107 
DEFORMATION MONITORING BY MEANS OF SAR INTERFEROMETRY 
IN THE NORTH-BOHEMIAN AREA 
B. Knechtlova, I. Hlavacova 
Department of Mapping and Cartography, Faculty of Civil Engineering, Czech Technical University in Prague 
Thakurova 7, 166 29 Praha 6, Czech Republic 
barbora.knechtlova@fsv.cvut.cz, hlavacova@insar.cz 
http://www.insar.cz/ 
Commission VII, WG VII/2 
KEY WORDS: SAR Interferometry, Land Subsidence, Change Detection, Multitemporal Observations 
ABSTRACT: 
Radar interferometry allows for Earth-crust deformation mapping with the use of satellite images, without the necessity of expensive 
on-site measurements. Its accuracy may even reach several mm/yr in the theoretical case. However, it works only on objects which 
are stable in a long-time horizon - typically, these are artificial objects, such as roads, buildings etc. We use the radar interferometry 
to compute the subsidence on various places at the North-Bohemian coal basin where several communications were built on waste 
dumps. The method is based upon processing of stack of several scenes into interfograms and subsequent iterative adjustment of 
selected regions, which have coherence higher than chosen threshold. 
1. INTRODUCTION 
Radar interferometry is a method providing a possibility to map 
ground deformations in an area imaged by a satellite carrying 
synthetic aperture radar (SAR), without the necessity of 
expensive on-site measurements. The accuracy of this method 
may even reach several mm/yr in the theoretical case. However, 
this method is only usable in areas with no vegetation and for 
object, which are not expected to change during monitoring. 
Generally considered, it can be said that InSAR is suitable for 
artificial objects, such as building, roads, railways, etc. In 
addition, it is more important to measure the deformations for 
these artificial objects, than for forests or agricultural fields. 
We use 4 stacks of 12 or 13 scenes of the North-Bohemian area 
to assess the deformations in time in this area, particularly at 
some (coherent) parts of the scene: Most and Teplice sites. The 
North-Bohemian brown coal basin is a largely unstable area. In 
addition to many huge open mines, it contains also deep mines; 
some of them are very old and abandoned and may possess a 
potential danger for people living in those areas. More data and 
more areas will be processed in the future. 
2. THEORY 
Synthetic aperture radar (SAR) interferometry (InSAR) 
processes a pair of satellite SAR images, acquired by a satellite 
carrying SAR, which is ERS-1/2, ENVISAT, RADARSAT, 
JERS-1/2 and others. These scenes are complex-conjugate- 
multiplied, giving the multiple of their magnitudes (not 
important) and the difference of their phases, which is related to 
the difference of the distances between the satellite and the 
reflector in the two scenes. The phase map is called an 
interferogram. However, before actual postprocessing, the 
phase given by the flat-Earth and phase given by the DEM must 
be subtracted from the interferogram, and then the 
interferogram is considered to contain only the atmospheric 
signal, deformation signal and noise. 
The result of SAR interferometry may be a digital elevation 
model (DEM) or a map of Earth-crust deformations in the 
processed area. Due to the fact that the processing is time and 
memory requiring, we do not process the whole scene, only 
specific areas (usually towns and cities or other artificial 
objects). In some other interferograms, these areas are not 
coherent - these interferograms were excluded from the 
postprocessing. In addition, the urban areas are the most 
important to be monitored for subsidence. 
The stack method involves processing of several interferograms 
with a common master, with respect to which the deformations 
are related. In order to achieve a larger number of 
interferograms, some of the slave scenes are resampled in order 
to correspond exactly (i.e. with subpixel accuracy) to the master. 
Then, any of these resampled scenes can be used as a master for 
creating other interferograms, although the baseline parameters 
etc. do not change by resampling. The larger number of 
interferograms is not only usable in the case when some 
interferograms must be excluded due to bad coherence (bad 
coherence may be also crop-depending, i.e. different for 
different crops of the scene), but 
also for reducing erroneous influences in the adjustment and for 
ambiguity resolution, which is an important part of the 
postprocessing. 
Atmospheric signal is kind of strong, however, it is assumed to 
change only slowly in the space, and if not processing large 
area (which is our case), only a constant can be subtracted from 
the whole interferogram - we call this process phase reference 
to a certain point. The deformation signal is what we are 
looking for. 
The most significant limitation of InSAR is the ambiguity of the 
phase - the measured phase is always within the (-k, +7t)