The surface movement in an area of about 22 km
Most research of the movements of the Earth's surface above underground mines has focused on the direct impact of mining, i.e., the impacts that occur during the lifetime of the mine. This is entirely logical because the largest amount of movement occurs during that period. Additionally, during that period, the mining company can limit the hazards, e.g., by selecting a different mining method (e.g., room and pillar instead of longwall), by backfilling the mined-out area instead of creating a goaf, or by changing the mining geometry. However, by introducing the concepts of sustainable mining, the long-term impact of mining on its surroundings has been receiving greater attention. This means that the period after the mine's closure is a period that should not be neglected. Surface movements after closure, which is the topic of this study, should be investigated in detail. In the past decades, not only have individual mines in western Europe been closed, but coal production has stopped in entire coal basins. As a consequence, the deep underground was flooded because access to the underground facilities was sealed off, and the underground pumping stations were dismantled. This created a new hazard, i.e., the uplift of the surface caused mainly by the swelling of clay minerals in the argillaceous rocks in the coal strata (Bekendam and Pöttgens, 1995). Although the order of magnitude of the movements in such uplifts is smaller than the subsidence that occurs during mining, cases have been reported in which uplifts have damaged buildings and the surface infrastructure (Baglikow, 2011; de Vent and Roest, 2013; Caro Cuenca et al., 2013). Therefore, studying this phenomenon is more than a pure scientific exercise. To date, other researchers have focused mainly on understanding the phenomenon (e.g., Herrero et al., 2012) and identifying general trends, whereby the link with the rise in water level was an important issue (Caro Cuenca et al., 2013; Devleeschouwer et al., 2008). In this study, we tried to provide better quantification of the movement after closure and the difference between the residual downward movement and the ultimate uplift of the surface. To accomplish this, we studied the past mining directly underneath the observation points.
The underground coal mine of Houthalen, Belgium, was closed in 1992. For a period of nearly 2 decades (from 1992 through 2010), we analyzed the movements of the surface above the mine based on radar interferometry or Interferometry with Synthetic Aperture Radar (InSAR) measurements. The production of coal in this mine began in 1939, and, in 1964, the mine was merged (and connected underground) with the Zolder coal mine, which is situated to the west of the Houthalen mine. Production was stopped in both mines in 1992, and the access was sealed off. Hence, the underground pumps were also stopped, causing flooding of the underground work areas, the surrounding rock mass and caved zones.
Map of longwall panels in area studied, i.e., between a latitude of
51.01–51.05
Longwall mining with goaf was the method used in the mines, and different
coal seams were mined. The area in which the detailed study of surface
movement was conducted is situated from latitudes of 51.01–51.05
The coal strata in the Campine basin in northeast Belgium belong to the Upper Carboniferous strata (Westphalian unit), the time of the formation of many coal fields in Europe (Langenaeker, 2000; Vandenberghe et al., 2014). The top of the Upper Carboniferous strata generally occurs at depths of approximately 400–600 m. The waste rock within these coal strata is composed mainly of argillaceous rocks, like shale and siltstone, and of sandstone and thin (unmined) coal layers. The sandstone is classed as medium-strong, with a typical uniaxial compressive strength (UCS) of 90 MPa (Caers et al., 1997). However, values up to 160 MPa have also been measured. The other types of rocks are classified as weak rock; e.g., siltstone was tested with a UCS value from 17 to 68 MPa with an average of 46 MPa, and coal with a UCS value from 6 to 10 MPa with an average of 7 MPa. The average values of Young's modulus for these three types of rocks were determined as 28 GPa for sandstone, 9 GPa for siltstone and 1 GPa for coal (Caers et al., 1997). Overall, the successive strata are relatively thin (on the order of decimeters to meters in scale). The overburden is composed of weak to very weak geological material (e.g., sand, clay and chalk). Several aquifers and aquitards are present over the entire section of the overburden.
Radar interferometry or Interferometry with Synthetic Aperture Radar (InSAR)
is a recent remote sensing technique that allows the study of large time
series of surface movements (Akcin et al., 2010; Herrera et al., 2009;
Hongdong et al., 2011; Jung et al., 2007; Zhenguo et al., 2013). The movement
of reflective surfaces (i.e., the so-called permanent scatterers) is followed
during successive cycles of the satellite. There is high spatial coverage of
the areas studied, at least if the area corresponds to a built environment.
