coarse satellite orbit as possible, this paper employs precise 
orbits issued by TU Delft University. (f) atm is always not very 
easy to model and estimate, which can be removed afterwards 
as well as (f> noi by filtering technique. In the end we have got 
Qdef 
which is what we are interested in most. 
3 CASE STUDY 
3.1 Data Sets 
The basic data for two-pass D-InSAR is SAR pairs and external 
DEM. The SAR pairs are used to generate the interferogram 
comprising the deformation information. The aim of using 
external DEM is used to remove topographic contribution from 
the interferogram. 
This paper collects 6 ERS SAR SLC images covering Fushun 
city, Liaoning province, and their swathes are all about 
lOOkmx 100km. The time spans May 1993 to Jun 2000, and the 
frames are all 2763. The parameters of the example SAR pairs 
are listed in Table 1, the average spectral overlap is more than 
85%. The external DEM we adopted is USGS SRTM DEM 
issued on the internet (ftp://e0srp01u.ecs.nasa.gov). The DEM 
covering our research area available only has 3 arc-seconds’ 
(90m) sample spacing with 3.7 m standard deviation, seen in 
Figure 1-b. 
Figure 1-a SAR image acquired on July 14 1998 Figure 1-b. Mosaic SRTM DEM covering Fushun city 
No. 
Mission 
Orbit 
No. 
Master 
Slave 
ЛТ 
(days) 
B± 
(m) 
Spectral overlap 
Range 
Azimuth 
1 
ERS-1 
09667 
1993/05/22 
1995/04/10 
688 
-27 
97. 5 
96. 8 
2 
ERS-1 
19530 
1995/04/10 
1996/06/04 
421 
-397 
63.2 
83. 1 
3 
ERS-2 
05869 
1996/06/04 
1998/07/14 
770 
14 
98. 7 
99.3 
4 
ERS-2 
16891 
1998/07/14 
1999/09/07 
420 
65 
94. 0 
100 
5 
ERS-2 
22903 
1999/09/07 
2000/06/13 
280 
251 
76.8 
70.4 
Table 1. Basic parameters of interferometric pairs 
3.2 Data Processing 
To reduce errors from co-registration and flat earth phase 
removal, we used the precise ERS orbit data from Delft 
University of Technology. Consequently, we need execute 
general coarse to fine coregistration between master and slave 
images, and between simulated SAR image from SRTM DEM 
and master image. To suppress noise, the SAR images are 
processed by a multi-look operation of 5 pixels along the 
azimuth direction to obtain a final resolution of 20 x 20m, and 
filter the interferogram with a new modified Goldstein 
algorithm(Ding, 2006) in frequency field, and compared with 
the rectangular filtering, our method improved the quality of 
differential interferogram about 28%. The starting coherence 
threshold is set to the average coherence value 0.12 and the 
interferograms are unwrapped with the iterative disk masking 
algorithm. The unwrapped interferograms are mapped to slant 
range ground surface deformation, and then geocoded and 
projected to the vertical direction so as to reflect height change.