**1. Introduction**

There is general agreement that the total extraction mining method, such as longwall mining with the associated goaf, has an impact on the environment, e.g., the subsidence of the surface [1–5]. It also is obvious that this impact is permanent, but that issue is not the focus of this paper. The problem being addressed herein is whether the surface movement will be an ongoing phenomenon long after all mining activities have stopped or whether the movement will reach a state of complete, stabilized equilibrium after a certain period. Generally, the people in the industry, as well as researchers, have the opinion that the surface movement is limited in time. Often, it is claimed that subsidence only occurs over a period of a few months to a few years and that, subsequently, the ground is stable. Some researchers divide the process of subsidence into various phases [6–9]. They identify the first phase as initial subsidence and the second phase that is referred to as either principal, accelerated, or steady-state subsidence. The third phase is referred to as residual, differed, or delayed subsidence, and it is assumed to be limited in time. After the third phase, it is assumed that the subsidence has stopped and that the ground is stable [7]. Some examples of time periods for the residual phase from various places in the world are given by Al Heib et al. [7], and their conclusion is as follows: *"The duration of the residual subsidence phase is about 12 to 18 months, but this duration is often less long, i.e., about 3 to 4 months when the exploitation is carried out in an already disturbed zone (several seams, goaf,* ... *). There are some isolated cases resulting from geological contexts and*/*or particular exploitation, for these cases, the duration of residual subsidence can appreciably be raised and spread out over one period of 4 to 6 years."* The end of the residual subsidence phase is defined as either (1) when an amount of additional subsidence corresponding to a certain percentage of the subsidence during the two previous phases has been reached, (2) when the increments between successive measurements have become negligible, or (3) when a predefined fixed time period has been reached.

The basis for conducting the research presented in this paper was not a specific interest in long-term subsidence; rather, it was the occurrence of partly unexpected and new behavior of the surface movement several years after the closure of the last coal mines in northeastern Belgium around 1990 [10,11]. A few years after the closure of the underground infrastructure and the dismantling of the underground pumping stations, the surface started to move upward. This phenomenon first was observed in the Netherlands, where the coal mines were closed in the 1970 s [12]. Efforts to understand and explain this uplift phenomenon led to the study of the residual subsidence prior to the start of uplift, and hence, to the study of the long-term impact of deep coal mining.

The Campine basin belongs to the Upper Carboniferous strata (Westphalian stage). It is part of the South Permian basin of northwestern Europe [10]. Mainly coking or metallurgical coal was mined, with an average calorific value of 35 MJ/kg. The overburden contains sand, clay and chalk; i.e., all weak geological material. The total thickness of the overburden varies between 400 and 600 m. It contains several aquifers and aquitards [13]. The waste rock between the mined coal seams is composed mainly of shale, siltstone, sandstone, and thin (unmined) coal layers. Overall, the successive layers of waste rock are relatively thin (on the order of decimeter to meter in scale).

#### **2. Global View of Surface Movement after the Closure of Mines**

This paper presents data for the western part of the Belgian Campine basin in northeastern Belgium. Three coal mines were active in that part of the basin, i.e., from west to east, the Beringen, Zolder, and Houthalen mines [10]. The depths of the mining in the western part of the Belgian Campine basin varied from −485 m to −967 m. The mining height for a single longwall varied from 0.5 m to 3.3 m, with an average of 1.5 m. In most areas, 5 to 10 coal seams were mined above each other. Coal production started for the three mines between 1922 and 1939, and all of the production was terminated between 1990 and 1992. The underground pumping stations were dismantled, and the shafts were closed in the years after the production was terminated. This western part of the Campine basin was characterized by a period of additional residual subsidence that was followed by uplift. An earlier study showed that a residual downward movement occurred for the Houthalen coal mine until the end of 1999, which was followed by limited surface movement from 2000 through 2004, and a clear uplift since 2005 [13]. However, spatial variations were observed.

Taking the timing of closure and the subsequent periods of residual subsidence and uplift into account, the images of the European C-band ERS1/2 and ENVISAT-ASAR satellites were the most suitable. These data were acquired for research through a European Space Agency (ESA) research proposal [14]. The periods that were recorded for the two sets were from August 1992 through December 2000 (87 cycles of 35 days) and from December 2003 through October 2010 (72 cycles of 35 days), respectively. Radar interferometry or Interferometry with Synthetic Aperture Radar (InSAR) is very suitable for studying large time series over large surface areas. The movements of reflective surfaces (i.e., the so-called "permanent scatterers" or "reflectors") were followed during all successive cycles of the satellite. This allowed the acquisition of significant spatial coverage of the areas that were studied, at least if the areas were part of a built environment.

