H2-product: The Liège area
(Belgium), vertical displacements
revealed by PSI technique.
Dr Xavier Devleeschouwer
Pierre-Yves Declercq
Geological Survey of Belgium
E-mail: [email protected]
ROI : Liège city
Blue area has been processed by NPA during the second phase of Terrafirma.
The red rectangle corresponds to the target zone surrounding the Liège city (Walloon
Region), which has been selected for ground movement evaluation in old and abandoned
coal mining area.
102 scenes used between 23 April 1992
and 11 September 2005
Satellite data used from ERS1/ERS2 &
ENVISAT
Master Scene Date: 9 August 1998
Georeference (X,Y) accuracy:
± 4m, ± 10 m
Spatial distribution of
the Permanent Scatterers (PS)
STNicolas
• ROI covers 207 km² and contains 28,446 PS
Liège
• Average PS density = 137 PS/km2
• PS data are colour-classified using the
annual average velocity in mm/year
• Ground movements in the studied area are
directly identified.
• The black triangle corresponds to the
reference point used for the PSInSAR
processing by NPA
Histogram of the velocity values
Prior
conditions
before
interpolation:
to
verify
dependency and stationarity.
kriging
spatial
• Geostatistics assumes that the
second
order
stationarity
is
automatically verified in the data.
• The histogram shows the frequency
distribution of the PS velocity values.
The data are normally distributed.
Annual average velocities range between –19.41 and +13.42 mm/year. The distribution
shape is nearly close to a normal distribution with a calculated average of –0.066 mm/year.
This average velocity means that the area is globally in equilibrium. This is not the reality
as data are spatially distributed and a local average is thus more interesting.
The most frequent values are recorded between –4.5 and + 4.5 mm/year and correspond
to 28,406 PS for a total of 28,446 PS.
PS density
& Mean VEL
The spatial distribution of the 28,446 PS is
detailed using a fishnet of 1 km².
• A subsidence phenomenon close to the city
center of Liège with low negative annual
average velocity values (orange) along the
alluvial plain of the river Meuse.
St-Nicolas
• The municipalities of Saint Nicolas and
Beyne-heusay are highlighted by strong
positive velocities (dark and light blue)
LIEGE
BeyneHeusay
Concerning the density
• The PS density is highly heterogeneous with
high densities in the most densely and
populated urbanized center oriented mostly
SW-NE from Seraing to Herstal. PS density
ranges here between 100 and 481 PS/km².
• The southern part of the ROI has few PS
due to the presence of forest areas and
agricultural zones around the villages of
Chaudfontaine, Ougrée and Boncelles.
Spatial distribution of the PS data on a fishnet of one square
kilometer. Black number corresponds to the PS density per square
kilometer. Each square of the grid is colour-classified using the
mean annual average velocity value.
Variography & Kriging
The variogram is being modelised from 2
nested spherical models and a nugget effect
(C0). The slope break of the variogram at
200 m suggests a change in the behavior of
the processes responsible for the data
distribution. The sill is reached at 4,500 m,
providing a distance beyond which the
variogram remains essentially constant. The
nugget effect (C0) correspond to the γ(h)
value when h=0. It happens when PS are
close but their respective VEL is different.
According to the spherical model equation the
parameters are :
Direction: 0.0 Tolerance: 30.0
0.7
0.6
Variogram
0.5
0.4
0.3
C0=0.3
0.2
0.1
0
0
1000
2000
3000
4000
5000
6000
7000
Lag Distance (m)
Omnidirectional experimental variogram (black dots) (nbr
lag : 140, lag dist : 50 m) and modeling (continuous blue
line) based on the annual average velocities of the PS.
a1=200
C1= 0.1
a2=4500
C2=0.37
The nugget effect is tantamount to 39% of
the total variability. In other words, 39% of
the variation is random and unpredictable.
This variation has to be linked with
measurements errors and the natural
variability of the measured objects.
Kriging Interpolation
Two main uplifting areas are observed
opposed to the centre of Liège gently
subsiding along the axis of the river Meuse.
SaintNicolas
BeyneHeusay
The Saint-Nicolas uplift zone, with a
crescent form, extends from Seraing (left
side) up to Herstal and Cheratte (upper
right corner)
The Beyne-Heusay uplift is located on the
right bank of the river Meuse and extends
on
the
Chênee
and
Beyne-Heusay
localities.
The Saint-Nicolas area, on the left bank of
the river Meuse shows the strongest
positive annual average velocity values
ranging between 0.5 and 2.5 mm/year.
The mean annual average velocities of
Beyne-Heusay
are
lower
(0.5-1.25
mm/year) than those in the Saint-Nicolas
area.
Geology
The river Meuse separates more or less the
northern region composed of soft Mesozoic
sedimentary
series
covering
the
Paleozoic
formations from the southern area characterized
mostly by hard Paleozoic sedimentary series. The
coal basin of Liège is located in the Namur-Verviers
Synclinoria. The Carboniferous coal series are
composed mainly by a cyclic succession of shales
and sandstones with more than 30 exploited coal
seams.