In comparison to conventional leveling methods, the advantages of
radar interferometry include the following: (i) large areas can be covered for the same
effort (e.g., a full concession area of a mine), (ii) measurements are
conducted on a regular and frequent basis; i.e., one measurement per
satellite revolution (35 days for the datasets used in this research) and
(iii) a dense network of reflectors is available (sometimes every 10 to
20 m). One of the disadvantages is that, when no reflective surfaces are
identified in a specific zone, no information is available on the movement of
the surface. For example, this was the case for the area studied in the zones
composed of agricultural land, woodland and unused or semi-natural land.
Other problems were (1) that the recorded movement corresponds to the
reflection of a surface area of 4
In this study, the European C-band ERS1/2 and ENVISAT-ASAR satellite images
were used, which were available for research through a European Space Agency
(ESA) research proposal (Devleeschouwer et al., 2008). The recorded periods
were for both sets from August 1992 through December 2000 (87 cycles of 35
days) and from December 2003 through October 2010 (72 cycles of 35 days),
respectively. Generally, it is accepted that the linear velocities can be
estimated with accuracies of about 1 mm yr
Earlier research (Vervoort and Declercq, 2016) looked at annual increases in surface movement. It showed that, in this area at the end of the first period of observation (from August 1992 through December 2000), uplift had already been initiated in certain zones or for certain reflectors. In a similar way, it was observed that, at the start of the second period of observation (from December 2003 through October 2010), certain reflectors were still undergoing downward movement. Therefore, in the first instance, we looked at 5-year time zones in each observation period, which can be considered to be characterized by a pure downward movement (for the first observation period from mid-August 1992 through mid-August 1997) or a pure upwards movement (for the second observation period from mid-September 2005 through mid-September 2010). The remaining part of each observation period was also studied and for comparison purposes, a length of 2.5 years was chosen, i.e., the last 2.5 years of the first period and the first 2.5 years of the second period. These two 2.5-year time zones were from July 1998 through December 2000 and from December 2003 through June 2006, respectively. As the total first observation period was longer than 7.5 years and the second shorter, there was a gap between the time zones of 5 and 2.5 years in the first period and a small overlap in the second period, but the main advantage of doing so was that all time zones could be compared more easily. Hence, all scales for the graphs that correspond to the 2.5-year time zones are halved.
In this research, downward movement has a negative sign, and uplift has a positive sign; the same convention was used for the rate of movement (e.g., per year). However, when discussing the smallest (minimum) movement or the largest (maximum) movement, we considered the absolute value of the movement; in other words, when discussing the minimum rate, we did not apply the pure mathematical definition of minimum. For the area studied, no public data were available for the subsidence that occurred prior to the satellite observations.
Information on total surface movement during the two times zones of 5 and 2.5 years considered in the first observation period for the total area studied.
Distribution of total surface movement (in mm):
In the 5 years from mid-August 1992 through mid-August 1997, the area studied
was characterized by an overall downward movement (Table 1, Fig. 2a). Only
2 out of 1073 reflectors were characterized by small upward movements,
i.e., 3 and 6 mm. In the overall picture, these can be neglected. It also
justifies the choice of first considering the first 5 years instead of the
entire observation period. Among the reflectors, 69 % underwent residual
subsidence ranging from
If one looks at the spatial variation of the total surface movement over the 5-year time zone, it is apparent that the largest residual subsidence occurred in the northern central part of the area studied (Fig. 3a). Unfortunately, the reflectors were not evenly spread over the entire area. There were several zones with dimensions of a kilometer wide in which there were no reflectors at all. These zones, in this particular case, were farmland, woodland, unused land and a lake. The combination of large zones without reflectors and the large fluctuation between neighboring points motivated us to present the individual reflectors instead of calculating a contour map. The latter would result in too much loss of detail and would result in large uncertainties for certain sub-zones.