Figure 1 presents a global view of the surface movement in the period of mainly residual subsidence (Figure 1a), and in the period of mainly uplift (Figure 1b), for the entire surface area above and around the three coal mines. For comparison purposes, the surface movement was extrapolated to a total time period of 10 years for each Figure. Hence, Figure 1a covers the period from August 1992 through August 2002, and Figure 1b covers the period from March 2003 through March 2013. As the variation as a function of time is not always linear (see further), the increase of surface subsidence over the last 20 months in the first period was added to the value recorded for December 2000, while for the second period the increase in uplift during the first nine months and the last 29 months were added. Since 1 mm/year generally is accepted as a possible error on the trend of surface movement recorded by these satellites [15,16], only reflectors with a minimum additional movement of 10 mm over 10 years are presented. The outside borders of the zone of the longwall panels that were mined are presented by straight blue lines. The data clearly show that the reflectors with a significantly larger movement occurred mainly within these borders or in the immediate vicinity.

**Figure 1.** Spatial distribution of additional surface movement over a period of 10 years: (**a**) Period from August 1992 through August 2002, residual subsidence larger than −10 mm; (**b**) period from March 2003 through March 2013, uplift larger than 10 mm. The outside borders of the zone of the mined longwall panels are shown with the blue color.

However, some (smaller) movements are also recorded outside these borders. For a single panel, the angle of draw in the Campine basin is often assumed to be 45◦ [5]. In other words, the size of the zone of influence is approximately the depth of mining. Recent research has shown that this is an underestimation when looking at the entire coal basin. In Reference [5], this is illustrated by some examples: At about 3 km further than the boundary a systematic surface movement was recorded over a period of 20 years, while at a distance of 7 km no movement at all was recorded. As discussed later, by analyzing long-term frequent data series over large areas, new insights occur. With conventional levelling campaigns, one has the tendency to limit the area of monitoring and to stop monitoring once the difference in surface movement between successive levelling campaigns becomes equal to the accuracy of the levelling method.

Figure 2 presents the reflectors with an additional surface movement within ±10 mm over 10 years, as well as the reflectors with the opposite movement from the global trend (i.e., already uplift in the first period and still residual subsidence in the second period). In comparison to Figure 1, these two groups of reflectors now are situated mainly outside the mined area. In other words, one can conclude definitively that both types of movements (downward and upward) are linked to mining. The other conclusion is that two decades after the closure of the mine, there is still movement of the surface, and one cannot assume that the strata are stabilized.

**Figure 2.** Spatial distribution of the reflectors with an additional surface movement of ±10 mm over 10 years (grey squares), as well as the reflectors with the movement that is opposite of the global trend (plus and minus signs, respectively): (**a**) Period from August 1992 through August 2002; (**b**) period from March 2003 through March 2013.

Globally, about one and a half to two times more reflectors were detected with the more modern second satellite (ENVISAT-ASAR) than with the first (European C-band ERS1/2). Some sub-areas within the mined area do not have any reflectors, e.g., the band between the east-west, north-south coordinates (4.5 km, 10.5 km) and (9.0 km, 13.0 km). These gaps where there are no reflectors in Figure 1 correspond to zones composed of natural land, forest, lakes, and agricultural land. Also, the area to the northeast of the mined area is such land.

When comparing in detail the spatial distribution of the surface movements between Figure 1a,b, one cannot conclude that the sub-areas with large residual subsidence correspond with sub-areas of large uplift. For example, in the northwest, the large uplift values for coordinates (6, 12) in Figure 1b (uplift values of mainly 75 to 100 mm over 10 years) correspond mainly to residual subsidence values of −25 to −75 mm in Figure 1a. Something similar can be observed for the entire western part in the

northwest sub-area, i.e., within the east-west distance of 3 to 5 km and north-south between 11 and 14.5 km. Another example is the entire zone between east-west 11 and 17 km and north-south between 6 and 7 km. This zone is characterized by large uplift values and relatively low residual subsidence values. However, the zone around coordinates (8, 8) is characterized by high residual subsidence values (mainly −100 to −150 mm) and low uplift values (25 to 75 mm). All of these observations are in agreement with earlier research results [11,17] that showed that the spatial distributions of the residual subsidence values and the uplift values were different, and no correlation was observed between these two parameters. This information is useful in trying to understand the mechanisms that cause an uplift in comparison to the mechanisms that cause subsidence, as will be discussed later.
