Michele Crosetto 
  
  
  
—71*— No corr. 
  
Cot. opt. 
Corr rad. 
  
  
  
Correlation 
  
© Q Oc Oc OC coc c c oc = 
V) Quin © Wn QOO wv, cv cow ow c 
- (ye ND. (C Q C ec ww XO 0 O\ ee 
M E E MEE e! 
Distance [m] 
  
  
  
Figure 3: Ascending DEM. Autocovariance function of the height differences (InSAR versus reference DEM) for three 
grids: before correction of the atmospheric distortions (“No corr.”), after correction with optical data (“Corr_opt.”) and 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
-d for with radargrammetry data (“ Corr_rad.”). 
o the 
In the overlapping area between ascending and descending images 8 tie points were collected. The joint calibration was 
erformed using 14 GCPs for the ascending pair, 8 tie points and only 1 GCP for the descending pair. From the 
ff P gp p y gp 
€ sens descending pair we generated a 60 m spacing DEM, that was compared with the reference one (see statistics of the 
noticed processing type “without atmospheric corrections” in Table 6). The global bias (mean error) of the grid can be 
b 8 
teristics considered satisfactory, i.e. the joint calibration resolves quite well the geo-location of the generated 3D grid. There is 
n of the an important decrease of the DEM precision in mountainous areas. The grid, compared with the ascending one 
p g p 8 
1t 505 m described in section 2.1, has lower spatial resolution, covers only one third of the area and is less precise. These 
characteristics are related to the low coherence and to the high compression of the SAR images. 
spheric 
0 coarse Acquisition Date 4 June and 9 July 1993 
are not 7 
netic Baseline Length 160.7 m 
EM has Sub-image range dimension 1500 pixels 
buted in 
th of the Sub-image azimuth dimension 5000 pixels 
rid used Mean coherence of the filtered images 0.41 
InSAR geometry calibration Ascending and Descending Joint Calibration: 
Ms (see 14 GCPs Asc., 1 GCP Desc. and 8 Tie Points 
properly Table 5: Characteristics of the ERS-1 descending InSAR pair. 
e global 
ly). The 
try data Processing Terrain Mean Error Standard Deviation 
)EM are Type type [m] [m] 
properly 
i - 1. 17.96 
le more Without hilly/flat 1.83 
Atmospheric mountainous - 0.22 30.07 
Correction : 
entire area - 1.62 20.63 
[able 5). illy/ - 1.41 14.14 
) Atmospheric Correction hilly/flar 
al of the : : 
: : with mountainous 0.37 20.65 
mes Optical Data 
as 16 by P entire area -0.95 15.37 
adopted 
Table 6: Descending InSAR DEM results. 
  
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part Bl. Amsterdam 2000. 51