Spatial variation of total surface movement in the area between
a latitude of 51.01–51.05
Variation of the total surface movement along a north–south
transect, situated for a longitude between 5.37 and 5.38
Figure 3a shows that, although large movements may occur next to small
movements, clustering is present. For example, in the western and
southeastern parts of the area studied, the reflectors were mainly
characterized by a residual subsidence of
For the same (first) observation period, the last 2.5 years were also
analyzed (Table 1, Fig. 2b). As mentioned earlier, all scales were halved to
make the comparison easier, and the main reason for considering two time zones
was that we already expected a significant number of reflectors with uplift
at the end of the first observation period. About 8.5 % of the reflectors
underwent uplifts during this time zone of 2.5 years (Fig. 2b). Figure 3b
presents the locations of the corresponding reflectors. It is very clear that
these locations are complementary to the zone of the largest residual
subsidence observed in the first 5 years (Fig. 3a). The maximum subsidence
rate observed was about the same as during the first 5 years, i.e., about
When looking at the north–south transect (Fig. 4b), large subsidence values
occurred in a similar area as in the 5-year time zone. A peak was observed at
a latitude of about 51.04
Distribution of total surface movement (in mm):
Information on total surface movement during the two times zones of 5 and 2.5 years considered in the second observation period for the total area studied.
In the 5-year time zone from mid-September 2005 through mid-September 2010 at
the end of the second observation period, it becomes obvious that an uplift
over the entire area took place (Fig. 5a). Only 6 of the 1808 reflectors
had a slight downwards movement over this time period. This justifies the
choice of first looking at the last 5 years of the second observation period.
The smallest movement was
Spatial variation of total surface movement in the area between a
latitude of 51.01 and 51.05
There was a clear difference between the start and end of the second
observation period, justifying the splitting of the entire observation period
in two. Figure 5b shows the distribution of the increase in surface movement
over the 2.5 years between December 2003 and June 2006. About 6 % of the
reflectors still had undergone subsidence (in comparison to 0.3 % in the
last 5 years). The classes between 5 and 20 mm of total increase
(corresponding to an average annual rate between 2 and 8 mm yr
The map of the reflectors in the area studied now shows a completely
different picture (Fig. 6a) in comparison to the first observation period.
The largest uplift values were observed mainly in the central to southern
part. In the northern part, where the largest residual subsidence was
recorded, small uplift values were observed. In the 2.5-year time zone, from
December 2003 through June 2006, numerous reflectors still had undergone
subsidence in that northern part (Fig. 6b). In the eastern part (longitude
larger than 5.39
Figure 7 presents north–south transects that are similar to those for the
first observation period. In the last 5 years of observation (Fig. 7a), the
maximum uplift was observed at a latitude of about 51.024
In comparison to these north–south transects, the east–west transects
showed a smaller variation, except, of course, that the movement evolved to
zero away from the exploitation in the east. In comparison to the east of the
area studied, where there was no exploitation, the exploitation of the Zolder
mine bordered the exploitation of the Houthalen mine in the west. So, this
clearly affected the movement. As an example, an east–west transect is
presented in Fig. 8 for a latitude between 51.018 and 51.026
Variation of the total surface movement along a north–south
transect, situated for a longitude between 5.37 and 5.38
Variation of the total surface movement along a east–west transect,
situated for a latitude between 51.018 and 51.026
Smoothed curves fitted for the total surface movement along a
north–south transect, situated for a longitude between 5.37 and
5.38
As mentioned above, the movement is more complex than can be represented by a
single value. Hence, one should be careful in replacing the individual
measured values by an average or by a smoothed curve. However, for comparison
purposes, such smoothed curves were drawn for the north–south transects,
presented above. For the smooth curves of both of the 5-year time zones that
were studied, the following observations were made (Fig. 9a).
The absolute movement over 5 years is the same order of magnitude as
the residual subsidence and the uplift. The maximum observed movements were at different locations. In the phase of
residual subsidence, the maximum was situated around a latitude of
51.04 To the northern and southern end of the chosen transect, the movements
evolved towards zero, away from the exploitation. The curve of the uplift is very symmetric, which is not the case for the
curve of the residual subsidence. In Sect. 4.3, the mechanism behind the
uplift is further discussed. All these points are also visible when looking
at the smoothed curves for both 2.5-year time zones that were studied
(Fig. 9b). The only difference is that, in the southern part (more to the
south than a latitude of 51
When looking in detail at the movements at the locations of both maxima, the
above can be more quantified (Table 3, Figs. 10, 11). Around both maxima, the
10 nearest reflectors were selected. The reflectors were not necessarily the
same for the two observation periods, but they were the same within each of the two observation
periods. The number of 10 reflectors is a compromise between zooming in on a
particular area and having enough data to be statistically representative.
