IEEE Geoscience and Remote Sensing Magazine - March 2020 - 29

the noise points later during processing. Second, only the
PSCs identified previously are used to generate interferograms with a single-reference mode, and the flat Earth and
topographic phases are removed. Third, the PSCs are connected to a reference point to form a network. According to
the established second differential phase model

z i = W 2r {

N
SAR Images

Baseline
Estimation

4r
4r
dhB 9i +
dvTi + { res
i }, (5)
mRsini
m

PS
Identification

Interferogram
Generation

Reference
Point

Remove Flat
and Terrain
Phase

and the defined coherence coefficient c

1
c= N

res
-1 { i

, (6)

n=1

the linear deformation rate dv and elevation correction dh
are solved by max {c} using the constraint ; { res
i ; 1 r [4],
where W 2r {$} denotes a wrapping operation, and the residual phase comprises the nonlinear deformation phase comaps
nldef
ponent, APS, and random noise: { res
+ { i + { noise
.
i = {i
i
The max {c} is a nonlinear optimization problem due to
the unwrapped phase used in (5), and a simple periodogram algorithm is utilized to estimate the parameters. If
further refinement is needed, iteration can be considered.
The atmospheric phase is separated from the residual
phase using a filtering method and is compensated for using the phase generated in the second step. We can iterate
several times until the results are satisfactory, and noise
points can be eliminated in the iteration. Figure 10 shows
the PSI flowchart. PSI has the ability to suppress spatial
and temporal decorrelation significantly, even for interferometric pairs with large spatial and temporal baselines.
The introduction of PSI opened the door to study of the
time-series InSAR technique; however, PSs are often lacking in nonurban and mountainous areas, and low spatial
sampling density often leads to unreliable results.
SMALL BASELINE SET
The SBAS method utilizes DSs to retrieve deformation,
which solves the problem of PSI not utilizing PSs in nonurban areas. The key calculation steps include the following:
◗◗ For the first step, select the SAR image pairs whose perpendicular and temporal baselines are less than a certain threshold to generate interferograms as a way to
reduce spatial and temporal decorrelation.
◗◗ For the second step, complex multilook processing is
performed on the interferograms to enhance the scattering characteristics of DSs.
◗◗ For the third step, phase unwrapping is performed for
each interferogram using the MCF method.
◗◗ For the fourth step, the phase model is established, and
the SVD algorithm is used to estimate the deformation
and elevation correction.
During subsequent processing, nonlinear deformation
and APS can also be obtained by using a filtering method.
Figure 11 displays the SBAS flowchart. SBAS can enhance
MARCH 2020

Establish PS
Network

IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE

Parameter
Estimation

Integration

Filtering

Height
Correction

Deformation

APS

FIGURE 10. The PSI flowchart.

N SAR
Images
Baseline
Estimation

Interferogram
Generation
Complex
Multilook
Phase
Unwrapping

SVD

Height
Correction

Deformation

Model
Deformation

Residual
Phase

Filtering

High-Pass
Deformation

APS

FIGURE 11. The SBAS flowchart.

IEEE Geoscience and Remote Sensing Magazine - March 2020

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - March 2020