PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
January 2014
9
py top, bulk volume of vegetation, and
ground surface, one can derive height
differences between physical scat-
terers with differing scattering char-
acteristics (Lee and Pottier, 2009).
Physical radar backscattering models
for different vegetation types can be
developed to calculate the canopy
height, bare-Earth topography, and
other parameters based on measure-
ments from Pol-InSAR images.
Technical
Frontiers
PSInSAR – Monitoring Time-
series Deformation with mm-scale
Precision
When more than two SAR images of
a study area are available, multiple
InSAR images can be produced and
multi-interferogram InSAR processing
can be employed to improve the
accuracy of deformation measurements
(or other InSAR products) (Ferretti
et al
., 2001, 20011; Berardino
et al
.,
2002; Hooper
et al
., 2007; Zhang
et al
.,
2012). The goal of multi-interferogram
InSAR processing is to characterize the
spatial and temporal behaviors of the
deformation signal, various artifacts,
and noise sources (e.g., atmospheric
anomalies including radar-frequency
dependent ionospheric phase advance
and non-dispersive tropospheric phase
delay of the radar signals, orbit errors,
DEM-induced artifacts) in individual
interferograms, then to remove the
artifacts and anomalies to retrieve
time-series deformation measurements
at the SAR pixel level.
Among several approaches to
multi-interferogram analysis, per-
sistent scatterer InSAR (PSInSAR) is
one of the most promising for deriving
time-series deformation images.
PSInSAR uses the distinctive back-
scattering characteristics of certain
ground targets (PS) and unique
characteristics of atmospheric delay
anomalies to improve the accuracy
of conventional InSAR deformation
measurements (Ferretti
et al
., 2001).
The SAR backscattering signal of a
PS target has a broadband spectrum
in the frequency domain. As a con-
sequence, the radar phase of a PS
target correlates over much longer
time intervals and over much longer
baselines than non-PS targets. If the
backscatter signal from a given pixel
is dominated by return from one or
more PS scatterers, the pixel remains
coherent and the limitation imposed
by loss of coherence in conventional
InSAR analysis can be overcome. The
atmospheric contribution to the back-
scattered signal, DEM error, and orbit
error can be identified and removed
from the data using a multi-interfero-
gram iterative approach, and displace-
ment histories at PS pixels can be re-
solved with millimeter-scale accuracy.
If a sufficient number of PS pixels exist
in a series of interferograms, relative
displacements among them can provide
a reasonably detailed approximation
to the surface deformation field. The
result is akin to conventional geodetic
measurements made at a network of
benchmarks, as opposed to the more
complete spatial coverage provided by a
coherent InSAR deformation image.
Multi-track PSInSAR – Fusion
of Descending and Ascending
PSInSAR Tracks
Conventional PSInSAR can derive
time-series surface displacements only
in the satellite’s line-of-sight direction.
However, if multiple SAR images of
the same area from both descending
and ascending tracks are available,
PSInSAR analysis can be used to es-
timate the 3-D positioning of PSs over
time and to derive their displacement
histories in both the east-west and ver-
tical directions (by ignoring any dis-
An example of multi-temporal InSAR pro-
cessing over Hong Kong Disneyland and its
surroundings where most lands were re-
claimed from the sea. TerraSAR-X interfer-
ograms (2008-2009) with different temporal
intervals ranging from 11 to 77 days serve
as the basic input to multi-temporal InSAR
processing.
L-band radar image of Yunaska Island of the
Aleutian arc acquired by NASA’s UAVSAR
(uavsar.jpl.nasa.gov).