Table 3 presents the minimum, maximum and average of the total vertical
movement over the 5-year time zone. The variation of these values as a
function of time is plotted in Figs. 10 and 11. As could be expected based on
Fig. 9a, the difference between the two groups of curves is clear. For the
first observation period, there was a small overlap between the two groups;
i.e., the minimum of the residual subsidence of the location of the maximum
residual subsidence was slightly smaller than the maximum of the other
location studied, but the difference between the two averages was 27 mm over
the 5-year time zone. For the second observation period, there was no overlap
between the two groups. The difference between their averages over the
5 years was 20 mm.
Information on the two locations, corresponding to the zones with
approximately largest residual subsidence in the first period
(Max
Evolution of subsidence over 5-year time zone in the first observation
period (from mid-August 1992 through mid-August 1997):
Evolution of uplift over 5-year time zone in the second observation
period (from mid-September 2005 through mid-September 2010):
Indication of selected locations on map of exploitation panels in
area studied (between a latitude of 51.01 and 51.05
Table 3 provides a summary of some basic information on the exploitation just
underneath the two locations. The surface movement is of course not only
affected by the mining directly below, but also by the mining around the
locations. For an angle of draw of 45
Mining by the longwall method results in caving above the mined-out areas,
creating the goaf area. A roof height of 2 to 8 times the mined height
generally is considered to be sufficient to fill up the mined height, plus
the caved height (Peng, 1986). In the Campine basin, an average value of 5
times normally was assumed, corresponding to a bulking factor of 1.2. The
caved zone is composed of blocks of broken material and includes a large
number of small and large cavities. Hence, Young's modulus of this caved
material is several orders of magnitude smaller than that of the original
intact layers (Galvin, 2016). Over time, this volume is compressed
progressively, but it will never reach its original state. Apart from the
caving of the immediate roof layers, the rock further away fractures, and
sliding along the induced fractures occurs. Still further away from the
mining depth (i.e., closer to the surface), plastic and elastic deflections
of layers also occur. All these phenomena result in the occurrence of
subsidence at the surface. A typical trough shape is created, e.g., above and
around a single panel that has been mined. The zone of influence at the
surface is larger than the dimensions of the panel itself. By considering an
angle of draw of 45
To study the possible link of the residual subsidence with the original mining characteristics in more detail, several groups of locations were selected (Fig. 12). First, three locations were selected where, underneath, no mining had ever taken place (Table 4a). Second, two locations with a small amount of mining, i.e., two panels only and with a total mining height of 2 and 2.5 m, respectively (Table 4b). Third, three locations were selected with extensive mining, i.e., seven or eight panels and a total mining height of 9.2 to 10.3 m (Table 4c). As for the two locations with maximum movement (Table 3), the 10 reflectors in the most immediate vicinity were studied. It was not easy to find an adequate number of locations so proper analyses could be done; i.e., enough reflectors had to be present in both observation periods at a close distance, and the same mining conditions had to exist underneath these reflectors.
Information of selected locations, i.e., movement of 10 reflectors
around coordinates given over 5-year time zones in both observation periods
and mining characteristics underneath locations:
When one looks at the average total residual subsidence over the 5-year time
zone, one gets
By looking at the data of Table 4b and c as a function of the mining depth, no clear trend is observed. For the two locations with a limited amount of mining, the most shallow mining resulted in the largest residual subsidence, while, for the three locations with extensive mining, the largest residual subsidence was for the deepest exploitation.
Information on residual subsidence of the locations considered in Tables 3 and 4, reordered as a function of the most recent exploitation panel.
