72 
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008 
-* DlnSAR (27 Dec 2006 - 11 Feb 2007) 
Estimated subsidence from ground survey data (27 Dec 2006-11 
Feb 2007) 
Figure 10. Validation of DInSAR-derived subsidence profiles 
against ground truth. 
4.4 Accumulated Subsidence 
The accumulated subsidence has been derived for the period 29 
Jun 2007 to 14 Feb 2008 (230 days) using the 6 ascending 
ALOS PALSAR images (Figure 11). In the most ideal situation, 
the height displacement between 29 Jun 2007 and 14 Feb 2008 
should be directly calculable from the differential interferogram 
29 Jun 2007~14 Feb 2008. However, similar to other 
differential interferograms over long time spans, it is usually 
very difficult to unwrap the interferometric phase correctly due 
to high deformation gradient and decorrelation. Therefore an 
alternative technique was used to compute the subsidence for 
the period 29 Jun 2007~14 Feb 2008 by accumulating the 
subsidence from successive SAR pairs. The simplest way to 
determine the accumulated subsidence is to combine all 
subsidence maps generated from each ALOS pairs. However, 
the error in each DlnSAR result will also be accumulated. 
In order to reduce the geocoding error between each DlnSAR 
result and the noise due to accumulating the deformation, a 
simple approach has been developed, as outlined below. (1) All 
ALOS PALSAR images were first resampled with respect to a 
reference master. (2) DlnSAR analysis was then carried out to 
measure the deformation of the same points in each co- 
registrated images, i.e. interferograms 2-6 in this study (Table 
1). (3) The deformations calculated from the five differential 
interferograms were added together (in slant range) prior to 
geocoding the accumulated deformation map. (4) Geocode the 
accumulated deformation map. Given the assumption that the 
expectation of the atmospheric delay for a point in k 
acquisitions is 0 (Kampes et al., 2006), the atmospheric noise 
would also be reduced by this method. The descending pair is 
not included in the accumulated subsidence map because the 
descending image may measure different scattering objects to 
the ascending image for the same pixel. Hence the deformation 
measured from the descending pair may not be consistent with 
the ascending pair if their deformation values are accumulated. 
Horizontal deformation which causes inconsistencies between 
ascending pair and descending pair is another issue which needs 
to be considered. 
Figure 11. Accumulated subsidence of all subsidence maps for 
the period 29 Jun 2007 ~ 14 Feb 2008 (230 days). 
5. CONCLUDING REMARKS 
This study illustrated the capability of ALOS PALSAR for 
mine subsidence monitoring in Australia. Simulations have 
shown that the new satellites ALOS, TerraSAR-X and 
COSMO-SkyMed perform much better than the satellites 
launched before 2006 for this monitoring application. 
Differential interferograms from ALOS PALSAR and 
ENVISAT ASAR images with similar temporal coverage were 
generated. Strong phase discontinuities and decorrelation have 
been observed in almost all ENVISAT interferograms, whereas 
these issues are almost invisible in ALOS PALSAR 
interferograms due to its better spatial resolution and longer 
wavelength. Six successive subsidence maps derived from 
eight ALOS PALSAR images using both ascending and 
descending passes were obtained. More than 50cm subsidence 
has been found in the Westcliff Mine over 46 days. The 
DlnSAR results derived from ALOS PALSAR data were 
validated with ground survey data at both mine sites. RMSE of 
1.7cm and 0.6cm has been found in the Appin and Westcliff 
mine areas respectively. This study demonstrated an easy to 
implement approach to calculating the accumulated subsidence 
from a series of SAR images by resampling a series of SAR 
images into a reference master. This approach could minimise 
geocoding error between each DlnSAR result and the error due 
to accumulating the results over several DlnSAR results. 
ACKNOWLEDGEMENTS 
This research work has been supported by the Cooperative 
Research Centre for Spatial Information through Project 4.09, 
whose activities are funded by the Australian Commonwealth’s 
Cooperative Research Centres Programme. The Australian 
Research Council has been supporting DlnSAR research at 
UNSW over a number of years. The Australian Coal 
Association Research Program has also supported research for 
ground subsidence monitoring using DlnSAR. The authors wish 
to thank the European Space Agency and the Earth Remote 
Sensing Data Analysis Center (ERSDAC) for providing the 
ENVISAT ASAR and ALOS PALSAR data, respectively. 
METI and JAXA retain ownership of the ALOS PALSAR 
original data. The PALSAR L-l.l products were produced and 
distributed to the IAG Consortium for Mining Subsidence 
Monitoring. The authors also wish to thank BHP-Billiton for 
providing the ground survey data and other spatial data.