, less than
e summer
n (before
terised by
pment of
cquired in
based on
ion (center
area) and
Additional
taken from
w InSAR
Other data
sing were
S Imagine
yrams with
ently done
noving the
rferometric
ived from
ndem Data
ound to be
erence was
areas two
used: „the
97) and the
the DEM
> signal in
yy a higher
ping phase
it value- a
i algorithm
, 2. For all
vely high.
areas with
of the phase
5-days. The
m2.
(Figs. 2, 3):
terferogram
ine = 54 m.
rometrically
erpendicular
ue to small
), estimated
Processed interferograms (Figs. 2, 3) show the presence of
ellipsoid-shaped differential fringes which can be interpreted as
the effect of mining subsidence. The phenomena are located in
two areas: in the north — the vicinity of Bytom city and in the
south — between Katowice and Ruda Slaska. Spatial distribution
of the detected surface changes well correspond to the active
mining zones at that time. The detailed analysis shows that the
locations and shape of the fringes can be directly related to the
surface changes above the mining work faces.
In the Bytom area in many locations 2 or 4 fringes are visible
(the subsidence of ca 6 to 12 cm during 35-days). Such amount
of changes can be explained by reactivation of old abandoned
works in the cap of mining panel. This area was in the past a
centre of zinc and lead ores mining which caused weakening
and fracturing of the overburden.
In the area between Katowice and Ruda Slaska
to 6 cm per 35-days) have been noted.
ing surface changes
during October 1992 (04 Oct — 08 Nov). White lines are coal
mine lease zones
Fig. 2: Differential interferogram show
e
At each step of InSAR differential processing problems may
occur. To overcome them one needs quite some experience and
pay attention, consider some critical items, as listed hereafter:
Atmospheric effects: local variations in atmospheric properties
lead to differences in the path lengths between the two antenna
positions and the target area and giving rise to spurious phase
variations which are superimposed to the phase variations
resulting from the target area (Tayre & Massonnet 1994, 1996).
Only the comparison of a series of interferograms from the
same area can prove that no influence of atmospheric artifacts is
present. In our case the phase changes caused by atmospheric
effects are generally of greater extent than subsidence effect and
can be easily recognized. The processing of 3 interferograms
from subsequent 35-days can help to exclude all type of
artifacts as they would occur only in one of the three
interferograms. For final conclusions a comparison with mining
data and with meteorological data is still needed.
Temporal decorrelation: excessive large time intervals
between acquisitions of SAR scenes can result in a reduction of
coherence preventing the generation of interferograms due to a
temporal variation in backscattering properties of the target area
(FAO/ESA, 1993, Pratti et al., 1994) . The temporal separation
between the data in this project was 35 and 70 days. Some of
the couples were also acquired during different season. All 35-
day interferograms are characterized by good coherence in
urbanised areas. The coherence in arable areas is variable and
depends on the season. In all cases of 35-day interval the
coherence in agricultural regions was acceptable for
interferogram processing. However the 70 days interferograms
have lower coherence. A clear seasonal dependence have been
also observed.
c£. SED
>: Sb
5 P AR t E NS VE a E 2 t S
Pre ee eI
Fig. 3: Differential interferogram showing surface changes
during September 1993 (03 Sept — 08 Oct). White lines are coal
mine lease zones
DEM co-registration: couples of data with baselines higher
than 50 m usually need topographic effect removal. This
procedure is based on the co-registration of a differential
interferogram with a DEM and a subsequent extraction of the
topographic effects (the fringes due to topography). The result
may depend on the precision of co-registration with external
DEM. An error in DEM co-registration can make differential
detection impossible. In our case data with perpendicular
baseline less than 100 m were used; it means that one full fringe
is generated by topographic height difference of ca. 81 m. It is
important to consider the relief differences in area investigated.
In our case they are less than 150 m, therefore the topographic
effects caused by the relative long baseline are within one
fringe. Thus, such effects can be easly removed, even by means
of a coarse DEM.
For verifications different 35-days interferograms were
processed with the same reference image, but with different
baseline values: they show no influence on differential fringes
proving the consistency of the method.
Co-registration with mining data: the coal mining panels are
usually relatively small. The mining front of the panel is
usually ca 200 long (ca 10 pixels on the interferoram), so that
the error in co-registration larger than 4 to 5 pixels is
unacceptable. The mining maps are usually in specific co-
ordinate system of no standard projections and spheroid. The
comparisons of such different data require large number of
International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998 557
