mining has been theoretically described by Carnec et al. (1994) 
and tested in Selby Coalfield in the UK (Stow, 1996, Stow & 
Wright, 1997). 
    
   
. "nt 4 . t 4 p m a. RN E x A 
Panel PA TAA ta e SS MES X -. 
image of the study area; bright colours 
  
Fig.2: Intensity SA 
are urbanised areas. 
The SAR Interferometry (InSAR) is a technique for extracting 
information from the Earth's surface using the phase of a SAR 
signal. The height of a point on the Earth surface can be 
reconstructed from the phase difference between the signals 
arriving at the antenna during repeated observations of the same 
platform (Pratti et al., 1994, Solaas et al.; 1996). The phase 
difference is directly related to the difference in path lengths 
between the point on the Earth surface and the two positions of 
the antenna. If the positions of the antenna are known 
accurately then the path difference can be used to infer the 
position and the altitude of the target point on the Earth surface, 
using the interference pattern generated by the paths. 
With respect to the Earth, the ERS-1 and ERS-2 satellites go 
through a 35-days cycle of orbits. After this period of time (one 
or more orbit repeat cycles), the same area is visited again. 
Differential processing is used for surface deformation 
detection. It needs data sets acquired with very close viewing 
points (baselines 0 to 100 m). These very short baselines allow 
to neglect topographic influence on the results. On such 
interferograms one complete fringe represents a shift of half a 
wavelength of the backscattered microwave (the radar wave 
must cover the round-trip distance - forth and back — Pratti et 
al., 1994). For the satellites ERS-1 and ERS-2, a fringe marks a 
change of 3 cm in the amount of ground motion (in the 
direction of the satellite). Data with greater baselines can also 
be used, but the topographic influence must be removed using a 
conventional DEM (Digital Elevation Model) or another 
interferogram. 
ERS SAR DATA PROCESSING 
According to satellite data coverage, the following parameters 
have been selected: 
Track: 222, frame 2583 (shifted) — center of the area 
Track: 494, frame 2583 (shifted) — western part of the area 
The criteria used for data selection were as follows: 
a) baseline: less than 300 m (for DEM processing), less than 
100 m (for differential processing) 
b) limitation due to season: the best are from late summer 
(after crops) to early autumn or late autumn (before 
snowfalls). In Poland these periods are characterised by 
dry and stable weather, with little development of 
vegetation. 
c) limitation due to weather: images need to be acquired in 
dry weather condition. The selection was done based on 
meteorological observations from Katowice station (center 
of area), Bierun station (south part of the area) and 
Glubezyce station (west part of the area). Additional 
weather informations such as cloud cover were taken from 
NOAA quicklook images. 
For the interferometric processing the EarthView InSAR 
software versions 1.0.4 and 1.1.0 has been used. Other data 
combinations as well as conventional post-processing were 
applied Atlantis EarthView version 4.4.1 and ERDAS Imagine 
versions 8.2 and 8.3. The GIS analysis of interferograms with 
maps of mining activity and surveying data is currently done 
using ITC ILWIS 2.1. 
DEM GENERATION 
The Digital Elevation Models were needed to removing the 
topographic effect during differential interferometric 
processing. Due to lack of external DEM derived from 
topographic maps, the DEM generated from ERS Tandem Data 
has been used. The quality of tandem data was found to be 
generally well, except in forested areas were coherence was 
very low. To improve the DEM quality in these areas two 
different phase unwrapping algorithms were used: „the 
Interactive Disk Masking” - EarthView (Atlantis 1997) and the 
Constantini method (Constantini 1996). Only the DEM 
obtained by the Constantini algorithm senses the signal in 
forested areas. However, these areas are affected by a higher 
error. The DEM processed by EarthView unwrapping phase 
tool mask the forested areas (puts to 0 m height value- a 
threshold due to low coherence). 
For the topographic effect removal the Constantini algorithm 
was solely used. 
SURFACE CHANGE DETECTION 
The densely urbanised area is presented in Fig. 1, 2. For all 
interferograms estimated coherence is relatively high. 
Generally, on the interferograms are visible areas with 
ellipsoid-shaped fringes representing 2 or 3 cycles of the phase 
changes equal to ca 9 cm subsidence during the 35-days. The 
separate areas affected by changes covers ca 2 to 3 km2. 
For described area two interferograms are presented (Figs. 2, 3): 
1) October 1992 (04.10.92 - 08.11.92). Interferogram 
processed from data with perpendicular baseline — 54 m. 
For topographic effect removal the interferometrically 
computed DEM was used (Fig. 2). 
2) September 1993 (03.09.93 - 08.10.93). The perpendicular 
baseline for this couple of data was 20 m. Due to small 
temporal decorrelation (small season changes), estimated 
coherence was high (Fig. 3). 
556 International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 
  
in