Cretaceous formations are present on the
“Hesbaye” (NW) and “Herve” (East) plateaus and
are unconformably lying on the Paleozoic folded
and faulted subcrop. Locally, Oligocene sands and
Quaternary loess cover all these oldest deposits.
The Quaternary of the alluvial plain consists in
several heterogeneous stratified and lenticular
deposits. Gravels and sands are found as much as
clay, loams and peats or sediments with a high
organic material contents.
Geological
map
(1:40,000
scale-map)
superposed
on
topographic 1:100,000 scale-map from NGI (© National
Geographic Institute)
Bt: Burnotian | Cb3 : Coblencian | Co : Couvinian | Gva-b :
Givetian | Fr1 Frasnian | Fa1 : Famennian | T1: Tournaisian |
H1b,H2 : Houiller | Cp1, Cp2, Cp3, Cp4 : Campanian | M :
Maestrichtian | Om,On : Oligocene | al : modern alluvial
sediments, Quaternary
Uplift and the extent of coal mining activities
The extent of the coal mining activities figured
by the black polygons is superimposed to the
kriging interpolation based on the annual
average velocity of the PS.
Uplifting zones, in blue, fit relatively well the
areas where coal seams have been deeply
exploited.
The Saint-Nicolas and Beyne-Heusay uplift areas
are clearly separated by a wide red zone
characterized by negative annual average
velocities
(subsidence).
The
subsidence
phenomenon is clearly related to the extent of
the alluvial plain of the river Meuse. The
subsidence disappearing westward indicates that
the velocities of the uplift are higher than those
of the subsidence.
Extent of the exploited coal seams (after Calembert, 1955)
superimposed
to the kriging interpolation and background
topographic 1:100,000 scale-map from NGI (© National Geographic
Institute).
Coal concessions limits
in the Liège coal basin
The extension of the positive ground
movements (blue colours) are centered on
the coal concessions (black polygons).
• Saint-Nicolas-Seraing uplift covers directly 5
main coal concessions named, from north to
south, “Bonne-fin - Banneux - Batterie”,
“Patience et Beaujonc”, “Espérance et BonneFortune”, “Gosson-Kessales” and “Cockerill”.
• Beyne-Heusay uplift, on the right bank of
the river Meuse, is almost totally observed in
the Werister coal concession.
• Positive ground motions correspond to a
zone characterized by a crescent form
starting in the southwestern part (Seraing,
right bank of the river Meuse), crossing the
river Meuse towards the NNE through the
districts of Saint-Nicolas and Herstal, then
crossing the river Meuse towards the Cheratte
district of Visé.
Limits of the coal concession superimposed to the kriging
interpolation. Background topographic 1:100,000 scale-map
from National Geographic Institute.
Kriging interpolation
and mines closure dates
The three next slides highlight the evolution
of the coal mining activity (year of closure)
in the coal basin of Liège in 1965, 1969 and
1972.
The coal concessions are superimposed with
the kriging interpolation based on annual
average velocity.
In 1965, almost all the coal concessions
surrounding the city centre of Liège are
active.
1965
Limits of the coal concession in the coal basin of Liège
superposed to the kriging interpolation. Hatched surfaces
indicate active coal concessions in 1965. Background
topographic 1:100,000 scale-map from National Geographic
Institute.
Kriging interpolation
and mines closure dates
In 1969, the coal concessions of “Ans” and
“Abhooz et Bonne-Foi Hareng” located in the
northern part of the studied area have been
closed.
The coal concession of “Werister” located in
the southeastern part of the studied area has
been also closed.
1969
Limits of the coal concession in the coal basin of Liège
superposed to the kriging interpolation. Hatched surfaces
indicate active coal concessions in 1965. Background
topographic 1:100,000 scale-map from National Geographic
Institute.
Kriging interpolation
and mines closure dates
In 1972, only few concessions are still
active.
Three coal concessions (“Patience et
Beaujonc”, “Cockerill” and “Espérance et
Bonne Fortune”) are centered on the SaintNicolas - Seraing districts.
1972
Limits of the coal concession in the coal basin of Liège
superposed to the kriging interpolation. Hatched surfaces
indicate active coal concessions in 1965. Background
topographic 1:100,000 scale-map from National Geographic
Institute.
Number of exploited coal seams
Mining data in the coal concessions centered on
the uplift area of St-Nicolas-Seraing indicate a
deepness between 750 and 1000 meters.
These data reveal that the last collieries, still
active during the seventies in the Liège coal
basin, were thus not only the deepest
exploitation activities but also the ones with the
highest number of exploited coal seams.
Since the closure of the collieries, the
exploitation and groundwater pumping activities
were
abandoned.
The
rise
of
mining
groundwater after several years of aquifer
recharge leads to hydrostatic overpressure
resulting in several centimeters of elastic
rebound (uplift) in these previously subsiding
mining areas.
The number of exploited coal seams in the coal basin of Liège (courtesy of
Veschkens, ISSeP) is represented here in orange (between 11 and 16) and
red (between 17 and 25) colours superposed to the kriging interpolation.
Background topographic 1:100,000 scale-map from National Geographic
Institute.
