69
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008
In the test sites, a subsidence bowl with radius of 150m is
expected given that the width of each longwall panel is about
200-300m. Therefore, theoretically the expected maximum
deformation that can be detected (without phase discontinuity)
is approximately 8cm, 7cm, 48cm, 86cm, 39cm and 39cm for
the wavelengths of the ERS, ENVISAT, JERS, ALOS,
TerraSAR-X and COSMO SkyMed satellites respectively
(assuming resolution of 25m, 30m 18m, 10m, 3m and 3m
respectively), along the LOS direction.
A simulation is carried out to investigate this effect using a
subsidence model (Figure 2) derived from an ALOS PALSAR
DInSAR result. The model has a peak subsidence of 50cm. The
subsidence model is rescaled based on the ground resolution of
each satellite and is converted into absolute phase using
equations (2) & (3). Differential interferograms are simulated
by wrapping the absolute phase (Figures 2). The simulated
differential interferograms are then converted back into LOS
displacement by unwrapping the phase in the simulated
differential interferogram using the MCF method (Costantini,
1998). The temporal and spatial decorrelation is not considered
in this simulation. Phase saturation has been observed in both
differential interferograms derived from ERS and ENVISAT
data due to the high phase gradient in the subsidence model. In
contrast, the phase fringes in the differential interferograms
from ALOS, JERS, TerraSAR-X hnd COMOS SkyMed data are
reasonably clear.
Figure 2. Simulated differential interferograms from various
SAR satellites based on the subsidence model under noise-free
conditions.
Figure 3. Detectable subsidence errors with different
magnitudes of peak subsidence under noise-free conditions.
The simulation is repeated using a subsidence model with
different peak subsidence (from 5cm to 150cm). The detectable
subsidence errors (RMSE) with different peak subsidence are
shown in Figures 3, which shows that the ALOS, TerraSAR-X
and COSMO-SkyMed data are able to be used to measure
larger displacement with much lower errors. The L-band ALOS
PALSAR is able to maintain a low subsidence error with
relatively high maximum detectable subsidence. High RMSE is
observed in the ENVISAT and ERS results, with peak
subsidence greater than 10cm.
4. RESULTS
ALOS and ENVISAT images with similar temporal coverage
were searched for the test sites. The two-pass DInSAR
technique with a 25m resolution DEM was used to estimate the
location and amplitude of ground deformation. The
performance of earlier SAR satellites ERS-1/2 and JERS-1 have
already been discussed in a previous study (Ge et al., 2007).
4.1 ENVISAT ASAR
More than 90 ENVISAT images have been acquired over the
same site during the period 07 July 2006 and 10 March 2008.
The images were acquired from 7 different tracks, in both
descending and ascending passes, with four different imaging
modes. Although the location of the subsidence bowls can be
identified from many ENVISAT differential interferograms,
strong phase discontinuities and decorrelation have been
observed in almost all ENVISAT interferograms, and hence it is
not possible to generate displacement maps. Figure 4 shows an
example of a differential interferogram generated using
ENVISAT pairs for both mine sites. The interferogram derived
from ENVISAT pairs show phase saturation near the centre of
the subsidence bowl in the case of the Westcliff Mine, while the
fringes at the rims of that subsidence bowl are reasonably clear.
The phase of the interferogram in Figure 4 is unwrapped and is
converted into vertical displacement. The maximum subsidence
detected by the ENVISAT pair detected from the
interferograms is about 5cm, whereas the expected subsidence
is greater than 40cm. This is because the phases in the
ENVISAT differential interferograms fail to correctly
unwrapped due to phase saturation. The ENVISAT differential
interferogram again shows phase saturation in the centre of the
subsidence bowl in the case of the Appin Mine. Unlike the
subsidence bowl in the Westcliff Mine, the fringes at the rim of
the subsidence bowl in Appin are only clear in the upper parts
of the image (low vegetation area) and are very noisy for the
lower parts (heavily vegetated area). This suggested that
ENVISAT images can be affected strongly by vegetation.
4.2 ALOS PALSAR
There are 10 ALOS PALSAR acquisitions available for the
period from December 2006 to March 2008, from both
ascending and descending passes, with two different imaging
modes (FBS and FBD). Seven differential interferograms were
generated based on the ALOS PALSAR images (Table 1). The
ALOS PALSAR FBD data are oversampled by a factor of 2 in
the range direction so that they can be co-registered with ALOS
PALSAR FBS data for DInSAR processing. Figure 5 shows the
differential interferograms generated by ALOS PALSAR pairs
for a similar time period to the ENVISAT pairs (Figure 4). The
fringes in the differential interferogram derived from the ALOS
pair are very clear even at the centre of both subsidence bowls.