When comparing the two locations of maximum movement in Sect. 4.1, there was
the possibility that more residual subsidence occurred directly above the
more recent longwall panel. This would be logical. Therefore, Table 5
classifies the various locations as a function of the most recent longwall
panel underneath. Taking into account the large number of possible parameters
that influenced the results, the trend of these seven locations is indeed
that the locations above the most recent panels resulted in larger residual
subsidence. However, it must be pointed out that the location with the
second-most recent mining has undergone, on average, less movement than one of the
locations without mining underneath (i.e., No
When comparing the residual subsidence in the north–south transect (Fig. 9a)
with the map of longwall panels, one can observe that the zone of influence
is larger than expected based on the normally used values of the angle of
draw. Based on the latter values and the depth of exploitation, the influence
zone during the phase of subsidence should be limited to the zone between
a latitude of 50.995 and 51.06
What was explained in the previous section is the process that was initiated by
the caving process, and it can be seen as a mechanical stress-deformation
process that includes time-dependent aspects. However, once the underground
activities ceased and the underground access was closed off, including
dismantling of the pumping installations, the underground workings begins to be
flooded (Bekendam and Pöttgens, 1995; Caro Cuenca et al., 2013; de Vent
and Roest, 2013). In the beginning, the water finds its way through various
pathways, including open roadways, permeable faults, and volumes of loose
blocks. However, there is no reason the rock mass adjacent to the mined area or
between mined areas would not be submerged, and this leads to new processes.
In the literature (Herrero et al., 2012), the swelling of clay minerals of
argillaceous rocks under the influence of water is considered to be the main
factor for inducing uplift. Swelling is governed by the swelling pressure and
is, therefore, linked to the mining depth. Caro Cuenca et al. (2013) showed the direct correlation between the increase of the water level in the
underground areas and the uplift clearly. In all cases, the groundwater levels
even showed a very high correlation (
For the same average locations, as for the first observation period, the
minimum, average and maximum uplift of the 5-year time zone for 10
reflectors are given in Table 4. By considering the three groups as a
function of the amount of mining, one gets average uplifts of
Often, one links the largest uplift to zones with the largest subsidence, and estimates the total uplift to be 3 to 4 % of the total subsidence (Herrero et al., 2012). Bekendam and Pöttgens (1995) also concluded that, generally, the uplift is 2 to 4 % of the subsidence; the latter conclusion is for the same Campine basin, but above the Dutch coal mines to the east. This cannot be confirmed by the area studied here and, of course, for the time periods considered; only the residual subsidence rate is known. As pointed out earlier, no public data were available for the subsidence that occurred prior to satellite monitoring, but by applying the rule of thumb for estimating the total subsidence, one could estimate that the subsidence was from about 1.5 to 11 m in the area studied, and 3 % of this would mean that a total of 45 to 330 mm of uplift finally would occur above the mined-out area. If this were correct, then the uplift during the second observation period (until 2010) would have reached only the bottom part of this predicted range; in other words, one can still expect more uplift above the mining area and immediate surroundings.
As discussed in Sect. 4.2, the influence zone during the phase of subsidence
should be limited to the zone between a latitude of 50.995 and 51.06
Based on all of the information that was collected, there is no indication that the process of uplift is directly linked to the mining characteristics. It is more likely that the uplift as a result of the flooding is initiated at or close to the shafts, where most likely the deepest point is situated and where the pumping station was situated. From that central location, further flooding (in the horizontal direction) and rise of mine water (in the vertical direction) are extended, creating a further uplift at that central location and an initiation of uplift further away from the central area.
Of course, the fact that mining and caving have taken place has an effect. It is the main reason that water flows into the underground workings. However, the local situation (e.g., the depth, extent or time of mining) does not seem to have a very significant influence on uplift. When looking at the interpolated curve of Fig. 9a, no local irregularities are noted; the curve itself also is very symmetric, much more so than the curve of residual subsidence (Fig. 9b).
