In: Wagner W., Szekely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B 
378 
liable estimation of linear deformation. What’s more, by 
integrating the amplitude-based criterion for pixels selection, 
CPT can provide full-resolution deformation. In some sense, we 
can say CPT is a well-integrated technique of the main PSI 
techniques. 
2.5 Other techniques 
Some other multiple-interferogram techniques for deformation 
monitoring emerge and gained many uses as well, including, 
Interferometric Points Targets Analysis (IPTA) developed by 
GAMMA remote sensing research group (Wegmtiller et al., 
2000; Werner et al., 2003) in Switzerland, Spatio-Temporal 
Unwrapping Network (STUN) (Kampes et al., 2005) and phase 
gradient approach to stacking interferograms (Sandwell et al., 
1998; Raucoules et al., 2003; Rocca, 2007). Moreover, there is 
STB AS (Small Temporal BAseline Subset) for monitoring of 
wetland’s water level changes (Hong et al., 2008). 
2.6 Remarks on Advanced D-InSAR techniques 
Compared with Classical D-InSAR, which employs several 
SAR images (4 at most for the 4-pass version D-InSAR) to 
analyze a single deformation episode, Advanced D-InSAR 
technique fully exploits the SAR archives available, and we 
may consider it a postprocessing step (Usai, 2003; Berardino et 
al., 2002) applied to the set of D-InSAR interffograms that may 
be generated via already available interferometric data 
processing tools. Based on this, several considerations are in 
order. 
2.6.1 Foundation of Advanced D-InSAR 
The input of Advanced D-InSAR is a set of Differential 
interferograms. Therefore, a careful D-InSAR processing has to 
be implemented, controlling the quality of all major processing 
steps (e.g. image co-registration, phase unwrapping, etc.), 
guaranteeing a high quality set of input data for the Advanced 
techniques. This is of particular significance for PS method, 
where no compulsive constraints are enforced on temporal and 
spatial baseline and any noisy area existed will introduce mis 
registration problems. Phase unwrapping, on the other hand, 
always remaining the most delicate issue, behaves as a sparse 
and irregularly sampled data unwrapping problem in advanced 
techniques, and can be performed following a two-step 
algorithm (Ghiglia et al., 1998) 
1) estimation of the unwrapped phase differences between 
neighboring pixels; 2) integration of the gradient using one of 
the known techniques, such as minimum cost flow (Costantini, 
1998), weighted least mean squares (Ghiglia et al., 1998; 
Spagnolini, 1995), and branch and cut (Goldstein et al., 1988). 
2.6.2 Data acquisition 
Besides the large stack of SAR images required, uniform 
distribution of temporal and spatial baselines are always 
preferred in order to acquire more accurate and reliable 
information about the ongoing deformation. However, global 
availability of SAR acquisitions is somewhat limited. Many 
areas have few or no acquisitions unless the area of interest was 
previously tasked for imaging. For example, there are subsiding 
areas in Mexico and in the People’s Republic of China that 
have significant aquifer-system compaction problems, however, 
with limited ERS SAR coverage. For Envisat SAR coverage, 
the various selectable polarizations of the transmitted 
electromagnetic SAR signal may limit the availability of SAR- 
image pairs suitable for InSAR processing (Galloway et al., 
2007). 
2.7 Difference and similarity among Advanced techniques 
Besides several differences among the techniques detailed 
above, mainly relying in data requirements (minimum number 
of SAR images, more than thirty needed for PS for a well 
statistics estimation of phase stability), the limitations on 
baseline length (SBAS, LS, CPT), the need of multilooking 
(SBAS, LS, CPT), the multi-pair approach (SBAS, LS, CPT) 
for interferogram formation, there exist several similarities 
among them. 
2.7.1 Deformation extraction strategy 
All the techniques extract deformation through a two-step way, 
linear and nonlinear. In fact, we’d better regard the introduction 
of a linear model as a way to clean phase to make easier 
nonlinear estimation. Such a strategy, dividing and conquering, 
running through the whole signal isolation process, does help a 
lot (Blanco et al., 2007). 
2.7.2 APS estimation and removal 
The output of Advanced D-InSAR includes LOS displacement 
rate, DEM error, and Atmospheric disturbance, with the latter 
two byproducts indicating great superiority compared with 
Classical D-InSAR, and in some way justifying the need of a 
large number of images (Ferretti et al., 2000). 
After the estimation and removal of linear phase (linear 
deformation and DEM error phases), theoretically, three 
contributions still remain: APS, nonlinear deformation and 
noise. In practice, however, the noise contribution was 
mitigated to the minimum either due to the multilooking process 
in SBAS and CPT, or due to the neighboring differencing in PS 
and CPT, thus only APS being the target to be cleaned. Based 
on the observation that the atmospheric signal phase component 
is characterized by a high spatial correlation and exhibits a 
significantly low temporal correlation (random), the desired 
nonlinear deformation is estimated as the result of the cascade 
of a spatial low-pass and a temporal high-pass filtering 
operation, with APS removed( Ferretti et al., 2001; Berardino et 
al., 2002; Mora et al.,2003). 
2.7.3 Multi-plantform interferometry 
The frequency difference between ERS and EnviSAT, although 
a small shift, limits the possibilities of the generation of useful 
cross-interferograms (Monti et al., 2000). For a flat surface, 
theoretically, it’s possible to compensate for the 31 MHz center 
frequency difference unless the normal baseline reaches 2100m 
(Gatelli et al., 1994). Although some researchers found several 
pairs of images for successful crossing-interferometry by 
searching, with delicacy, the whole archives(Santoro et al, 
2007) , we should note that the consequence of such large 
baselines are, on the one hand, the restrictive elevation of 
ambiguity with respect to the image, around 4.5m, which makes 
the interferograms sensitive to topography. On the other hand, 
interferograms with large baseline are extremely sensitive to 
volumetric decorrelation, which poses great limitation in urban 
areas. 
Again, Advance DInSAR techniques circumvent the above 
dilemma elegantly, exploiting pointwise targets, as in the case 
of the Permanent Scatterer approach (Arrigoni et al., 2003; 
Ferretti et al., 2004; Wegmuller et al., 2005) that allows 
investigating the temporal evolution of the detected 
displacements by analyzing full-resolution (single look) 
interferograms, or in the SBAS and CPT cases, by considering 
ERS and EnviSAT as independent subsets, searching for a least