PSInSAR data versus
Karst hazard database
Karst atlas consists provides the description of
each karstic feature situated in the Walloon
area.
Most of the carbonate rocks of Devonian and
Carboniferous age are subject to karstic
phenomena
The superposition of the karstic features on the
kriging interpolation reveals a subsidence bowl
closely located with numerous karstic sinkholes
(green triangles) and stream sinks in the south
of the studied area.
Karstic features (courtesy of DGATLP, Walloon Region)
superposed to the kriging interpolation. Background topographic
1:100,000 scale-map from National Geographic Institute.
Karst hazard level
PS observations were undertaken on the field
close to the black karstic zones indicating high
vulnerability level.
A few PS present in these areas are clearly
identified on residential buildings, have a good
coherence factor and some of them indicate
very low VEL value.
The superposition of karst vulnerability zones
with PS data is efficient in high PS density areas
or need to monitor ground movements with
artificial corner reflectors.
PSInSAR data versus
GPS Station
The GPS station of the Sart-Tilman provides weekly
information concerning its own positioning for a time
period starting in July 2004
One PS present at a close distance (< 20m) has a good
coherence of 0.86 and a VEL value of –1.72 mm/year.
The annual average velocity calculated from the GPS time
serie corresponds to –2.45 mm/year.
The difference between the two techniques (< 1mm/year)
is acceptable.
The mechanism responsible
displacement is unclear.
for
the
vertical
ground
PSInSAR data versus
levelling (downstream area)
The AIDE (Association Intercommunale pour le
Démergement et l'Epuration) has made several
levelling campaigns during the last 50 years to
determine the evolution of the ground movements
along the river Meuse. Levelling campaigns have
thus begun during the fifties and were used to
monitor and cartography the mining subsidence.
After several years of equilibrium, an uplift
phenomenon was monitored for the first time
during the years 1977-1978.
• The levelling reference point, not figured here,
for this levelling loop is located 4-5 km to the
northeast in a very stable area confirmed by the
annual average velocity of 0.11 mm/year of the PS
in a 100 m cluster around the reference point.
• A buffer with a 100 m diameter centred on each
levelling points is associated with a cluster of PS.
All the PS inside this cluster are used to calculate
the mean annual average velocity (mm/year) of
the cluster.
Available levelling points (courtesy of AIDE) downstream area of
the Liège centre along the river Meuse superposed to the kriging
interpolation. Background topographic 1:100,000 scale-map from
National Geographic Institute.
Comparison of the mean average annual velocity in mm/year for the PS clusters (red bars) with the
levelling vertical displacement rate of the benchmarks (blue bars).
• Vertical average annual velocity of the clusters ranges between -1.09 and + 0.25 mm/year.
• PS density in the clusters varies between 1 and 26.
• Average coherence of each cluster of PS is always higher than 0.73.
• Calculated velocity values for the levelling campaigns correspond to ten years (between 1991 and
2001) and range between - 1.27 and +1.55 mm/year. Some points show discrepancies between positive
velocity value calculated for levelling and negative value determined by the PSInSAR technique.
• A difference < 1 mm/year is observed for more than 80% of the points where both velocity values are
available  correlation is clearly observed between PS data and these terrestrial geodetic measurements
Conclusions (1/2)
Kriging interpolation highlights two main ground movements. Positive annual average velocity values are
observed on the hills that surround the urban centre of Liège on both sides of the river Meuse. On
the opposite a sinuous lineament characterized by negative annual average velocity values fits the
alluvial plain of the river Meuse.
The evolution of the coal mining activities in the coal basin of Liège reveals that the last collieries, still
active during the seventies in the Liège coal basin, were thus not only the deepest exploitation (7501000 m) activities but also the ones with the highest number of exploited coal seams.
Kriging interpolation was also compared to the presence of karstic features in the southern part of the
studied area and to available geotechnical maps. There are some good correlations between some
karstic features and PS data but it’s seems difficult at that time without detailed fieldwork to
establish clearly a link between the presence of karstic processes and ground movements observed
on buildings in the area. An increase of the PS density is necessary to go further in details. The
comparison with geotechnical maps doesn’t reveal any general trend regarding any kind of data
represented. Only limited and localized places related to the presence of an old sand quarry, a
mineshaft, backfill materials or in the alluvial plains reveal subsidence phenomenon.
The northern levelling site (downstream of the Liège centre, close to Herstal) shows that a difference
between the PSI technique and the leveling is lower than 1 mm/year.
The GPS station of the Sart Tilman provides weekly information for a time period between July 2004 and
December 2005. The annual average velocity given by the closest PS corresponds to -1.72 mm/Yr.
The annual average velocity of –2.45 mm/Yr, calculated from the time series, corresponds to the
vertical movement of the GPS antenna. The difference is lower than 1 mm/Yr and thus considered as
reliable. Unfortunately, we are not able to explain the subsidence value observed here
Conclusions (2/2)
Coal exploitation and groundwater pumping activities were abandoned thirty years ago. Even after a long
period of time, an uplift is still observed. An elastic rebound, related to the recharge of the mine
aquifer in a previously subsiding basin during mining activities, is still observed in some parts of the
Liège coal basin.