Most research of surface movement above underground mines
focuses on the direct effect of mining, i.e., within the lifetime of the
mine, and less attention is given to the long-term impact of mining on
surface movements. At the end of the last century, several coal basins
were closed in Europe, and researchers began to observe a new phenomenon,
i.e., the uplift of the surface as a consequence of the flooding of the
underground workings (Bekendam and Pöttgens, 1995). In addition, cases were
reported of damage to buildings and infrastructure during the uplift phase
(Baglikow, 2011; de Vent and Roest, 2013; Caro Cuenca et al., 2013). During
that period, satellite images with frequent and detailed measurements of the
surface movement over large areas became available, so this topic could be
studied further. To date, the focus has been mainly on understanding the
phenomenon (e.g., Herrero et al., 2012) and identifying general trends, like
for example the link with the rise in the water level (Caro Cuenca et al.,
2013; Devleeschouwer et al., 2008). In this study, the residual subsidence
after closure, as well as the initiation and further evolution of the uplift
were investigated for an area of 22 km In the first 5 years following the closure of the coal mine (between
mid-August 1992 and mid-August 1997), the area studied was still
characterized by an overall downward movement; the average residual
subsidence was Although large residual movements may occur next to small movements,
clustering was present, and it resulted in areas with, on average, smaller
residual subsidence and other areas with larger values; certainly when
looking at the north–south transects, there was a clear zone in which the
maximum residual subsidence occurred. In absolute terms, the rate of uplift was about the same order of magnitude
as the residual subsidence, but, in fact, it was slightly larger; an average
rate of uplift of 9 mm yr The zone in which the maximum uplift occurred was clearly at a different
location from the zone with the maximum residual subsidence. The curve of the uplift along north–south lines was very symmetric, which
was not the case for the curve of the residual subsidence. There was no clear sign that the amount of mining underneath a relatively
small area had an effect on the residual subsidence. However, there was some
indication that the locations above the most recent panels resulted in
larger residual subsidence values. There is not a simple one-on-one
relationship with the time since exploitation. The zone of influence was
larger than one would expect based on the normally used values of the angle
of draw and depth of mining. Based on all of the information that was collected, there was no
indication that the process of uplift was directly linked to the mining
characteristics. It is more likely that the uplift as a result of flooding
was initiated at or close to the shafts; from that central location, the
additional flooding (in the horizontal direction) and rise of mine water (in
the vertical direction) were extended, creating additional uplift at that
central location and initiating uplift further away from the central area.
Most concepts that one finds in textbooks dealing with surface subsidence
above longwalls considers either the impact of mining a single panel or a
relatively simple mining geometry and/or mining sequence (e.g., mining a
single seam with adjacent panels, which are mined in a successive sequence).
The latter is certainly typical for several countries, including the large
coal producers, such as Australia, South Africa and the United States. In Europe, the
situation was often different (Preusse et al., 2013). For the mine studied, a
total of 10 seams were partly mined over a time period of 60 years (between
1932 and 1992) at depths varying from 539 to 967 m. However, the situation
was not significantly different when a shorter time is considered. For
example, in the 1970s, seven seams still were being mined at depths varying
from 556 to 824 m. As one also observes on the map of longwall panels
(Fig. 1), there was no systematic geometry or a systematic approach of mining
the different panels. These observations probably explain why no clear link
has been established between mining characteristics and residual subsidence.
The entire area was rather in movement. For the amount of uplift, such
one-on-one relationships were nonexistent. As illustrated above, one can best
visualize the uplift as starting at or close to the shafts, whereby a further
uplift occurred in the following years at that central location, and uplift
was initiated farther away from this central area. This seems to be in
accordance with the process of flooding the underground and the systematic
rise of the water level. It will be interesting to investigate the further
evolution of the uplift, when more recent satellite data become available.
The process of subsidence and the one of uplift are entirely different. The first is caused by a caving process and is mainly a mechanical stress-deformation process, including time-dependent aspects, while the process of uplift is caused by the swelling of the clay minerals in the argillaceous rocks in the coal strata, due to flooding. Hence, one cannot assume that the areas where one has the greatest risk for damage to infrastructure due to subsidence are the same areas for the hazards linked to the uplift process.
As for this area, no one-on-one relationships could be clearly identified between the surface movement and the mining characteristics, future research of this multivariate problem could benefit from using techniques including unsupervised learning and supervised learning (Noack et al., 2014). It would be best to have data on the initial subsidence, the residual subsidence and the uplift, combined with data on the mining characteristics.
In this study, the European C-band ERS1/2 en ENVISAT-ASAR satellite images were used, which were made available by ESA for research. To obtain these data, researchers who are interested have to submit a research proposal to ESA.
The author declares that he has no conflict of interest.
The input by Pierre-Yves Declercq from the Geological Survey of Belgium, Royal Belgian Institute of Natural Sciences, Brussels, is greatly acknowledged for providing the necessary basic data on surface movements and mining characteristics. Edited by: T. Glade Reviewed by: two anonymous referees