRE- 
PROCESSING: 
S Sisto and homogenising-compatibilising of 
endogeneous and exogeneous parameters: 
- the 3 multitemporal recordings SAR-ERS-1 PRI; 
- repetitive recordings LANDSAT MSS & TM and 
SPOT; 
_ aerial repetitive panchromatic photos, IR and 
colour IR; 
- topographic maps with contour lines at 0.25m, 
pedological maps; 
á piezometric profiles, hydrological data, meteo 
data, quantitative and qualitative profiles, coastal 
evolution determinations; 
b) The three ERS recordings rectifying, taking into 
account the first one; 
- The registration of the LANDSAT MSS & TM 
and SPOT recordings; 
  
  
Fig.4.&S.Classification of the different turbidity level waters on the 
Landsat TM(1,2,3,4) image, from August 1989.& Classification of the 
different turbidity level waters on the Landsat MSS image, of July 
1975. 
- The fusion between MSS & TM and SPOT 
recordings with SAR - ERS taken as reference (the 
best resolution recording has been taken as base) 
using polynomial registration; for a high accuracy 
of registration - fusion of the aero-satellite images 
bi have tried to modify their geometry by enlarging 
them; 
- Marine drilling profiles and their afferent classical 
determinations positioning on the aerial satellite 
recordings; 
- Radiometric seasonal profiles, built on 
characteristic and suggestive directions, basically 
on the “Great Deltaic Profile” and over different 
ecosystems (see fig 3); 
559 
3.2. Preliminary processing of SAR - ERS 
recordings, taken as relative reference level: 
- ERS recordings segmentation: 
(a) on ecosystems basis; 
(b) on hydrographic basins and microbasins basis; 
(c) visual, with successive adjustments using pdf 
(Probability Density Function); 
- Speckle attenuation using: 
- adaptive filters (Lee, Kuan, Frost, Gamma 
filters); The most adequate filter has been found to 
be the Gamma filter. 
- Geometric filters; these filtrations are performed 
on the whole recording or on wide hethero-geneous 
areas, respectively on image segments, with 
visualisation before and after the application of the 
filters, with comparisons of the obtained pdf; 
(Probability Density Function); 
- Colour - composed RGB and IHS system 
processing, as well as transforms from RGB to IHS 
colour spaces on the whole recording (or wide 
areas) and on image segmentation respectively. 
  
  
EH os 
il ] T-28,25 
  
  
  
  
Harta temperaturilor (TM6) 
Fig.6.Temperature map on the Landsat TM6 image from August 1989. 
4. IMAGE PROCESSING, INTERPRETATION AND 
OBTAINEED RESULTS. 
4.]  Processings using radar recordings 
combinations and recordings from optical domain: 
supervised and unsupervised classifications using 
M.L.R. (maximum likelihood rule) and NN (nearest 
neighbour) algorithms, results comparisons with 
confusion matrix use; 
4.2 Because the principal components analysis 
(PCA) is a statistical technique, that transforms a 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996