2012
equation,
tter InSAR.
| persistent
CATTER
ork
1al network
structure of
jection and
solution of
ution. That
ed in radar
ndscape is
elationship
ace will be
se the sites
conversion
dicates the
e to image
' persistent
shows the
atter B and
t the link is
AR image
radar slant
lection of
es, because
D between
puted with
) increases
| the local
ts extreme,
shortening
ighbouring
| is usually
pairs with
persistent
' network
> NDPs of
sidues that
tent scatter
V
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B7, 2012
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia
3.2 Construction of three-dimensional persistent scatter
Delaunay network
To make the relationship of persistent scatter pairs consistent
with geography space, the three-dimensional Delaunay network
algorithm (Zhou et al. 2007) is introduced to construct
persistent scatter network. After persistent scatters are
identified from time serial radar images, the geodetic
coordinates of persistent scatters are derived from the
DEM interferometry with two tandem images and the geodetic
height H is interpolated from external DEM such as SRTM
DEM. Then the geodetic coordinates (LBH) are transformed to
Cartesian coordinates (XYZ). Finally, the three-dimensional
persistent scatter Delaunay network based on Cartesian
coordinate system is constructed with three-dimensional
Delaunay network algorithm. This network is further reformed
by cutting the arc longer than 1km. The optimised network is
then used to establish persistent scatter pairs.
To show the difference between three-dimensional Delaunay
network and two-dimensional network, the comparison of the
two kinds of network is conducted by simulating in Matlab.
The simulated images are demonstrated in figure 3. Geographic
wol uu
e
^i uid
19
80^
Y
(a) The surface generated (b) Planar Delaunay network
by Peaks function in Matlab
terrain is firstly simulated with Matlab peaks function. The
simulated geographic terrain is showed as surface image in
figure (a). Then 51 points are randomly selected as persistent
scatters. The two-dimensional network (figure (b)) and
three-dimensional network (figure (c)) of 51 persistent scatters
are constructed respectively according to planar coordinates
(XY) and Space rectangular coordinates (XYZ). From the view
of shape and structure, two-dimensional network are
significantly different from three-dimensional network. The
number of triangles is 89 in two-dimensional network but it's
9] in three-dimensional network. That means persistent scatter
pairs formed by two-dimensional network are less than that
formed by three-dimensional network. Furthermore, the
three-dimensional network established with persistent scatter
geographic coordinates is fixed as long as the geographic
coordinates of persistent scatters are determined.
Correspondingly, the persistent scatter neighbourhood derived
from the three-dimensional network are definite. On the
contrary, the neighbourhood in two-dimensional network
established in image coordinate system are varied with the
image resolution and projection while the geographic scene is
converted to image.
63782.
N 6.3782
6.3782. =<
2 Lem
6.3782 xc
6.3782 —
6.3782 S
6.3782
(c) Three-dimensional Delaunay network with
three-dimensional coordinates
Figure 3. The Delaunay networks generated respectively by planet coordinates and three-dimensional coordinates
4. EXPERIMENTAL DATASET AND RESULTS
4.1 Experimental dataest
In the last century, the urban area of Shanghai was found
beginning to subside because of the excessive exploitation
of underground water (Zhang, 2002, Zhang, 2005, Ye et al.
2005). InSAR based on three-dimensional persistent
scatter network and two-dimensional network respectively
is used to detect the ground subsidence of Lujiazui in
Shanghai. Figure 4 displays the experimental area of
interest (AOI) marked by a box onto the master amplitude
image, where the inset shows the enlarged multi-image
reflectivity map derived by averaging all the image
patches of the AOI. The AOI covers the rectangle
geographic scope ranging from 121.44584°E to
121.58915°E and 31.20618°N to 31.288°N. The total area
is about 33km? 26 single look complex (SLC) SAR
images taken by ERS-1/2 during 1992 through 2002 are
utilized. The SAR image taken by ERS-2 on Jun 4, 1996
was chosen as the common master image and the
remaining 25 images were used as the slave images. Thus
59
Fig. 4. The experiment area marked by
a box onto the master amplitude image
25 differential interferograms were generated by the
“two-pass” method (Gabriel et al. 1989, Massonnet et al.
1993, Zebker et al. 1994b). Table 1 lists the parameters of
all the images, including spatial and temporal baseline
with respect to the master image.