The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Voi. XXXVII. Part B7. Beijing 2008 
174 
2.2 Data set 
3. INTERFEROMETRIC PHASE MODEL 
The 26 scenes archived complex (SLC) images (Track:3,Frame: 
2979) are chosen as the long-term experimental data acquired 
by C-band ERS-1/2 sensors covering the time span from 1993 
to 2000 in the test area. Due to enough dataset, a unique image 
is selected as the master image to generate temporal 
interferograms. The interferometric combinations ensure the 
uniform temporal and spatial baseline distribution, as shown in 
figure.2. 
Baseline Distribution 
Figure 2. Interferometric combinations in long time series 
The latest 9 scenes of complex (SLC) images (Track:3,Frame: 
2979) acquired by C-band ASAR sensors onboard the Envisat 
covering the time span from 2003 to 2005 in the test area are 
chosen as the short-term experimental data. In order to exploit 
sufficient interferograms, multiple master images are chose to 
generate multiple interferometric combinations in overlapping 
time intervals with the uniform temporal and spatial baseline 
distribution, as shown in figure.3. 
Baseline Distribution 
The two PS-InSAR techniques are both model-based, firstly the 
interferometric phase model can be expressed as the sum of the 
following four contributions, 
(f>., = è + è + a + n . 
' int ' topo ' dejo atmos noise 
(1) 
Where (j) topo - the phase caused by topography 
$ defo = phase caused by a displacement of the 
target along the line of sight (LOS) of sensor 
a atoms = the phase caused by atmospheric artefacts and 
imprecise orbit data. 
n noise = the phase caused by decorrelation noise. 
By removing the topography phase, the differential 
interferometric phase model can be obtained in the Equation.2. 
Ó A(r = Ójr + Ó. TíZ. ~bT2 . (i\ 
'dip ~defo 'topo-error atmos noise \¿) 
Where Vtopo -error is the phase caused by topography error. It is 
expressed as Equation.3 
u topo-error 
An B, 
£ 
/1 R -sin# 
(3) 
Where /1 = radar wavelength 
R = range distance from master sensor to ground 
B L = the local perpendicular baseline 
6 = the local incidence angle 
£ = the DEM error 
As the ground deformation in urban is a low velocity process, a 
linear model can basically fit the deformation change. So (¡) defo is 
explicitly written as Equation.4. 
«W. =-- vT W 
Where v is the deformation velocity in the LOS direction 
(subsidence velocity is approximatively equal to vertical 
component of v), and T is the temporal baseline between two 
SAR acquisitions 
The two PS-InSAR techniques are primitively based on the 
Figure 3. Interferometric combinations in short time series interferometric phase model represented by Equation. 1. 
Shuttle Radar Topography Mission (SRTM) DEM of low 
resolution of 90m X 90m is used as the external DEM to remove 
the topography phase. The DEM is enough for our study 
because the test area is a flat area with maximum topographic 
relief less than 30m. 
4. PS TECHNIQUE 
In the processing system of PS technique, the PSs are initially 
identified by means of amplitude dispersion threshold method 
(Ferretti et al., 2000), which is usually an effective detection 
method in a large amount of dataset. Based on the 
interferometric model, spectral estimation strategy is utilized to 
estimate the valid phase terms (e.g. subsidence velocity) on PSs 
in time series. And disturbing phase terms (e.g. atmospheric