nr*» iuxs trjza 
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008 
102 
As far as phase unwrapping algorithm is concerned, the freely 
available phase unwrapping software called SNAPHU 
developed by Chen and Zebker (Chen C W, Zebker, 2001) have 
been chosen, after phase unwrapping, all pixels were calibrated 
with respect to an area that was assumed stable, the zero 
deformation value was taken as the mean of the phase values 
located within a 3*3 neighbourhood in an area with high 
coherence, in our case, the stable area was selected near to 
Purple Mountain which is in the north east of the study area, 
then, for each selected pixel, a joint estimation of DEM errors 
and the model parameter vector p is carried out by solving the 
equation( 10) directly using the least square method. 
and the reference point for each image used for phase 
calibration is marked with R. 
-10 
■-15 
,W 
No. 
Master 
(y-m-d) 
Slave 
(y-m-d) 
^i(m) 
B\\ (m) 
T(d) 
Sub. 
1 
1999-10-18 
1999-07-05 
-98.2 
39.9 
105 
1 
2 
1999-10-18 
1999-04-26 
-77.4 
42.4 
175 
1 
3 
1999-10-18 
1998-12-07 
-48.8 
23.1 
315 
1 
4 
1999-10-18 
1996-08-19 
-86.9 
-3.0 
1158 
1 
5 
1999-07-05 
1999-04-26 
34.9 
2.6 
70 
1 
6 
1999-07-05 
1998-12-07 
86.2 
-16.8 
210 
1 
7 
1999-07-05 
1996-08-19 
21.9 
-43.0 
1050 
1 
8 
1999-04-26 
1998-12-07 
53.2 
-19.3 
140 
1 
9 
1999-04-26 
1996-08-19 
43.7 
-45.6 
980 
1 
10 
1998-12-07 
1996-08-19 
-86.5 
-26.1 
840 
1 
11 
2000-04-10 
1997-12-22 
-66.5 
-118.9 
840 
2 
12 
2000-04-10 
1997-11-17 
-73.2 
-80.3 
875 
2 
13 
1997-12-22 
1997-11-17 
-67.5 
38.2 
35 
2 
Table 1:13 interferograms generated by 8 ERS2 SAR images 
... 
>4% 
: r - -JET- - v I »*•' 
M ■ v. %■ - .» - 1 * 
"0 50 100 150 200 ¿50 
Fig. 2: Generated Delauney triangulations 
In order to provide an overall picture of the detected 
deformation, the false colour map representing the cumulative 
LOS deformation value for each investigated pixel was 
presented in Fig. 3, superimposed on the greyscale 
representation of the SAR image amplitude, areas where the 
measurements accuracy affected by decorrelation phenomena 
(coherence below the threshold) have been excluded from the 
false colour map, to be noted that, the image acquired on august 
19,1996 was assumed as reference image with zero deformation, 
Fig. 3: LOS cumulative deformation maps superimposed on the 
SAR amplitude of the investigate area 
Significant deformation is clearly shown in Fig. 3, it can be 
easily found that there are a large amount of sparsely distributed 
deformation pixels along the Yangtze River and the Qinhuai 
New River, most of the deformation areas are located in the 
west and south of the Qinhuai River, while in the north and east 
of the Qinhuai River, nearly no deformation was found; 
Another large deformation area is in the centre of the 
investigated area, which is on the alluvial floodplain of the 
ancient Qinhuai River, and now is the old city of the Nanjing. 
The largest deformation is 12.0cm, that the mean velocity of the 
subsidence is about 3.3cm/year, this agrees well with geodetic 
measurements carried out by some researchers. 
In order to demonstrate the capability of the proposed approach 
to follow the temporal evolution of the detected deformation, 
some examples are provided in the following Fig. 4, presents 
the chronological sequence of the computed deformations for 
points (marked by A,B,C and D in Fig. 5) located in the area of 
maximum subsidence, except the point C which is near the 
reference point R, it can be seen that the observed deformations 
are characterized by a rather continuous subsidence 
phenomenon from 1996 until 2000, to be noted that, before the 
year 1999, all points of surface subsidence are small, in the 
image sequence of the December 7, 1998, even a amount of the 
increase in performance, subsidence become dramatically since 
the year 1999; it should be demonstrated whether this 
phenomenon has some relations with the Yangtze River water 
level change, particularly in the summer of 1998, the 
devastating floods of the Yangtze River lead to increased 
groundwater level. 
From the above case study of ground deformation in Nanjing 
area, it can be seen that SBAS-DInSAR method have the 
following advantages: (1) Because of using small baseline 
subsets, this method has coherent area significantly larger than 
that of CR and PS method, not only to say traditional pointwise 
measurement such as GPS and levelling; (2) Ground control 
points is not necessary for SBAS-DInSAR method , (3) the use